Intelligent Systems Reference Library 193

Valentina E. Balas
Vijender Kumar Solanki
Raghvendra Kumar   Editors

Further Advances in
Internet of Things
in Biomedical and
Cyber Physical
Systems
Intelligent Systems Reference Library

Volume 193

Series Editors
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
Lakhmi C. Jain, Faculty of Engineering and Information Technology, Centre for
Artificial Intelligence, University of Technology, Sydney, NSW, Australia;
KES International, Shoreham-by-Sea, UK;
Liverpool Hope University, Liverpool, UK
The aim of this series is to publish a Reference Library, including novel advances
and developments in all aspects of Intelligent Systems in an easily accessible and
well structured form. The series includes reference works, handbooks, compendia,
textbooks, well-structured monographs, dictionaries, and encyclopedias. It contains
well integrated knowledge and current information in the field of Intelligent
Systems. The series covers the theory, applications, and design methods of
Intelligent Systems. Virtually all disciplines such as engineering, computer science,
avionics, business, e-commerce, environment, healthcare, physics and life science
are included. The list of topics spans all the areas of modern intelligent systems
such as: Ambient intelligence, Computational intelligence, Social intelligence,
Computational neuroscience, Artificial life, Virtual society, Cognitive systems,
DNA and immunity-based systems, e-Learning and teaching, Human-centred
computing and Machine ethics, Intelligent control, Intelligent data analysis,
Knowledge-based paradigms, Knowledge management, Intelligent agents,
Intelligent decision making, Intelligent network security, Interactive entertainment,
Learning paradigms, Recommender systems, Robotics and Mechatronics including
human-machine teaming, Self-organizing and adaptive systems, Soft computing
including Neural systems, Fuzzy systems, Evolutionary computing and the Fusion
of these paradigms, Perception and Vision, Web intelligence and Multimedia.
Indexed by SCOPUS, DBLP, zbMATH, SCImago.
All books published in the series are submitted for consideration in Web of Science.

More information about this series at http://www.springer.com/series/8578

Valentina E. Balas Vijender Kumar Solanki
• •

Raghvendra Kumar
Editors

Further Advances in Internet

of Things in Biomedical
and Cyber Physical Systems

123
Editors
Valentina E. Balas Vijender Kumar Solanki
Department of Automatics Department of Computer Science
and Applied Software and Engineering
Aurel Vlaicu University of Arad CMR Institute of Technology (Autonomous)
Arad, Romania Hyderabad, Telangana, India

Raghvendra Kumar
Department of Computer Science
and Engineering
GIET University
Gunupur, Odisha, India

ISSN 1868-4394 ISSN 1868-4408 (electronic)

Intelligent Systems Reference Library
ISBN 978-3-030-57834-3 ISBN 978-3-030-57835-0 (eBook)
https://doi.org/10.1007/978-3-030-57835-0
© Springer Nature Switzerland AG 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

The main objective of this book publication is to explore the concepts of Internet of
Things, biomedical and cyber physical systems along with the recent research and
development. It also includes various real-time applications and case studies in the
field of engineering and technologies used. As populations grow and resources
become scarcer, the efficient usage of these limited goods becomes more important.
The content of the book is divided into four different sections.

Section I: Distributed Sensor Networks

Chapter 1 discussed cluster formed by this low-energy node will terminate pre-
maturely and waste entire network resource. ILEACH is measured one of the finest
of them. To improve service life, sensor nodes with high residual energy and short
distance from the base station (BS) are chosen as cluster head (CH) nodes. Then
intelligently manage these nodes to create clusters to maximize the lifetime of the
WSN and minimize the average energy consumption. The TDMA protocol is used
for intra-cluster communication. In this article, we propose a reform of the ILEACH
protocol by acquaint with cluster communication, in which the cluster heads are
organized in a hierarchical structure, additional optimizing the life of the WSN. The
simulation outcomes illustrate that the improved algorithm in this respect outper-
forms the LEACH protocol.
Chapter 2 proposed framework consolidates wearable sensors to quantify
physiological and natural parameters. A passage is acquainted with giving infor-
mation preparing, a neighborhood web server, and a cloud association. A wearable
sensor on a laborer and natural sensor on a wanderer that can transmit the infor-
mation to the client by means of a door for example server, gives offer notice and
cautioning component for the clients. Live health examination taken for laborers
who work in an underground like tunnels, shafts, etc., it has an Individual database
of laborers and contrasts it, and current essential tangible qualities separate to
workplace information. Live update will screen from the control room, and it can

v
vi Preface

direct the specialist if any medical problem occurs and furthermore can maintain a
strategic distance from the undesirable passing.
Chapter 3 discussed cloud computing has the implementation of traditional IT
for high implementation time. The security is the main problem of the big data
analytics for implementing governance and risk management. The factors for
affecting the cloud computing have identified for the hybrid cloud computing.
Chapter 4 focuses on DoS attacks in cognitive radio networks (CRNs). The
presence of malicious users is threat for enhancing the effective spectrum utiliza-
tion, and this threat may be an active or passive. In an active attack, malicious user
will deliberately upset the primary user framework. A passive attack relates to the
circumstance in which a malicious attack endeavors to translate source data without
infusing any data or attempting to alter the data, i.e., it will tune in to the trans-
mission without cooperating with other users. The network consists of two users
such as primary users and secondary users where the main impact occurs on pri-
mary users. The network performance parameters such as packet delivery ratio,
packet loss ratio, bandwidth usage and end to end delay are analyzed in CRN to
detect DoS attacks.
Chapter 5 indicated that the alternating deposition direction strategy allows
achieving thin walls with more regular height. The roughness of the side surface
of the thin walls is about 0.23 mm. The microstructure of ER70S6 thin walls
changes from region by region: The upper region consists of lamellar structures; the
middle region features granular structures; and the lower region shows mixed
lamellar and equiaxed structures. The hardness also varies according to these three
regions. The upper region shows the highest average value of hardness
(
191 HV), followed by the lower region (
178 HV) and the middle region
(
163 HV). Finally, the ER70S6 walls built by WAAM exhibit the anisotropy in
terms of tensile strengths in the horizontal and vertical directions.
Chapter 6 analyzed the factors that impact the precisional control process
of the DC motor such as disturbance, the temperature effect on coil resistance and
the temperature effects on magnetic fluxes. From that, we suggest a method of the
actuator fault estimation to apply for the process of eliminating fault which will be
performed in the future. First, a nonlinear mathematical model of the DC motor
under the action of temperature is constructed to control the system. Second,
building the inequalities based on the reconstruction of unknown input observer
(UIO) with considering disturbance is constructed to estimate the actuator faults
based on Lyapunov’s stability condition and a linear matrix inequality
(LMI) optimization algorithm in order to obtain the control signal error asymp-
totically stable. Finally, the numerical simulation process is done to show the
obtained result of the proposed method.
Chapter 7 analyzed the problems occurred for maintaining big data processing
and maintaining it. Hadoop Distribution File System is developed to produce the
solutions for the big data challenges with the concept of acquire, organize, analyze
and decide using analytic application.
Preface vii

Section II: Intelligent System Design and Applications

Chapter 8 discusses an objective of this application to make it interactive and make

facilities to the user for solving daily problems. Citizens can use this system to
search for information and send necessary documents within the electronic gov-
ernment for the Ministry of Commerce in client side for solving their problems.
Also, it is used to complete their information in the database system by using full
name and ration card number of family to enter the system. Citizen can follow this
case and can find the result for this request from the system.
Chapter 9 proposed a new approach by combining automatic syntactic features
with pre-trained word embedding in deep learning method—Bidirectional Long
Short-Term Memory (BiLSTM) for Vietnamese Named Entity Recognition. The
proposed system has achieved good results for the Vietnamese NER problem on the
VLSP 2016 dataset.
Chapter 10 proposed a new design of convolutional neural networks (CNNs) and
principal component analysis (PCA). The proposed system is estimated using
five datasets, (Mhearth), (Sensors-Activity-Recognition-Dataset-SHOAIB),
(REALDISP), (REALWORLD) and (Activity Recognition Dataset), where the
accuracy equals 99.8%, 99.44%, 99.85%, 96.90%, 98.68%, Sequentially.
Chapter 11 discussed the overall and detailed designs of the robot are presented.
Also, some main research results relevant to the implementation and pilot appli-
cations of the robot are shown. It was demonstrated that the robot prototype was
effectively implemented and tested with the use of TIG/MIG/MAG welding
methods. The use of the robot in a welding cell reduces the production cost,
improves the product quality and optimizes the manpower used in the welding
process.
Chapter 12 proposed a modified reversible parallel and serial adder/subtraction
circuit using dual key gate (DKG) and SG. The performance of parallel
adder/subtractor circuit design using dual key gate and serial adder/subtractor using
dual key gate with SG is simulated and synthesized using Xilinx. The performance
of this circuit is compared with existing design using Feynman gate and toffoli gate
based on complexity, low power and garbage input/outputs.
Chapter 13 designed by adopting the elbow inlet for generating the swirl flow.
The temperature difference of steel pipes for the original furnace was 55.4°C.
However, the temperature difference of a bundle of steel pipes for the new model by
using downward inlet angle of 30 degrees was 13°C. Moreover, temperature dif-
ference by modified model had more uniform than those of by Prieler et al. of 50°C.
Effect of Reynolds number, the horizontal and the vertical steel spacing on the
temperature difference among the steel pipes were also examined. The transient
simulations were performed to investigate steel pipe temperature during annealing
process. In short, the results revealed that a new designed inlet configuration with
this kind of geometry of an annealing furnace could be applied to minimize the
temperature difference among steel pipes in steel annealing process.
viii Preface

Chapter 14 designed a real-time facial expression classification the same system

using Deep Convolutional Neural Networks (DCNN) that could be used for
security checks in public places. The design of the system is conducted in five
stages. Initially, facial expression images for the seven categories are captured,
preprocessed, and loaded into the system, so that there are ten images for each
of the category, namely angry, disgust, sad, fear, neutral, happy and surprise. The
network architecture is then defined by repeating convolution layer, batch nor-
malization layer, rectified linear layer (ReLU), max pooling layer three times.
A fully connected layer is then defined for training followed by a SoftMax layer.
Chapter 15 aimed to find the interdependency between cost and time problems in
construction projects and determine which problem has major impact on the other
problem to find its solution. The results show PSO is very fast in finding the
interdependency between the problem and the method in searching which is very
smooth as it finds the solution in the third iteration and that the effect on the
problem one on two is slightly different from the two on one which means that both
require solution and it’s the same applied on the problem two and three which
indicate that the construction phase is very critical and require great attention.

Section III: IoT Applications in Biomedical Engineering

Chapter 16 discussed a novel strategy for automated water irrigation together with a
platform for pest detection, which can be used to control the water level and water
the crops in agricultural lands, accordingly, is presented in this paper. The water
pump is triggered depending on the water level in the soil. In addition, we have
implemented a new algorithm in this method to identify the pests in the plants. It
will take reasonable measures to eradicate it, depending on the nature of infection.
The proposed algorithm employed is built on the extraction of appropriate features
from the plant leaves, and those features are utilized for classification.
A comparison of the proposed algorithm with current algorithms like k-NN and
decision tree was set up to yield admirable results.
Chapter 17 created a small medical system within the context of the Health
Telematics software that enables specialist doctors to use telescope tracking,
long-range assistance and telecommunication from mobile health providers. The
system enables critical biosignals and photographs of patients to be submitted to the
hospital from site. Data are transmitted over a GSM network or Wi-Fi connection.
Owing to the need to share and record data during telemedicine sessions, we have
aligned the consulting network with a digital database that can store and handle the
“ambulance” program captured data.
Chapter 18 proposed a greenhouse automation system based on Arduino for the
monitoring of temperature, humidity and moisture of the soil. Arduino can obtain
data on the environmental conditions of the greenhouse from various sensors and
transfer the data to the ESP8266 module. Consequently, it is possible to change the
state of greenhouse control devices like fans, lamp heater and water pump in
obedience to the necessary conditions of the crops. These parameters are modified
Preface ix

by the type of plant to maximize their growth, the Aloe Vera plant was used in this
project. For the architecture of the Internet of Things was used Blynk coming from
the embedded board and the communication link with the Blynk Server was
through the Wi-Fi protocol. Results indicate that the system allows the control and
monitoring in real time of the greenhouse correctly. As a future improvement, it is
intended with the data obtained, to search for the best optimal conditions for plant
growth through artificial intelligence.
Chapter 19 analyzed and discussed the impact of artificial intelligence on the
HRM process. Application of AI tools for screening of candidates, engagements of
employees and their career enhancement have been explained thoroughly. Some
challenges in HR practices based on data science are extensive HR phenomenon,
and there are many restraints due to small datasets. In this chapter, we have ana-
lyzed the gaps between the realism of artificial intelligence in managing human
resources and expectations. Suggestions have also been discussed for the progress
to be made. The methods which have sustainable HR and talent acquisition, training
through the utilization of technology, have also been discussed in the present
chapter. Practical examples to understand the collaboration of working with AI are
considered as well.
Chapter 20 identified the emergency call headed back to the location and to
monitor the congestion system in order to provide efficient facilities. This journal
also sets out a method that uses a ZigBee component and Internet of Things (IoT) to
transmit the treatment request from the ambulance to the nearby hospitals, while
ambulance attaining the road junction, the smart traffic system which in turn
changes the traffic signal cycle. This system can be implemented throughout the city
thereby reducing the delay.
Chapter 21 proposed a method, which will ensure road safety, women safety as
well elder people safety. Overall, the major threats and difficulties faced by people
will be prevented or solved using this application.
Chapter 22 deals with the basic introduction to the concept of Internet of Things
(IoT), which is a concept that enables a device to connect to the Internet or other
devices, hence forming a giant ecosystem. The IoT platform acts as the brain of the
system, while the devices linked together via IoT function as limbs.

Section IV: Cyber Physical System Framework and Applications

Chapter 23 introduced multiple linear models fuzzy objectives as both the objective
functions and variables coefficients (time) fuzzy numbers Trigonometric function
was used to convert the fuzzy numbers to the normal formula and then construct the
mathematical model and solve it using the goal programming method.
Chapter 24 presented an effective technical solution for integrating and con-
trolling a heavy robot of which all joints are driven by hydraulic actuators. The
robot is designed to support workers for transferring hot and heavy workpieces
between a heating furnace and a hydraulic press machine of a hot press forging
shop floor. The control system of the robot is integrated mainly based on the
x Preface

industrial PLC units, which also plays a role as a central control unit for controlling
all the components of the entire forging station. The control algorithm for the
system integration is validated through functional tests and experiments that are
carried out with a real forging shop floor at The Mechanical Company Ltd. No. 83,
Yen Bai Province, Vietnam.
Chapter 25 used simple IR system to compute the percentage of grams of
plagiarized texts, done in two methodologies: (i) percent of plagiarism in suspicious
document (a file in dataset) and (ii) percent of plagiarism in a file in dataset (to
suspicious file). And then the Precision, Recall, F-measure and Error rate are
estimated. In case of execution time, the proposed method is four times faster than
winnowing algorithm.
Chapter 26 proposed method uses Named Entity Recognition technique with a
recurrent neural network model in combination with conditional random field
model to extract asset features, thereby building a regression model to evaluate the
price of assets based on the attribute set. The method works relatively well with a
dataset of mobile phone descriptions with high accuracy.
Chapter 27 discussed CrowdBC, a blockchain-based decentralized framework
for crowd sourcing, within which a requester’s task are getting to be solved by a
crowd of workers without trusting on any third party, users’ privacy is often
guaranteed, and only low transaction fees are required. In particular, we introduce
the architecture of our proposed framework, supported which we provide a concrete
scheme. We further implement a software model on Ethereum public test network
with real-world dataset. Experiment results show the feasibility, usability and
scalability of our proposed crowd sourcing system.
Chapter 28 gives a empirically investigation, what effect chatbots have on
subjective well-being of the students and how subjective well-being can be mea-
sured in terms of this research context. The major findings of the study are that
students’ subjective well-being has been measured into three dimensions. The first
dimension of subjective well-being is life satisfaction which represented that overall
students feel satisfied after conversing with the chatbot. The second dimension is
positive affect which is also increasing after conversing with the chatbots. The third
dimension is negative affect which is decreasing among the students after con-
versing with the chatbots. So, it is recommended to the college and university to
implement chatbots in their premises for the wellbeing of their students.
We are sincerely thankful to Almighty to supporting and standing in all times
with us, whether its good or tough times and given ways to conceded us. Starting
from the call for chapters till the finalization of chapters, all the editors have given
their contributions amicably, which itself a positive sign of significant team works.
The editors are sincerely thankful to all the members of Springer (India) Private
Limited, especially Prof. (Dr.) Lakhmi C. Jain, S. Tigner and Aninda Bose for the
providing constructive inputs and allowing opportunity to edit this important book.
We are equally thankful to a reviewer who hails from different places in and around
the globe shared their support and stand firm toward quality chapter submission.
The rate of acceptance we have kept as low as 16% to ensure the quality of work
submitted by author. The aim of this book is to support the computational studies at
Preface xi

the research and post-graduation level with open problem-solving technique, and
we are confident that it will bridge the gap for them by supporting novel solution to
support in their problem solving. At the end, editors have taken utmost care while
finalizing the chapter to the book, but we are open to receive your constructive
feedback, which will enable us to carry out necessary points in our forthcoming
books.

Arad, Romania Valentina E. Balas

Hyderabad, India Vijender Kumar Solanki
Gunupur, India Raghvendra Kumar
About This Book

The edited book covering the further advances in the fields of Internet of Things,
biomedical engineering, and cyberphysical system with recent applications. It is
covering the various real time, offline applications and case studies in the fields of
recent technologies and case studies of Internet of Things, biomedical engineering,
and cyberphysical system with recent technology trends. In the twenty-first century,
the automation and management of data are vital, in that the role of Internet of
Things proving the potential support. The book is consisting of excellent work of
researchers and academician who are working in the domain of emerging tech-
nologies, e.g., Internet of Things, biomedical engineering, and cyberphysical sys-
tem. The chapters covering the major achievements by solving and suggesting
many unsolved problems, which am sure to be going to prove a strong support in
industries toward automation goal using of Internet of Things, biomedical engi-
neering, and cyberphysical system.

xiii
Key Features

1. This book will provide in-depth knowledge about enhancements of Internet of

Things in related fields.
2. Technical approach in solving real-time/offline Internet of Things applications in
biomedical and cyberphysical system.
3. Practical solutions through case studies in Internet of Things, biomedical and
cybersecurities.
4. Companies may get different ways to monitor biomedical data from various
medical sensors and modify their processes accordingly to prevent unauthorized
leakage of the data.
5. The interdisciplinary tools & cases of Internet of Things, biomedical and
cyberphysical system.

xv
Contents

Part I Distributed Sensor Networks

1 Energy Efficient Multi-hop Routing Techniques for Cluster Head
Selection in Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . .. 3
G. Hemanth Kumar, G. P. Ramesh, and C. Ravindra Murthy
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2 Intelligent Wearable Sensor Band for Underground
Working People . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 11
S. Karthikeyan, G. Sethuram Rao, M. S. Kowshik, P. Mohan Raj,
G. Vishal, R. Juliet, P. Swetha, and T. Veronica
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Hardware Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 Rover Implementation . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.2 Specialist Implementation . . . . . . . . . . . . . . . . . . . . . 13
2.2.3 Edge Gateway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Programming Components . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.1 Wearable Communication . . . . . . . . . . . . . . . . . . . . . 15
2.3.2 Programming Implementation . . . . . . . . . . . . . . . . . . 16
2.3.3 Cloud Implementation . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4.1 Sensors’ Presentation . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

xvii
xviii Contents

3 Hybrid Cloud Computing Model for Big Data Analytics

in Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
R. Sheela Daniel, S. Raja, P. Ebby Darney, and Y. Harold Robinson
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.2 Methods and Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4 Bio-Inspired Search Optimization for Intrusion Detection System
in Cognitive Wireless Sensor Networks . . . . . . . . . . . . . . . . . . . . .. 33
M. S. Vinmathi, M. S. Josephine, and V. Jeyabalaraja
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Impact of Security Attacks on Cognitive Network Work . . . . . 34
4.3 Algorithm for Search Optimization . . . . . . . . . . . . . . . . . . . . . 36
4.3.1 Elephant Search Algorithm (ESA) . . . . . . . . . . . . . . . 36
4.4 Implementation and Results . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.5 Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . . 40
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal
Arc Welding Robot: The Surface Roughness, Microstructures
and Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 43
Van Thao Le, Dinh Si Mai, Van Chau Tran, and Tat Khoa Doan
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 44
5.2 Materials and Experimental Methods . . . . . . . . . . . . . . . . . .. 45
5.3 Experimental Results and Discussion . . . . . . . . . . . . . . . . . .. 47
5.3.1 Effects of Depositing Strategies on the Shape
of Thin Walls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.3.2 Surface Roughness Analysis . . . . . . . . . . . . . . . . . . . 48
5.3.3 Microstructure Analysis . . . . . . . . . . . . . . . . . . . . . . 49
5.3.4 Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . 50
5.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
6 An Application of Observer Reconstruction to Estimate Actuator
Fault for DC Motor Nonlinear System Under Effects
of the Temperature and Disturbance . . . . . . . . . . . . . . . . . . . . . .. 55
Tan Van Nguyen and Xuan Vinh Ha
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
6.2 DC Motor Model Formulation . . . . . . . . . . . . . . . . . . . . . . . . 56
6.3 UIO Design for Nonlinear System . . . . . . . . . . . . . . . . . . . . . 58
6.4 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.4.1 Simulation Without Disturbance d ðtÞ ¼ 0 . . . . . . . . . . 65
6.4.2 Simulation with Disturbance d ðtÞ ¼
0:025 randomð2; tÞ . . . . . . . . . . . . . . . . . . . . . . . . .. 66
Contents xix

6.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7 Enhanced Hadoop Distribution File System for Providing
Solution to Big Data Challenges . . . . . . . . . . . . . . . . . . . . ....... 71
A. Essakimuthu, R. Karthik Ganesh, R. Santhana Krishnan,
and Y. Harold Robinson
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.2 Solution for Big Data Challenges . . . . . . . . . . . . . . . . . . . . . . 74
7.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

Part II Intelligent System Design and Applications

8 Electronic Public Distribution System
in Electronic Government . . . . . . . . . . . . . . . . . . . . . . . ......... 87
Israa M. Hayder, Dalshad J. Hussein, Hussain A. Younis,
and Hameed Abdul-Kareem Younis
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.2 E-Government . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
8.4 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
8.4.1 Application Servers (Tomcat Apache) . . . . . . . . . . . . 90
8.4.2 Proposed E-Government Schemes . . . . . . . . . . . . . . . 91
8.5 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.5.1 E-Governments in the Middle East . . . . . . . . . . . . . . 92
8.5.2 Futures This Project . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.5.3 Running the Project . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.5.4 Main Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
8.5.5 User Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.5.6 Main Site for Use . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
8.5.7 Staff Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
8.5.8 Admin Main Page . . . . . . . . . . . . . . . . . . . . . . . . . . 94
8.5.9 Searching Operation . . . . . . . . . . . . . . . . . . . . . . . . . 97
8.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
9 Combining Syntax Features and Word Embeddings
in Bidirectional LSTM for Vietnamese Named
Entity Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Bui Thanh Hung
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
9.2 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
xx Contents

9.3 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

9.3.1 Word Embeddings . . . . . . . . . . . . . . . . . . . . . . . . . . 103
9.3.2 Bidirectional Long Short Term
Memory (Bi-LSTM) . . . . . . . . . . . . . . . . . . . . . . . . . 104
9.4 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
9.4.1 VLSP Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
9.4.2 Experiment Results . . . . . . . . . . . . . . . . . . . . . . . . . . 107
9.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
10 Human Activity Recognition by Deep Convolution Neural
Networks and Principal Component Analysis . . . . . . . . . . . . . . . . . 111
Amir A. Aljarrah and Ali H. Ali
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
10.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
10.3 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
10.4 Deep Learning Architectures . . . . . . . . . . . . . . . . . . . . . . . . . 115
10.4.1 Convolutional Neural Network (CNN) . . . . . . . . . . . . 115
10.4.2 Convolutional Layer (CL) . . . . . . . . . . . . . . . . . . . . . 116
10.4.3 Batch Normalization Layer . . . . . . . . . . . . . . . . . . . . 117
10.4.4 Rectified Linear Unit (ReLU) Layer . . . . . . . . . . . . . . 117
10.4.5 Pooling Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
10.4.6 Fully-Connected Layer . . . . . . . . . . . . . . . . . . . . . . . 118
10.4.7 Softmax Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
10.4.8 Classification Layer . . . . . . . . . . . . . . . . . . . . . . . . . 119
10.5 Architecture of the Method . . . . . . . . . . . . . . . . . . . . . . . . . . 119
10.5.1 Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
10.5.2 Normalization of Data . . . . . . . . . . . . . . . . . . . . . . . . 123
10.5.3 Principle Component Analysis . . . . . . . . . . . . . . . . . . 124
10.6 Experiments and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
10.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
11 Development of a New 6 DOFs Welding Robotic System
for a Specialized Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Truong Trong Toai, Duc-Hoang Chu, and Chu Anh My
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
11.1.1 TIG Welding Method . . . . . . . . . . . . . . . . . . . . . . . . 136
11.1.2 MIG/MAG Welding Method . . . . . . . . . . . . . . . . . . . 137
11.1.3 Plasma Arc Welding (PAW) . . . . . . . . . . . . . . . . . . . 138
11.1.4 Submerged-Arc Welding (SAW) . . . . . . . . . . . . . . . . 138
11.1.5 Laser Beam Welding (LBW) . . . . . . . . . . . . . . . . . . . 138
11.1.6 Industrial Robots . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
11.2 Design of a 6DOFs Welding Robot . . . . . . . . . . . . . . . . . . . . 140
11.2.1 Functional Requirements . . . . . . . . . . . . . . . . . . . . . . 140
11.2.2 The Overall Structure of the Welding System . . . . . . . 140
Contents xxi

11.2.3 The Welding Arm . . . . . . . . . . . . . . . . . . . . . . . . . . 142

11.2.4 The Workspace of the Robot . . . . . . . . . . . . . . . . . . . 142
11.2.5 The Control Software of the Robot . . . . . . . . . . . . . . 144
11.3 Implementation and Testing for the Robot . . . . . . . . . . . . . . . 145
11.3.1 Implementation of the Robot . . . . . . . . . . . . . . . . . . . 145
11.3.2 Accuracy Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 147
11.3.3 Testing and Application . . . . . . . . . . . . . . . . . . . . . . 147
11.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
12 Low Power Reversible Parallel and Serial Binary
Adder/Subtractor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
N. Bhuvaneswary, S. Prabu, S. Karthikeyan, R. Kathirvel,
and T. Saraswathi
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
12.2 Realization of Basic Reversible Logic . . . . . . . . . . . . . . . . . . 152
12.3 Design of Reversible Parallel Adder/subtractor . . . . . . . . . . . . 154
12.4 Design of Reversible Serial Adder/Subtractor . . . . . . . . . . . . . 155
12.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
12.6 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
12.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
13 Computational Fluid Dynamics Analysis of Reduction
of Temperature Difference for a Bundle of Steel Pipes Inside
Annealing Furnace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Lam Hai Dinh, Tan Van Nguyen, and Tu Thien Ngo
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
13.2 Annealing Furnace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
13.2.1 The Original Annealing Furnace . . . . . . . . . . . . . . . . 162
13.2.2 The New Designed Annealing Furnace . . . . . . . . . . . 163
13.3 Mathematical Model and Boundary Conditions . . . . . . . . . . . . 163
13.4 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
13.4.1 The Flow Field and Steel Pipes Temperature
of the Basic Annealing Furnace . . . . . . . . . . . . . . . . . 165
13.4.2 The Flow Field and Steel Pipes Temperature
of the Modified Annealing Furnace . . . . . . . . . . . . . . 165
13.4.3 Effect of Reynolds Number at the Inlet, the
Horizontal and Vertical Steel Spacing on the Steel
Temperature Difference Among Steel Pipes . . . . . . . . 166
13.5 Transient Simulation of Steel Pipes Bundle Temperature
During Annealing Process . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
13.6 Total Deformation of Steel Pipes During
Annealing Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
13.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
xxii Contents

14 A DCNN Based Real-Time Authentication System

Using Facial Emotions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
A. Praveen Edward James, Mun Hou Kit, and T. Anthony Snow Ritta
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
14.2 Design Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
14.2.1 Image Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . 177
14.2.2 Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
14.2.3 Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
14.2.4 Convolution Layer . . . . . . . . . . . . . . . . . . . . . . . . . . 179
14.3 Algorithmic Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
14.4 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
14.4.1 Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
14.4.2 Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
14.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
15 Interdependence of Construction Projects Problems Using PSO . . . 187
Abtehaj Hussein, Rouwaida Hussein Ali, Hafeth I. Naji,
and Naji Muter
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
15.2 Research Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
15.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
15.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Part III IoT Applications in Biomedical Engineering

16 Automated Irrigation System with Pest Detection Using IoT
with OTSU Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
A. V. Kalpana, T. Chandrasekar, S. Rukmani Devi,
and T. C. Jermin Jeaunita
16.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
16.2 Background and Related Work . . . . . . . . . . . . . . . . . . . . . . . 210
16.3 Proposed Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
16.4 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
16.5 Conclusion and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . 215
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
17 IoT Based Telemedicine System . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
L. K. Hema, Rajat Kumar Dwibedi, R. Karthikeyan, and V. Vanitha
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
17.2 Biomedical Automation in Internet-of-Things . . . . . . . . . . . . . 221
17.3 Existing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
17.4 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Contents xxiii

17.4.1 Sensing, Processing, Communication Plane . . . . . . . . 222

17.4.2 Data Concentration and Cloud Computing . . . . . . . . . 222
17.4.3 Analytics, Pre and Post Processing by Physician . . . . 223
17.5 Data Acquisition and Transmission in Ambulance
Using IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
17.6 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
17.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
18 Tracking Greenhouses Farming Based on Internet
of Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Sebastián Gutiérrez, Rafael Rocha, David Rendón,
Juan Carlos Bernabé, Luis Aguilera, and Vijender Kumar Solanki
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
18.2 Aloe Vera Plant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
18.2.1 Soil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
18.2.2 Irrigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
18.2.3 Indoor Cultivation . . . . . . . . . . . . . . . . . . . . . . . . . . 230
18.3 System Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
18.4 Hardware of the System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
18.4.1 Temperature and Humidity Sensor . . . . . . . . . . . . . . . 231
18.4.2 Light Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
18.4.3 Water Sensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
18.4.4 Actuator and Power Supply . . . . . . . . . . . . . . . . . . . . 233
18.5 Data Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
18.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
18.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
19 Impact of Internet of Things and Artificial Intelligence on Human
Resource Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Sonal Pathak and Vijender Kumar Solanki
19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
19.1.1 Employees as Resource . . . . . . . . . . . . . . . . . . . . . . . 240
19.1.2 Management of Human Resource . . . . . . . . . . . . . . . 241
19.2 Traditional Methods of Human Resource Management
Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
19.3 Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
19.3.1 Importance of Artificial Intelligence in Transitioning
Human Resource Management Practices . . . . . . . . . . 249
19.3.2 How Artificial Intelligence is Reinventing Human
Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
19.4 Comparison Between Traditional HRM
and AI-Based HRM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
19.5 Implications of Artificial Intelligence in Human Resources
Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
xxiv Contents

19.6 Limitations of Artificial Intelligence Based HR Tools . . . . . . . 260

19.6.1 Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
19.6.2 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
19.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
19.8 Future Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
20 IoT Based Intelligent Ambulance Monitoring and Traffic
Control System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
J. Jijin Godwin, B. V. Santhosh Krishna, R. Rajeshwari,
P. Sushmitha, and M. Yamini
20.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
20.2 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
20.2.1 Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
20.2.2 Hardware Components . . . . . . . . . . . . . . . . . . . . . . . 272
20.3 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
20.3.1 Preemption Strategy . . . . . . . . . . . . . . . . . . . . . . . . . 273
20.3.2 Path Selection Strategy . . . . . . . . . . . . . . . . . . . . . . . 275
20.4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
20.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
21 IoT Based Emergency Alert System . . . . . . . . . . . . . . . . . . . . . . . . 279
K. Chinnusamy, D. Nandhini, A. S. Subhashri, M. Hemavathy,
V. Baskar, M. Kavimani, and M. Vignesh
21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
21.2 Proposed Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
21.2.1 Wireless Alert Through Internet of Things (IoT) . . . . 280
21.2.2 PIC 16F877A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
21.2.3 Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
21.3 Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
21.3.1 Zigbee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
21.4 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
21.4.1 Geo Fencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
21.5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
21.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
22 An Industrial Internet of Things Approach for Pharmaceutical
Industry Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Deepak Kumar Sharma, Gurmehak Kaur, and Mohita Sharma
22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
22.1.1 Introduction to IoT . . . . . . . . . . . . . . . . . . . . . . . . . . 290
22.1.2 Introduction to Pharmaceutical Industry . . . . . . . . . . . 291
22.2 Background and Research on IoT . . . . . . . . . . . . . . . . . . . . . 291
Contents xxv

22.3 Technologies Used with IoT . . . . . . . . . . . . . . . . . . . . . . . . . 293

22.3.1 Short Range IoT Network . . . . . . . . . . . . . . . . . . . . . 293
22.3.2 Medium Range IoT Network . . . . . . . . . . . . . . . . . . . 294
22.3.3 Long Range IoT Network . . . . . . . . . . . . . . . . . . . . . 295
22.4 Slow Growth in the Pharma Industry . . . . . . . . . . . . . . . . . . . 296
22.5 IoT in Pharma Industry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
22.6 Advantages of IoT in the Pharmaceutical Industry . . . . . . . . . 298
22.6.1 IoT Solves the Previously Discussed Problems
in the Following Ways . . . . . . . . . . . . . . . . . . . . . . . 298
22.6.2 Uses and Advantages of IoT
in the Pharma Industry . . . . . . . . . . . . . . . . . . . . . . . 299
22.7 Architecture of IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
22.8 Challenges and Drawbacks of IoT . . . . . . . . . . . . . . . . . . . . . 302
22.8.1 R&D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
22.8.2 Manufacturing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
22.8.3 Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
22.9 Applications of IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
22.10 Future of IoT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
22.11 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

Part IV Cyber Physical System Framework and Applications

23 Solving the Problem of Fuzzy Transportation Using Linear
Programming and Goal Programming . . . . . . . . . . . . . . . . . . . . . . 313
Hasanain Hamid Ahmed
23.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
23.2 Concept of the Transport Problem . . . . . . . . . . . . . . . . . . . . . 314
23.3 Methods of Solution of Transport Model . . . . . . . . . . . . . . . . 314
23.4 Concept of the Goal Programming . . . . . . . . . . . . . . . . . . . . . 314
23.5 Methods of Solving Goal Programming . . . . . . . . . . . . . . . . . 315
23.6 Fuzzy Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
23.6.1 Functions Affiliation to the Fuzzy Group . . . . . . . . . . 316
23.6.2 Linear Affiliation Function . . . . . . . . . . . . . . . . . . . . 316
23.7 Robust Ranking Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
23.8 Mathematical Model of the Problem of Multi-Objective
Fuzzy Transportation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
23.9 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
23.9.1 Building the Mathematical Model of the Fuzzy
Transport Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 319
23.10 Explanation the Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
23.11 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 328
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
xxvi Contents

24 Effective Solution to Integrate and Control a Heavy Robot Driven

by Hydraulic Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
Chu Anh-My
24.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 331
24.2 Description of the Robotic Cell . . . . . . . . . . . . . . . . . . . . . . . 334
24.3 The Control System Design . . . . . . . . . . . . . . . . . . . . . . . . . . 337
24.4 Validation and Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
24.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344
25 Design and Implementation of Arabic Plagiarism
Detection System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Zahraa Jasim Jaber and Ahmed H. Aliwy
25.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
25.2 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
25.3 Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
25.4 Outline of the Proposed System . . . . . . . . . . . . . . . . . . . . . . . 349
25.5 Implementation and Results . . . . . . . . . . . . . . . . . . . . . . . . . . 354
25.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
26 Application of Artificial Intelligence to Asset Pricing
by Vietnamese Text Declaration . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Tran Ngoc Thang, Dao Minh Hoang, Tran Thi Hue,
Vijender Kumar Solanki, and Nguyen Thi Ngoc Anh
26.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
26.2 Theoretical Preliminary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
26.2.1 Named Entity Recognition . . . . . . . . . . . . . . . . . . . . 361
26.2.2 Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . 362
26.2.3 Conditional Random Field . . . . . . . . . . . . . . . . . . . . 362
26.2.4 Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
26.3 Solution Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
26.3.1 Phase 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
26.3.2 Phase 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
26.4 Computational Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . 365
26.4.1 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
26.4.2 Implementation and Evaluation of Model . . . . . . . . . . 365
26.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
Contents xxvii

27 A Survey on Decentralized Crowdsourcing

Using Blockchain Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
M. Preetha, K. Elavarasi, A. Mani, E. Pavithra, P. Sudharshna,
and S. Rakshini
27.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
27.1.1 Blockchain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
27.1.2 Working of Blockchain . . . . . . . . . . . . . . . . . . . . . . . 374
27.1.3 Smart Contract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
27.1.4 Crowdsourcing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
27.1.5 Consensus Algorithms . . . . . . . . . . . . . . . . . . . . . . . 375
27.2 Literature Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
27.3 Real Time Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
27.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
28 A Study of Student’s Subjective Well-Being Through Chatbot
in Higher Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Shivani Agarwal and Nguyen Thi Dieu Linh
28.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
28.1.1 Background About Chatbot . . . . . . . . . . . . . . . . . . . . 385
28.1.2 Application of Chatbot in Business . . . . . . . . . . . . . . 385
28.1.3 Scope of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . 386
28.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
28.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
28.3.1 Participants and Procedure . . . . . . . . . . . . . . . . . . . . 390
28.3.2 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
28.3.3 Objectives of the Study . . . . . . . . . . . . . . . . . . . . . . . 391
28.3.4 Hypothesis of the Study: . . . . . . . . . . . . . . . . . . . . . . 391
28.4 Result and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
28.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
28.6 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402
About the Editors

Valentina E. Balas, Ph.D. is currently Full Professor in the Department of

Automatics and Applied Software at the Faculty of Engineering, “Aurel Vlaicu”
University of Arad, Romania. She holds a Ph.D. in applied electronics and
telecommunications from Polytechnic University of Timisoara. Dr. Balas is author
of more than 270 research papers in refereed journals and international conferences.
Her research interests are in intelligent systems, fuzzy control, soft computing,
smart sensors, information fusion, modeling, and simulation. She is the
Editor-in-Chief to International Journal of Advanced Intelligence Paradigms
(IJAIP) and to International Journal of Computational Systems Engineering
(IJCSysE), member in Editorial Board member of several national and international
journals and is evaluator expert for national and international projects. She served
as General Chair of the International Workshop Soft Computing and Applications
in seven editions 2005–2016 held in Romania and Hungary. Dr. Balas participated
in many international conferences as Organizer, Session Chair, and member in
International Program Committee. Now she is working in a national project with
EU funding support: BioCell-NanoART = Novel Bio-inspired Cellular
Nano-Architectures—For Digital Integrated Circuits, 2M Euro from National
Authority for Scientific Research and Innovation. She is a member of EUSFLAT,
ACM and a Senior Member IEEE, member in TC—Fuzzy Systems (IEEE CIS),
member in TC—Emergent Technologies (IEEE CIS), member in TC—Soft
Computing (IEEE SMCS). Dr. Balas was Vice-president (Awards) of IFSA
International Fuzzy Systems Association Council (2013–2015) and is a Joint
Secretary of the Governing Council of Forum for Interdisciplinary Mathematics
(FIM), A Multidisciplinary Academic Body, India and recipient of the “Tudor
Tanasescu” Prize from the Romanian Academy for contributions in the field of soft
computing methods (2019).

Vijender Kumar Solanki, Ph.D. is an Associate Professor in Computer Science

& Engineering, CMR Institute of Technology (Autonomous), Hyderabad, TS,
India. He has more than 14 years of academic experience in network security, IoT,
Big Data, Smart City and IT. Prior to his current role, he was associated with

xxix
xxx About the Editors

Apeejay Institute of Technology, Greater Noida, UP, KSRCE (Autonomous)

Institution, Tamilnadu, India and Institute of Technology & Science, Ghaziabad,
UP, India. He is member of ACM and Senior Member IEEE.
He has attended an orientation program at UGC-Academic Staff College,
University of Kerala, Thiruvananthapuram, Kerala & Refresher course at Indian
Institute of Information Technology, Allahabad, UP, India.
He has authored or co-authored more than 60 research articles that are published
in various journals, books and conference proceedings. He has edited or co-edited
14 books and Conference Proceedings in the area of soft computing.
He received Ph.D. in Computer Science and Engineering from Anna University,
Chennai, India in 2017 and ME, MCA from Maharishi Dayanand University,
Rohtak, Haryana, India in 2007 and 2004, respectively and a bachelor's degree in
Science from JLN Government College, Faridabad Haryana, India in 2001.
He is the Book Series Editor of Internet of Everything (IoE): Security and
Privacy Paradigm, CRC Press, Taylor & Francis Group, USA; Artificial
Intelligence (AI): Elementary to Advanced Practices Series, CRC Press, Taylor &
Francis Group, USA; IT, Management & Operations Research Practices, CRC
Press, Taylor & Francis Group, USA and Computational Intelligence and
Management Science Paradigm, (Focus Series) CRC Press, Taylor & Francis
Group, USA.
He is Editor-in-Chief in International Journal of Machine Learning and
Networked Collaborative Engineering (IJMLNCE) ISSN 2581-3242 ; International
Journal of Hyperconnectivity and the Internet of Things (IJHIoT), ISSN 2473-4365,
IGI-Global, USA, Co-Editor Ingenieria Solidaria Journal ISSN (2357-6014),
Associate Editor in International Journal of Information Retrieval Research
(IJIRR), IGI-GLOBAL, USA, ISSN: 2155-6377 | E-ISSN: 2155-6385 . He has
been guest editor with IGI-Global, USA, InderScience & Many more publishers.
He can be contacted at [email protected]

Dr. Raghvendra Kumar is working as Associate Professor in Computer Science

and Engineering Department at GIET University, India. He received B. Tech,
M.Tech., and Ph.D. in computer science and engineering, India, and Postdoc
Fellow from Institute of Information Technology, Virtual Reality and Multimedia,
Vietnam. He serves as Series Editor Internet of Everything (IOE): Security and
Privacy Paradigm, Green Engineering and Technology: Concepts and Applications,
published by CRC press, Taylor & Francis Group, USA, and Bio-Medical
Engineering: Techniques and Applications, published by Apple Academic Press,
CRC Press, Taylor & Francis Group, USA. He also serves as acquisition editor for
Computer Science by Apple Academic Press, CRC Press, Taylor & Francis Group,
USA. He has published number of research papers in international journal
(SCI/SCIE/ESCI/Scopus) and conferences including IEEE and Springer as well as
serve as organizing chair (RICE-2019, 2020), volume editor (RICE-2018), keynote
speaker, session chair, co-chair, publicity chair, publication chair, advisory board,
technical program committee members in many international and national confer-
ences and serve as guest editors in many special issues from reputed journals
About the Editors xxxi

(Indexed By: Scopus, ESCI, SCI). He also published 13 chapters in edited book
published by IGI Global, Springer, and Elsevier. His research areas are computer
networks, data mining, cloud computing, and secure multiparty computations,
theory of computer science and design of algorithms. He authored and edited 23
computer science books in fields of Internet of Things, data mining, biomedical
engineering, big data, robotics, and IGI Global Publication, USA, IOS Press
Netherland, Springer, Elsevier, CRC Press, USA.
 Part I
Distributed Sensor Networks
Chapter 1
Energy Efficient Multi-hop Routing
Techniques for Cluster Head Selection
in Wireless Sensor Networks

G. Hemanth Kumar, G. P. Ramesh, and C. Ravindra Murthy

Abstract Sensors usually operate on battery power, which limits energy consump-
tion. Energy efficient optimization algorithms allow nodes to use smartly without
wasting battery consumption. Hierarchical routing protocol is the finest recognized
protocol for improving power consumption in wireless sensor network. The LEACH
protocol not take into account the remaining energy of the node when selecting the
cluster header, it is possible to select a node with a slightly lower energy as the
cluster header. In this way, the cluster formed by this low-energy node will terminate
prematurely and waste entire network resource. ILEACH is measured one of the
finest of them. To improve service life, sensor nodes with high residual energy and
short distance from the base station (BS) are chosen as cluster head (CH) nodes.
Then intelligently manage these nodes to create clusters to maximize the lifetime
of the WSN and minimize the average energy consumption. The TDMA protocol
is used for intra-cluster communication. In this article, we propose a reform of the
ILEACH protocol by acquaint with cluster communication, in which the cluster
heads are organized in a hierarchical structure, additional optimizing the life of the
WSN. The simulation outcomes illustrate that the improved algorithm in this respect
outperforms the LEACH protocol.

Keywords LEACH · ILEACH · OHILEACH · TDMA

G. Hemanth Kumar (B) · C. Ravindra Murthy

Department of E.I.E, Sree Vidyanikethan Engineering College, Tirupati, Andhrapradesh, India
e-mail: [email protected]
G. P. Ramesh
Department of ECE, St. Peter’s Institute of Higher Education and Research, Avadi, Chennai,
Tamilnadu, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 3

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_1
4 G. H. Kumar et al.

1.1 Introduction

Wireless Sensor Network (WSN) is a group of a variety of sensor nodes with limited
thorough, reckoning, and Communication abilities these sensors are positioned exten-
sively with one or more base stations. WSN offers a wide range of applications such as
pressure, humidity, temperature, military scouting disaster management, and forest
inspection etc. [1]. State-of-the-art setups, sensor nodes with limited battery power
are used randomly. The choice of routing methods is an important issue for the effi-
cient provision of collected data from source to destination. The routing techniques
are used in these category of networks must ensure minimum energy depletion, as
sensor battery replacement is often impossible. The choice of routing methods is
a significant issue for Distribute sensing data efficiently from source to destina-
tion. The routing approaches used in these categories of networks often make it
impossible to replace the sensor battery, so they need to guarantee minimal energy
consumption. Depending on the application and network architecture, for WSN has
proposed and developed numerous energy-efficient routing protocols [2]. However,
the rechargeable batteries used in the WSN will not be recharged, if it is recharged,
so an improvement is required: Therefore, to improve performance, you need to
improve the lifetime of the network and this can be analysed with the available node
energy as data is transferred from the source to the target. Based on the organization
of sensor networks, routing can be categorized as flat, location-based routing and
hierarchical routing [3]. Clustering algorithms offer an energy-efficient technique
to exploit the lifespan of WSNs by splitting the sensor nodes into clusters, which
would then have the option of internally choosing a cluster header. Every cluster
header would then gather packets from all nodes in the cluster and send information
to the central station. The selection of cluster heads is one of the best-known cluster
algorithms i.e. LEACH (Low Energy Adaptive Clustering Hierarchy). Each sensor
node is assigned the probability of becoming a cluster head [4]. Hierarchical routing
offers enhanced energy effectiveness and ascend ability due to its structure. This type
of protocol is separated into network groups, and some nodes are selected as superior
nodes based on specific standards. These superior nodes are known as cluster heads
(CHS), acquire, integrate, and suppress the data acquired from neighbouring nodes,
and eventually compile information to the Base station. The cluster head provides
extra facilities for additional nodes and therefore uses more power than other nodes
in the cluster. Cluster rotation is mutual technique to reduce rotational energy inside
the cluster. The rest HR protocol was anticipated by Zelman et al. [5] well-known as
LEACH. In this document, we suggest the improved LEACH (ILEACH) protocol,
which selects cluster heads based on various thresholds. The probability of the new
cluster head selection is calculated the number of initial energy and adjacent nodes.
On a rotation basis, a headset member receives and sends data from neighbouring
nodes Total results to the base station. Data Acquisition For a given number of sensor
nodes, the number of control and management nodes can be adjusted to systemati-
cally reduce the number of system nodes energy consumption, which increases the
lifetime of the network [6].
1 Energy Efficient Multi-hop Routing Techniques for Cluster Head Selection … 5

1.2 Related Work

The author discussed Comparison of performance between particle swarm optimiza-

tion and genetic algorithms by a new cost function, the determination of which is
to minimize distance within a cluster and optimize network power consumption
[7]. Improved clustering routing algorithm built on LEACH and PEGASIS based
load balancing in WSN. In clustering, the node removes the node from the cluster
head, removes the CH to the base station, and combines clusters with a minimum
of complex distance. This strategy balances the grid energy efficiently [8]. With a
single transition, a trust mechanism that identifies and eliminates malicious nodes
established among sending and neighbouring nodes, hybrid games with unceasing
monitoring and forwarding strategies, which can efficiently decrease error rates
when detecting packet loss on unreliable wireless channels [9]. To suppress attacks
by attackers, the author develop security mechanisms based on monitoring nodes,
which are heterogeneous energy nodes that have only monitoring but no forwarding
capabilities [10]. In the structure of a cluster network, sensor nodes are alienated
into clusters. The Cluster Head is not merely answerable for supervision of sensor
member nodes, but also controls the aggregation and transfer of data among clusters.
The oldest standard WSN clustering protocol is LEACH, which dynamically creates
clusters to optimize network power consumption [11]. M-GEAR is a location-based
protocol that divides the sensor field into four logical areas based on their position
in the sensing field. In this algorithm, if the distance from the sensor node to the
BS exceeds a predetermined threshold and the connection between the sensor node
and the BS is based on the cluster, the centre of the sensitive field has a gateway
sensor node [12]. The HEED algorithm is a common generic routing protocol. In
particular, the HEED algorithm synthesizes the mixture of the residual energy and
the secondary parameters of the node for the periodic selection of the cluster head
channel. In spite of the fact that this algorithm speeds up clustering and reduces inter-
cluster communication costs, the inter-cluster competition eliminates the possibility
of connecting some sensor nodes to the cluster [13–16] .

1.3 Methodology

Assume an opaque sensor network of uniform, energy-constrained nodes the data

should be reported to the aggregation node. In LEACH, integrated with TDMA-based
MAC protocol Clusters and simple “routing” protocols. LEACH divides nodes into
clusters and divides a dedicated node, cluster head, in each cluster is in control for
making and retaining TDMA schedules and other nodes of the cluster are partici-
pating nodes [14]. All participating nodes are assigned TDMA time slots that can be
used to exchange data between the participant and the cluster head. With the exclu-
sion of their time intervals, participants can spend their time in a state of sleep. The
cluster head equals the acquired data of its participants and transfers it to the sink
6 G. H. Kumar et al.

node or other participant nodes for further processing. Each node decides whether
it will become the cluster header independently of the other nodes, so there is no
signalling traffic associated with the choice of cluster header. The protocol is based
on loops, that is, all nodes decide whether to be cluster headers at the same time,
and then the non-cluster header nodes must be associated with the cluster header.
The non-cluster header selects the cluster header based on the received signal level.
A cluster-divided network is a time variable, and the protocol assumes global time
synchronization. Once cluster has been formed, each cluster head selects a random
CDMA code for its cluster. The content that it sends, and what it needs to use by
member nodes.
The protocol is structured in cycles and each cycle is divided into an installation
phase and a stable state phase (Fig. 1.1). The installation phase starts from the node
to the cluster head. In the next stage of advertising, the head of the cluster informs
its neighbours with an Advertising packet. The heads of the cluster use the CSMA
protocol to fight for media without additional conditions for unknown terminal prob-
lems. The non-cluster head node selects the ad package with the strongest received
signal level. In the next cluster setup step, participants use the CSMA protocol again
to notify their cluster heads. After TDMA scheduling randomly selects CDMA codes
and broadcasts information to plan the sub-step. Then, the steady state TDMA mode
begins. Due to conflicts in advertising or joining data packages, the protocol has no
guarantee that non-clustered nodes are part of a cluster. However, this can ensure
that the nodes belong to the maximum one cluster. Open the cluster head during
the whole round and you need to switch member nodes Start with the setup phase,

Fig. 1.1 Organization of LEACH cycles

1 Energy Efficient Multi-hop Routing Techniques for Cluster Head Selection … 7

sometimes in the steady state phase, depending on their location to schedule TDMA
cluster. According to the protocol defined, LEACH will not be able to cover large
topographical areas of square miles, because clusters of heads two miles from the
shell can Not enough energy to reach the receiver. This may be the limit if you can
schedule a cluster header to be sent using other cluster headers. During the cluster
setup stage, the node will randomly generate a number from 0 to 1 and including
0 and 1. Suppose the generated random number is less than the threshold T(n), the
node in the round will become the head of the cluster. The calculation of T(n) is
based on the subsequent formula:
⎧ p
⎪
⎨    , N ∈ G
T(n) = 1 − P r mod 1p
⎪
⎩
0, otherwise

In the above equation, p is the percentage of Cluster nodes take into account the
total number of nodes, i.e. The probability that a node will become the head of a
cluster; r refers to current rounds (cycles), N is the number of nodes; G is a set of
nodes that not come to be cluster head in 1/p cycle. The node selected as the cluster
header is then sent to its neighbouring node as cluster header information, and the
remaining nodes select the cluster to join and notify the corresponding cluster header
in accordance with the strength of the broadcast signal it receives. The cluster head
then creates the TDMA, creates a time slot for each node in the cluster, and sends
them to them in a broadcast form. Thus, each node can send data in its own time
interval, while in another time interval the node goes into sleep mode, which saves
energy. During a stable data transfer phrase, member nodes (non-clustered nodes)
in the cluster will transmit the tracked data to the corresponding cluster heads for a
given time interval. The transmitted phrase can be divided into several frames; each
frame is determined by the number of nodes in the cluster. The data that each node
sends in its own time interval is only part of the frame. At the end of each round, the
heads of the clusters and groups will be re-elected, which requires a certain amount of
energy. To reduce the load on the system, the duration of each round of stabilization
is much longer than the cluster creation time 6. For a cluster header, it remains in a
communication state, so that it can receive the acquired data from nodes in its cluster
every time. Once all the data has been received from the participating nodes, the
cluster head will process data such as data fusion to reduce redundant data. Finally,
the cluster head transmits the merged data to its own cluster head, for non-clustered
nodes they transmit data at their own time interval, and at other times they turn off
the wireless module to save power.

1.4 Results

See Figs. 1.2 and 1.3.

8 G. H. Kumar et al.

Fig. 1.2 Round number versus average energy

Fig. 1.3 Round number versus dead nodes

1.5 Conclusion

OHILEACH is a generic routing algorithm for cluster routing protocols that has
several advantages. Selecting the cluster header causes premature disconnection of
the node, resulting in loss of network attributes. When choosing a cluster head,
consider node energy stops so that low energy nodes do not become cluster heads. We
propose an ILEACH protocol reform that is accustomed to cluster communication.
Cluster communication is organized in a hierarchy to further optimize the life of the
1 Energy Efficient Multi-hop Routing Techniques for Cluster Head Selection … 9

WSN. In this research, we experiential that the OHILEACH protocol attains the best
energy organization intervals, such as the number of rounds compared to the average
energy and the number of rounds compared to the dead node, resulting in the longest
lifetime.

References

1. Pino-Povedano, S., Arroyo-Valles, R., Cid-Sueiro, J.: Selective forwarding for energy-efficient
target tracking in sensor networks. Signal Process. 94, 557–569 (2014)
2. Pantazis, N.A., Nikolidakis, S.A., Vergados, D.D.: Energy efficient routing protocols in wireless
sensor networks: a survey. IEEE Commun. Surv. Tuts. 15(2), 551–591 (2013)
3. Karaki. A.L., Kamal, A.E. Routing techniques in wireless sensor networks: a survey. IEEE
Wirel. Commun. 6–28.
4. Fu, C., Jiang, Z., Wei, W.E.I., Wei, A.: An energy balanced algorithm of LEACH protocol in
WSN. Int. J. Comput. Sci. 10(1), 354–359 (2013)
5. Heinzelman, W.R., Chandrakasan, A., Balakrishnan, H.: ‘Energy efficient communica-
tion protocol for wireless microsensor networks. In: Proceedings of 33rd Annual Hawaii
International Conference on System Science, vol. 2, p. 10 (2000).
6. Bingcai, C., Huazhuo, Y., Mingchuan, Y., et al.: A intercluster multi-hop routing protocol
improved based on LEACH protocol. Chin. J. Sens. Actuat. 27(3), 373–377 (2014)
7. Latiff, N.M.A., Tsimenidis, C.C., Sharif, B.S.: Performance comparison of optimization algo-
rithms for clustering in wireless sensor networks. In: Proceedings of the IEEE International
Conference on Mobile Adhoc and Sensor Systems, Pisa, Italy, pp. 1–4 (2007)
8. Zhang Z., Zhang, X.: Research of improved clustering routing algorithm based on load
balance in wireless sensor networks. In: Proceedings of the IET International Communication
Conference on Wireless Mobile and Computing, Shanghai, China, pp. 661–664 (2009)
9. Liao, H., Ding, S.: Mixed and continuous strategy monitor-forward game based selective
forwarding solution in WSN. Int. J. Distrib. Sens. Netw. 2015, 35978 (2015)
10. Hu, Y., Wu, Y.M., Wang, H.S.: Detection of insider selective forwarding attack based on monitor
node and trust mechanism in WSN. Wirel. Sens. Netw. 6, 237–248 (2014)
11. Heinzelman, W.R., Chandrakasan, A.P., Balakrishnan, H.: Energy-efficient communication
protocol for wireless microsensor networks. In: Proceedings of the 33rd Annual Hawaii
International Conference on System Sciences, Maui, HI, USA, 7 January 2000, pp. 1–10
12. Nadeem, Q., Rasheed, M.B., Javaid, N., Khan, Z.A., Maqsood, Y., Din, A.: M-GEAR:
Gateway-based energy-aware multi-hop routing protocol for WSNs. In: Proceedings of the
2013 Eighth International Conference on Broadband and Wireless Computing, Communication
and Applications (BWCCA), Compiegne, France, 28–30 October 2013, pp. 164–169
13. Lin, C.H., Tsai, M.J.: A comment on “HEED: a hybrid, energy-efficient, distributed clustering
approach for ad hoc sensor networks.” IEEE Trans. Mob. Comput. 5, 1471–1472 (2006)
14. Sipon Miah, M., Koo, I.: Performance analysis of ILEACH and LEACH protocols for wireless
sensor networks. J. lnf. Commun. Converg. Eng. 10(4), 384–389 (2012)
15. Kumar, G.H., Ramesh, G.P., Avadi, C.: Novel gateway free device to device communication
technique for IoT to enable direct communication between homogeneous devices. Int. J. Pure
Appl. Math. 118(16), 565–578 (2018)
16. Ramesh, G.P., Aravind, C.V., Rajparthiban, R., Soysa, N.: Body area network through wireless
technology. Int. J. Comput. Sci. Eng. Commun. 2(1), 129–134 (2014)
Chapter 2
Intelligent Wearable Sensor Band
for Underground Working People

S. Karthikeyan, G. Sethuram Rao, M. S. Kowshik, P. Mohan Raj, G. Vishal,

R. Juliet, P. Swetha, and T. Veronica

Abstract This paper presents an Intelligent wearable sensor band for underground
working people. The security and soundness of laborers are significant for under-
ground individuals. The proposed framework consolidates wearable sensors to quan-
tify physiological and natural parameters. A passage is acquainted with giving infor-
mation preparing, a neighborhood web server, and a cloud association. A wearable
sensor on a laborer and natural sensor on a wanderer that can transmit the information
to the client by means of a door for example server, gives offer notice and cautioning
component for the clients. Live health examination taken for laborers who work in an
underground like Tunnels, Shafts, etc., it has an Individual database of laborers and
contrasts it and current essential tangible qualities separate to workplace informa-
tion. Live update, will screen from the control room and it can direct the specialist if
any medical problem occurs and furthermore can maintain a strategic distance from
the undesirable passing.

Keywords Sensor · Cloud · Natural parameters · Tunnels · Security

2.1 Introduction

The Wearable Sensor Network (WSN) is being prospected in numerous applica-

tions like home security, savvy spaces, ecological checking, combat zone recon-
naissance, target following and also consists of various little, low-fueled, vitality
gained sensor hubs with detecting, preparing the information, and remote corre-
spondence [1]. The Performance of the remote sensor arranges (WSN) in ecological
observing and human services applications has been improved as of late. There-
fore, the client can comprehend the continuous physiological and ecological infor-
mation from neighborhood internet browser or portable applications anyplace and

S. Karthikeyan (B) · G. S. Rao · M. S. Kowshik · P. M. Raj · G. Vishal · R. Juliet · P. Swetha ·

T. Veronica
Department of Electronics and Communication Engineering, Velammal Institute of Technology,
Panchetti, Chennai, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 11

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_2
12 S. Karthikeyan et al.

whenever [2–4]. Wearable body territory arrange (WBAN) is one of the WSN that
is commonly utilized in social insurance situations to see the physiological signs
which may build the way of life, and as needs be wellbeing, for instance, a wearable
band, which computes the heartbeat rate and hemoglobin level present inside the
blood by utilizing a human services observing framework [5]. Aside from medicinal
services applications, WBANs have likewise been familiar with screen situations.
For instance, the work screens inside the meanderer can ascertain the temperature,
gas (like dangerous gases and non-poisonous inside the environment) and weight
esteems for security applications [6–9]. Indoor ecological monitoring is controlled
by the can configuration of a wearable sensor. A sufficient opportunity for both
ecological and physiological surveillance would not be provided there [10]. For eg,
the research unceasingly screens the world and the subject’s intensity for persistent
respiratory disease. Security is amazingly significant for the business work envi-
ronment, particularly for laborers continually exchanging working situations among
indoor and outside. In outside conditions, pressure and other dangerous substance
are destructive to human wellbeing [11]. To hinder laborers from being presented
to any unsafe and risky circumstances, some physiological parameters of laborers
ought to try and be checked; Temperature and pulse rate are the key parameters
considered in current measurement works dependent on WBAN [12]. The primary
usually observed parameters are among different wearable ecological testing appli-
cations, temperature and weight. During this article, we propose a wearable sensor
coordinating system that is appropriate for the financial work setting with related
protection and health applications [13].
The framework design is appeared in Fig. 2.1. The wearable sensor arrange
comprises of various wearable sensors that are equipped for speaking with the cloud
server. The Environmental sensors observing including surrounding temperature,
weight, elevation and gas content; the Wearable sensor is for physiological signs
checking including oxygen level and heartbeat rate. Two portable technologies are
used in our work, including Wi-Fi for short-term transmission of information and
Routers for long-run transmission of information [14]. It empowers the short-go
information to be transmitted at longer separations and associated with the on the
web. Wearable sensors are designed to converse with each other on various subjects
for effective functionality, if appropriate. In addition, an input is executed to store,
save and transfer data to the cloud system. Checked information are frequently shown
from a web server situated inside the passage and a web webpage inside the cloud
server. If a crisis state is unlikely to be observed, the system will give customers alerts
to their Smart Phones, such as Smartphone or Desktop. The remainder of this paper
is written as follows: in Sects. 2.2 and 2.3, the usage of hardware and programming
is provided, Sect. 2.4 provides some test results; in Sect. 2.5, the end and potential
works are finally condensed.
2 Intelligent Wearable Sensor Band for Underground Working People 13

Fig. 2.1 The architecture of the proposed WBAN

2.2 Hardware Components

2.2.1 Rover Implementation

The Rover description shows up in Fig. 2.2a. It fuses a force supply, one Camera,
RaspberryPi3+ Module and 4 ecological sensors. A Rover is appended with
MCP3008 IC, L293D Motor Driver and 2 DC motors operated by battery charging
and voltage regulation regulate the voltage of the battery at steady voltage (6 V).
The Raspberry Pi3+ Model with meanderer, which is utilized for the long-run corre-
spondence at 300 Mbps and it has most extreme band of 5 GHz for WLAN. The
long-range information must be conveyed from the Planet to the remote entrance.
The computer obtains the Worker’s data attached to the wrist. Temperature (LM35),
pressure (BMP180), and gas (such as MQ2 and CO2 ) are four ecological sensors
selected. Such sensors are chosen because of their exceptional, good precision and
low power consumption. Figure 2.2 Gas sensor—MQ2 and CO2 .

2.2.2 Specialist Implementation

Figure 2.2b presents the graph of the Worker. It includes a force supply, and hub
Microcontroller (MCU), and two physiological sensors. A low-power supply to
control the battery voltage at 12 V for the circuit. Internal heat level sensors (LM35)
14 S. Karthikeyan et al.

Fig. 2.2 a. Gas sensor b.

Raspberry Pi 3B+ in Rover
setup c. over all integration
part with Raspberry pi 3B+
2 Intelligent Wearable Sensor Band for Underground Working People 15

Fig. 2.3 Both physiological and environmental

and rate sensors are associated with the MCU by adaptable wires. Both the wellness
parameters are transmitted to the server through remote transmission (WBAN).

2.2.3 Edge Gateway

Passage computer architecture is referred to as Fig. 2.3. It consists of 1 Raspberry Pi

Model 3, remote modules, Fig. 2.2b MCU node with SPO2 sensor, and supply device
for power. The Raspberry Pi operates on the Raspbian platform and is a functional
platform for open-source Linux. This embraces a few distinct dialects that include
Java, Node.js, Python, C, and C++. The low power consumption, which needs just
2 A and 5.5 V power source, That a conveyable force bank can legally regulate. A
Raspberry Pi connects with the Pi to get the wanderer’s remote knowledge.

2.3 Programming Components

2.3.1 Wearable Communication

Wearable connectivity helps to include important health warning notices to some

or more of the on-site staff. Therefore, they can respond to emergency situations as
16 S. Karthikeyan et al.

early as possible without getting gateway alerts. Through rover will relay two forms
of RF packets: packets class 1 and packets class 2. Class 1 packet is general details
regarding the world that are transmitted when no dangerous substance is detected.
This message’s targeting destination is also the portal. Class 2 packet is distributed
after discovery of dangerous conditions.
This packet is additionally distributed because of the connection to certain
connected apps. The sensor node continues tracking environmental data in contin-
uous monitoring mode, without reaching low-power mode. After initialization a rover
can reach idle mode and listen to the RF channel for incoming messages. If data is
provided a data separation feature can verify if the data may be a alert message
from other Safe Nodes with unsafe environmental data. If it is dangerous data, the
consumer will be alerted through the application of a Smart Device. The program
algorithm will then miss the notification and switch to idle mode. In case no informa-
tion is collected, the meanderer must within a period gage and record inline sensor
details. On the off probability that any unsafe situations may be found, a Classifica-
tion 2 packet will be configured and the RF package will be distributed to some or
all others including the entrance. The Class 1 packet is formed and sent to the portal
on the off probability that a hurtful situation is found.

2.3.2 Programming Implementation

The design incorporates five parts: (1) WSN Director of Management (WDM), (2)
Data Process Provider (DPM), (3) Database Manager (DBM), (4) Regional Web
Server, and (5) Internet. For eg, Raspberry pi 3+ and Node MCU, a software written
in Python is used to communicate with the numerous remote modules. Right now,
program peruses information from the wanderer and laborer which pass the infor-
mation to server. The information put away in the neighborhood database can be
recovered later on for additional examination. A community server operating plat-
form that depends on Node.js, HTML, CSS, and JavaScript is built to slowly display
sensor details. Door details can be transferred to the cloud through Wi-Fi, Ethernet,
and organizing cells.

2.3.3 Cloud Implementation

The cloud service is supported in the US technology expert company Global Ocean.
The server operates on Ubuntu 16.04.5 with 2 GB of RAM and 25 GB of plate
capacity. In terms of Node.js and Node-RED, a cloud-put together platform is built,
like the portal web execution. In turn a Mosquito expert is added and configured on the
cloud platform. As an enhancement which can communicate within the platform and
cloud with the MQTT professional. In addition, MySQL database is implemented
on the server for storing of knowledge. To ensure the best possible protection of
2 Intelligent Wearable Sensor Band for Underground Working People 17

information, accreditations are needed if any client wants to access the server data.
The server is configured to attempt once regularly to do a scheduled validation of
the Ubuntu picture to insure the data is not lost.

2.4 Results and Discussion

2.4.1 Sensors’ Presentation

Some continuous evaluation from various sensors worn by one subject is provided
in the Fig. 2.4. The qualities demonstrate that the subject is outside and the wear-
able gadget in wrist. Clearly, when the subject is outdoors, it can be found that the
ambient temperature, beat rate and weight are greater than inside passages. Since
hotter temperature can assimilate more dampness, the weight is lower when the
subject is inside passages. For gas fixation, it is higher in inside passages when
contrasted with outside. Internal heat level and pulse information is additionally
introduced in the figure. For the exception of when the subject is within passages
from just below 35 °C to around 38 °C, the internal heat volume begins to increase.
The subject’s pulse perusing is changed at around 100 beats for every moment. The
Oxygen level in the blood consistently somewhere in the range of 95 and 99% as
would be expected, on the off chance that it goes irregular it will diminish underneath
90%.

Fig. 2.4 Software for video

streaming 1
18 S. Karthikeyan et al.

2.5 Conclusion

Right now, present a keen wearable sensor band for underground working individ-
uals for wellbeing and security estimations. The framework is set up to live both
physiological and ecological information shaping a system from wearable sensors
joined to laborers and give the data to the control space for checking. Highlights
like sensor hub equipment and programming structure, portal and cloud usage are
examined. During this venture, we will stay away from the undesirable demise and
furthermore keep away from the basic issues confronting the laborers.

References

1. Selkar, R.G., Thakare, M.: Brain tumor detection and segmentation by using thresholding and
watershed algorithm. Int. J. Adv. Inf. Commun. Technol. 1, 321–324 (2014)
2. Alam, M.S., Rahman, M.M., Hossai, M.A.: Automatic human brain tumor detection in MRI
image using template-based K means and improved fuzzy C means clustering algorithm. Big
Data Cogn. Comput. 3(27) (2019)
3. Prastawa, M., Bullitt, E., Ho, S., Gerig, G.: A brain tumor segmentation framework based on
outlier detection. Med. Image Anal. 8, 275–283 (2004)
4. Devkotaa, B., Alsadoona, A., Prasada, P.W.C., Singhb, A.K., Elchouemic, A.: Image segmen-
tation for early stage brain tumor detection using mathematical morphological reconstruction.
Procedia Comp. Sci. 125, 115–123 (2018)
5. Mathur, N., Meena, Y.K., Mathur, S., Mathur, D.: Detection of Brain Tumor in MRI Image
Through Fuzzy-Based Approach in High-Resolution Neuroimaging-Basic Physical Principles
and Clinical Applications. Rijeka, Croatia, InTech (2018)
6. Arzoo, M., Prof, A., Rathod, K.: K-means algorithm with different distance metrics in spatial
data mining with uses of NetBeans IDE 8.2. Int. Res. J. Eng. Technol. 4, 2363–2368 (2017)
7. Karthik, R., Menaka, R.: A multi-scale approach for detection of ischemic stroke from brain
MR images using discrete curvelet transformation. Measurement 100, 223–232 (2017)
8. Wu, Y., Zhu, M., Li, D., Zhang, Y., Wang, Y.: Brain stroke localization by using microwave-
based signal classification. In: IEEE International Conference on Electromagnetics in Advanced
Applications, pp. 828–831 (2016)
9. Ay, H., Furie, K.L., Singhal, A., Smith, W.S., Sorensen, A.G., Koroshetz, W.J.: An evidence-
based causative classification system for acute ischemic stroke. Ann. Neurol. 58(5), 688–697
(2005)
10. Database, Brainweb: http://www.bic.mni.mcgill.ca/brainweb/. Accessed 10 Sept 2018
11. Ramesh, G.P., Aravind, C.V., Rajparthiban, R., Soysa, N.: Body area network through wireless
technology. Int. J. Comput. Sci. Eng. Commun. 2(1), 129–134 (2014)
12. Nithya, V., Ramesh, G.P.: Wireless EAR EEG signal analysis with stationary wavelet transform
for co channel interference in schizophrenia diagnosis. In: Balas, V., Kumar, R., Srivastava, R.
(eds.) Recent Trends and Advances in Artificial Intelligence and Internet of Things. Intelligent
Systems Reference Library, vol 172. Springer (2020)
13. Fiebach, J.B., Schellinger, P.D., Geletneky, K., Wilde, P., Meyer, M., Hacke, W., Sartor,
K.: MRI in acute subarachnoid haemorrhage; findings with a standardised stroke protocol.
Neuroradiology 46(1), 44–48 (2004)
14. Fiebach, J.B., Schellinger, P.D., Jansen, O., Meyer, M., Wilde, P., Bender, J., Hähnel, S.: CT
and diffusion-weighted MR imaging in randomized order. Stroke 33(9), 2206–2210 (2002)
Chapter 3
Hybrid Cloud Computing Model for Big
Data Analytics in Organization

R. Sheela Daniel, S. Raja, P. Ebby Darney, and Y. Harold Robinson

Abstract Hybrid cloud computing model is constructed to produce the infrastruc-

ture for the organization. The service oriented architecture is used to support the big
data for the business intelligent solutions. The architect decision has provided the
Oracle based big data application with Hadoop technique. The cloud computing has
the implementation of traditional IT for high implementation of traditional IT for
high implementation time. The security is the main problem of the big data analytics
for implementing Governance and risk management. The factors for affecting the
cloud computing has identified for the hybrid cloud computing.

Keywords Private cloud · Public cloud · Hybrid cloud · Security · Organization ·

Risk management · Oracle · Big data

R. S. Daniel
Department of Civil Engineering, SCAD College of Engineering and Technology,
Cheranmahadevi, India
e-mail: [email protected]
S. Raja
Department of Mathematics, SCAD College of Engineering and Technology, Cheranmahadevi,
India
e-mail: [email protected]
P. E. Darney
Department of Electrical and Electronics Engineering, SCAD College of Engineering and
Technology, Cheranmahadevi, India
e-mail: [email protected]
Y. H. Robinson (B)
School of Information Technology and Engineering, Vellore Institute of Technology, Vellore, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 19

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_3
20 R. S. Daniel et al.

3.1 Introduction

The cloud computing trend and how cloud computing will become the essential part
of the organization’s infrastructure implementation in the future [1]. Hybrid cloud
computing vital for organizations has the Hybrid cloud will become essentials for
the organizations in future because it will offer a single cloud solution with the
mixture of public and private cloud computing services [2]. Organizations will be
concentrating more on data and application integration, relating exterior and interior
applications with hybrid computing implementations [3]. Cloud computing service
provider assistance on cloud consumption has the Organizations will need the help
of cloud computing service to liaise with IT department to discuss the purchasing
and cloud solution options [4]. Figure 3.1 demonstrates the cloud computing models
that the cloud computing has separated into 3 models of public cloud, private cloud
and hybrid cloud. It demonstrates the control for three types of services, abstraction
and the purpose for producing the efficient modelling.
Big data getting developed and established that EMC implements that many
companies are utilizing different sources to accomplish superior consideration of
their, customers, associates, employees and processes [5]. Figure 3.2 demonstrates
the produced oracle solution for the big data related issues that the HDFS have the

Fig. 3.1 Cloud computing models

3 Hybrid Cloud Computing Model for Big Data … 21

Fig. 3.2 Oracle solution for Big data

oracle NoSQL database and the enterprise application for acquire section, Hadoop
for organize section that has the oracle big data connectors, The data warehouse and
In database analytics has been demonstrated for the analyze and analytic applications
for decide implementations [6].
Companies are exploring and processing more than organized and transaction
based data which contains videos, social networking, RFID logs, sensor networks,
search indexes, environmental conditions and medical scans [7]. Universal method-
ology has discussed to develop strategies for big data to capture the right metadata
which will be analysed, maintained and store quickly [8]. It will provide the organiza-
tions to understand the large data to design business intelligence solutions which can
later link to build the strong Service Oriented Architecture (SOA) [9]. The variety
of digital sources which big data can support and organization can fully utilized
different pieces of the detailed information from different sources [10]. Figure 3.3

Fig. 3.3 Architect decision

22 R. S. Daniel et al.

Fig. 3.4 Traditional IT with cloud computing

illustrates the Architect decision for providing oracle and big data appliance with
Hadoop and it is connected with the oracle and big data to produce the external table
for oracle SQL developer with SQL tools.
Big data supports different type of sources like media and entertainment, health-
care, life sciences, video surveillance, transportation and logistics [11]. It will help
organization’s those trading globally which will provide the organization’s to make
their decisions, efficiently and well-timed based on their transactions [12]. There
are several differences for the cloud computing and traditional IT Eco system that
the service providers have the connection with the data centres and networks for
providing the support for hardware and software specifications with high amount
of implementation period and customization about the expensive for mobility and
adaptability [13]. The cloud eco system has the multi user based environment with
the service providers. The on demand and virtualization are the other parameters of
cloud based systems and it is demonstrated in Fig. 3.4.

3.2 Methods and Materials

The challenges involved in the implementation of big data as organization go further

along with it. One of the main challenges involved the lack of staff with big data and
analytical skills and second one it’s not limited to local contexts, it’s more global [14].
The big data contains several prospects but dealing with fast progresses enhances
challenges, budget and complications [15]. The organization’s thinking about big
data as problem or an opportunity, only 30% reflect that big data as problem because
only large volume of data is not only the solution for some organizations. On the
other hand, 70% of the organization sees it as a prospect to thorough studies of
their data which contains new evidences about their consumers, associates, costs,
processes and procedures which can be used for business proficiencies. Figure 3.5
3 Hybrid Cloud Computing Model for Big Data … 23

Fig. 3.5 Oracle integrated information architecture

demonstrated the integrated format for information with Oracle for big data cluster
with DBMS, NoSQL and HDFS. The security and governance for the management
with the organize for Hadoop based MapReduce network.
Infrastructure as a Service (IaaS) is the service provide infrastructure as a service,
in which company outsource the hardware, networking gears, servers and storage.
Companies like Microsoft, Amazon and Telstra and other sel.\lice providers delivers
its personal equipment and responsible for administration, monitoring, marinating
and running it. Application platform as a Service (ApaaS) is the service provides the
whole application as a service; in which company lease the hardware, networking
gears, servers, internet and storage. Companies like Google App Engine, Force.com
and application space to charge virtualised severs and related facilities and services
for developing and running applications. Software as a Service (SaaS) is the service
provides the software application as a service, in which company uses the applica-
tion over the internet. Companies like Google, salesforce.com and other application
service providers deliver OneSource and licensed software and applications. Service
Oriented Architecture (SOA) is the service which allows companies to maintain
and support communication between two different computing services or entities.
Communication between two commuting entities and services are independent of
each other, like Amazon Flexible Payment and PayPal. Delivery models of cloud
computing is about the different delivery models of cloud computing which includes
Public, Private and Hybrid Cloud. Figure 3.6 demonstrates the Physical architect for
combined analytics that has the oracle big data appliance, oracle big data connectors,
Oracle Exadata and Oracle Exalytics.
24 R. S. Daniel et al.

Fig. 3.6 Physical architect for combined analytics

3.3 Results and Discussion

Advanced analytics is the combination of different procedures and tools categories

which includes prognostic analytics, data mining, statistical analysis, complex SQL,
data visualization, artificial intelligence, natural language processing and database
approaches which maintain analytics. Visualisations are the concept that the big data
provides Advance Data Visualisations (ADV) which characterizes millions of data
points. ADV supports different sets of data varieties and data structure which is not
adjustable into computer screen. Real time Organization can observer and measure
different business processes more regularly because of the bid data tools and analysis.
Structure and unstructured data: Organizations are handling unstructured and struc-
tured data more efficiently than ever before. This provides organizations improved
data mining and fraud finding techniques to process under layers of organizational
data. Hadoop Distributed File System (HDFS) demonstrates the big data is varied in
terms of data types, HDFS will help to convert these different data types into data
structure so it can fit into the traditional data base management system. MapReduce
will provide distributed parallel processing for different types of data collections
which will make analysis easy on big data.
Private cloud deployed on company’s own site or data centre belongs to the service
provider. The private cloud makes logic to be deployed and contain within the zones
of company infrastructure which will provide more security and control over the
company’s infrastructure. It will provide access control to companies which will
assure certified access to infrastructure and its applications, hardware utilisation and
assets management. Vendor hosted or partner allowed the companies to move their
applications to the service provider secure data centres. Companies will be saving
cost rather than investing and buying the new technology not only equipment but
also release pressure on in house resources and it is demonstrates in Fig. 3.7.
Public cloud allows organization to utilized infrastructure and applications
through the internet. Organization’s end users will be accessing applications, servers,
3 Hybrid Cloud Computing Model for Big Data … 25

Fig. 3.7 Private Cloud

networks and services on demand through web browser on their laptops or worksta-
tions. Companies no longer required buying the expensive equipment, when appli-
cations and services are accessible through internet on little cost. The main problem
moving with public cloud is security and regulatory issues. and it is demonstrated in
Fig. 3.8.
Hybrid will provide the combination of both private and public cloud to the orga-
nizations. It will provide the organizations the choice to switch between private and
public on peak processing times, when resources required over the network. The
companies will deployed the application which is more important and essential on
the private cloud behind the firewall, like less security services and applications can
place on public clouds. Hybrid model will allow the companies to maintain and
control peak time periods, save costs and security on precarious applications and this
model is demonstrated in Fig. 3.9.
Cloud-centric deployment has the Organizations will be focusing and exploring
different options and prospects to transfer current enterprise workload on application
infrastructure or cloud system. To make most out of the cloud model organizations
need to plan applications deployment where most of the workload is vastly adjustable
and scalable. Upcoming operational models and data centre based on cloud model
that in coming years most of the organization will be designing and developing its
own cloud computing mode to support its operational model and data centres. Future
developments, issue identification and analysis in company A is the enormous big
mining giant which has several mines located all over the world. Each site has its
26 R. S. Daniel et al.

Fig. 3.8 Public cloud

Fig. 3.9 Hybrid cloud

own business rules and doing things according to their requirements which is not
ideal if company want to implement a global and standardise solution. Company A
is using Oracle stacks which are using Oracle Warehouse Builder (OWB), Hyperion
Performance Suite, Business Intelligence Tool (OBIEE) and Ellipse as ERP system
(Also based on Oracle) and oracle databases. The company is expanding quickly
in each area which involves different subject areas like assets management, mobile
3 Hybrid Cloud Computing Model for Big Data … 27

assets, procurement, sales, commercial, planning, finance, HR, budgeting and fore-
casting. The result of that data sets are getting bigger and larger, transactions and
operations are expanding enormously every day which need proper analysis, data
quality and to make business operations uniform and standardise each site and loca-
tion. A Company’s corporate affairs and legal business functions and department are
getting busier than ever before which needs to look for every media to get the right
and proper information at right time. To address the future requirements,
Company A needs to expand its data warehouse and business intelligence oper-
ations to make most out the business value which comes from data. Company A
has different sites in Canada, Chile, Peru, Argentina, South Africa and Australia,
each site has its own central offices where they perform all the business operations
according to their needs which at some stage need to standardise and make it global
to produce a single solution. To address large data sets, RFID Assets, social media
for legal and corporate laws, so company will analyse them properly need to address
some of the changes in the future. The process is explaining and hovering curtains
from big data analytics which can provides benefits to Company A explain in the
table in many ways.
The unstructured and structure data, social media, news and blogs from anywhere
from the internet are demonstrated in the detail. It is also explained about the inte-
gration of the data to address these requirements before entering into the data ware-
house environment. Company A needs an integrated architecture which will support
both traditional data warehouse and big data implementation. For the best result big
data needs to be added into the existing data warehouse platform, so analysis can
be used for the business intelligence system for better visualisation. Company A
is using heavily Oracle products, even the hardware The Solaris boxes which will
make implementation and integration of big data into existing data warehouse system
easier if Company A will keep the same vendor.
The different data sources are captured and stored which later processed into
traditional DBMS (OLTP), files, and distributed clustered systems such as NoSQL
and Hadoop Distributed File System (HDFS). Some of the architecture from big
data appliances to the reporting level that the Company A is already using Oracle
stream for warehousing and reporting purposes, the only piece is missing to bring
Big data analysis and tool in place for detailed level of analysis. The different data
sources are captured and stored on big data appliances, this unstructured information
will process into the format by using Hadoop, MapReduce and NoSQL which can
make sense for RDBMS system to integrate Data Warehouse system in place. This
information later processed for reporting and visualisation purposes.
Sandboxes are the copies or segment of huge datasets. For the experimentation
purposes Company A needs sandboxes to store segment of massive Company A
datasets from different data sources. It will provide the data analyst to play with the
data and do the analysis with any tool they want to use to make sense of the data to
the business. The data stored on the sandboxes store for certain period of time then
discard the data after the analysis. Company A corporate private cloud implements the
Private cloud deployed on Company A’s own site or data centre belongs to the service
provider. It will give company the option where small number of business users are
28 R. S. Daniel et al.

Fig. 3.10 Factors affected for cloud computing

located provide application and data access through clouds rather than spending and
investing a small data centre for 50–100 users. Figure 3.10 demonstrates the Factors
affected for cloud computing that the neutral has the highest amount of affected
factors other than Laws.
Moving big data into clouds utilizes the Private cloud implementation will also
give Company A the benefit to move Big data scenarios into the cloud in future if it has
too. If information is not located inside the company and it is coming from outside the
firewall then it can be stored on the cloud for analysis. This information is relatedto
tweets, feeds, internet, news, web data and events. Expensive maintaining cost when
small number of business users on different site Some of the sites in Peru and Chile
have only 30 users which will not going to increase in next 20 years, Company A has
got data centres and application centres, which is costing maintaining and licensing
cost. Private Cloud needs to be implemented for small number of business users,
which will provide low maintenance cost and no licensing cost In future big data
analysis from different sites can place on the clouds and used in EDW system for
analysis.
Resultant ICT impact and business or regulatory requirements has been described
that customer data is the precious assets for the organizations which required vigilant
governance and protection. Managing big data is not easy and it will require lot of
precaution and architectural framework to make sure the quality and security of the
underlying data. The data governance issues related to more towards security. Big data
provide the detailed level of analysis and sensitive information to trade secret, finan-
cial archives and knowledgeable belongings. Make information available centrally
will provide informal and treasured target for attackers or hackers which can harm
the status of the organizations. Figure 3.11 demonstrates the Internet based cloud
computing factors that have the email in spite of internet related cloud computing
model.
3 Hybrid Cloud Computing Model for Big Data … 29

Fig. 3.11 Internet based cloud computing factors

The application policy and governance program offer organization best practices
when providing business approaches and application schemes. The cloud computing
providers have extensive regularity ambiguities and absence of standardisation which
needs to address that sort of issues as well. Some of the companies will provide corpo-
rate companies to do penetration testing which will highlight where the loop poles
are in the implementation. Companies also implement their own security controls
deployed on its own sites for cloud computing to look after the security and privacy
issues. In relation to private cloud implementation for Company A requirement are
given below:
• Cloud service provider
• Mutual agreement on contract
• Security and privacy controls form Telstra
• Security and privacy controls within company A for the security
• Penetration testing required in certain time frames which provides by different
companies to find loop holes
• Applications needs to decide by the infrastructure team to place on private cloud
• Data ownership
• Make sure the regulations from the country regularity authority body if exists
• Redesign and revisited the privacy and security policy framework.

3.4 Conclusion

With all these benefits there are some challenges when it comes to big data and cloud
computing. Big data contains data ownership, data quality, data governance and
lack of staff type of challenges which require proper consideration, investigation and
30 R. S. Daniel et al.

deployment as part of big data implementation. On the other hand, in cloud computing
also contains security, privacy and data ownership types of obstacles which also
need proper analysis and deployment for cloud implementation. One of the main
obstacle facing in the implementation of big data and clouds are governance and
privacy issues which involves re-designing organization’s policies, impact analysis
and mitigation strategies for the effective execution. Massive research, developments
and investments are still improving the big data technology by different IT vendors
which can predict the long lasting future of big data and cloud technologies in IT
and business world.

References

1. Hussain Iqbal, M., Rahim Soomro, T.: Big data analysis: apache storm perspective. Int. J.
Comput. Trends Technol. 19(1), 9–14 (2015)
2. Krishnan, R.S., Julie, E.G., Robinson, Y.H., Kumar, R., Son, L.H., Tuan, T.A., Long, H.V.:
Modified zone based intrusion detection system for security enhancement in mobile ad-hoc
networks. Wirel. Netw. 1–15 (2019)
3. Akuma, S., Iqbal, R., Jayne, C., Doctor, F.: Comparative analysis of relevance feedback methods
based on two user studies. Comput. Hum. Behav. 60, 138–146 (2016)
4. Lopez, J., Rios, R., Bao, F., Wang, G.: Evolving privacy: from sensors to the Internet of Things.
Future Generat. Comput. Syst. 75, 4657 (2017)
5. Balaji, S., Julie, E.G., Robinson, Y.H., Kumar, R., Thong, P.H., Son, L.H.: Design of a security-
aware routing scheme in mobile ad-hoc network using repeated game model. Comput. Stand.
Interfaces 66 (2019)
6. Gong, B., Veeravalli, B., Feng, D., Zeng, L., Wei, Q.: CDRM: a cost-effective dynamic repli-
cation management scheme for cloud storage cluster. In: 2010 IEEE International Conference
on Cluster Computing, September 2010, pp. 188–196 (2010)
7. Li, M., Wang, X., Gao, K., Zhang, S.: A survey on information diffusion in online social
networks: models and methods. Information 8, 118 (2017)
8. Mittal, H., Saraswat, M.: An optimum multi-level image thresholding segmenta-tion using
non-local means 2d histogram and exponential kbest gravitational search algorithm. Eng. Appl.
Artif. Intell. 71, 226–235 (2018)
9. Robinson, Y.H, Jacob, I.J., Julie, E.G., Darney, P.E.: Hadoop MapReduce and dynamic intelli-
gent splitter for efficient and speed transmission of cloud-based video transforming. In: IEEE—
3rd International Conference on Computing Methodologies and Communication (ICCMC),
pp. 400–404. IEEE (2019)
10. Mika, V., Daniel Graziotin, M., Kuutila, M.: The evaluation of sentiment analysis—A review
of research topic, venues, and top cited papers. Science Direct 27, 16–32 (2018)
11. Hogenboom, F., Frasincar, F., Kaymak, U., De Jong, F.: An overview of event extraction from
text. In: Workshop on Detection, Representation, and Exploitation of Events in the Semantic
Web (DeRiVE 2011) at tenth international semantic web conference (ISWC 2011), vol. 779,
pp. 48–57 (2011)
12. Thomaz, G.M., Biz, A.A., Bettoni, E.M., Mendes-Filho, L., Buhalis, D.: Content mining frame-
work in social media: A FIFA world cup 2014 case analysis. Inf. Manage. 54(6), 786–801
(2017)
13. Krishnan, R.S., Julie, E.G., Robinson, Y.H., Raja, S., Kumar, R., Thong, P.H., Son, L.H.: Fuzzy
logic based smart irrigation system using Internet of Things. J. Clean. Prod. 252(10) 119902
(2020)
3 Hybrid Cloud Computing Model for Big Data … 31

14. Reddick, C.G., Chatfield, A.T., Ojo, A.: A social media text analytics framework for double-
loop learning for citizen-centric public services: a case study of a local government facebook
use. Gov. Inf. Q. 34(1), 110–125 (2017)
15. Lakshminarayanan, K., Krishnan, R.S., Julie, E.G., Robinson, H.Y., Kumar, R., Son, L.H.,
Hung, T.X., Samui, P., Ngo, P.T.T., Bui, D.T.: A new integrated approach based on the iterative
super-resolution algorithm and expectation maximization for face hallucination. Appl. Sci. 10,
718 (2020). https://doi.org/10.3390/app10020718
Chapter 4
Bio-Inspired Search Optimization
for Intrusion Detection System
in Cognitive Wireless Sensor Networks

M. S. Vinmathi, M. S. Josephine, and V. Jeyabalaraja

Abstract In modern times, security concerns have been susceptible. Kaspersky

intelligence believes which denial of service attacks has a vital effect on the network
relative to many other remote security threats. The current work focus on DoS attacks
in cognitive radio networks (CRNs). The presence of malicious users is threat for
enhancing the effective spectrum utilization and this threat may be an active or
passive. In an active attack malicious user will deliberately upset the primary user
framework. A passive attack relates to the circumstance in which a malicious attack
endeavors to translate source data without infusing any data or attempting to alter
the data i.e., it will tune in to the transmission without cooperating with other users.
The network consists of two users such as primary users and secondary users where
the main impact occurs on primary users. The network performance parameters such
as packet delivery ratio, packet loss ratio, bandwidth usage and end to end delay are
analyzed in CRN to detect DoS attacks.

Keywords Bio-inspired algorithm · Search optimization · Cognitive WSN ·

Intrusion detection · Elephant search algorithm

M. S. Vinmathi (B)
Department of Computer Science Engineering, Dr. M.G.R. Educational and Research Institute
(Deemed to Be University), Maduravoyal, Chennai 600095, India
e-mail: [email protected]
M. S. Josephine
Department of Computer Applications, Dr.M.G.R. Educational and Research Institute (Deemed to
Be University), Maduravoyal, Chennai 600095, India
e-mail: [email protected]
V. Jeyabalaraja
Department of Computer Science Engineering, Velammal Engineering College, Surapet, Chennai
600066, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 33

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_4
34 M. S. Vinmathi et al.

4.1 Introduction

The Cognitive Radio (CR) is one of the emerging technologies in wireless commu-
nication focused at enhancing the methods by which the radio is utilized. The drastic
increase in the number of users, data rates and wide range of applications in wireless
communication systems by the users results in scarcity of the spectrum. The spectrum
needs to be utilised effectively in order to make use of the available limited resources.
The key interest on spectrum utilization is created in cognitive radio networks. The
Primary Users (PUs) are the legitimate user frequency band and the Secondary Users
(SUs) are the unlicensed users or opportunistic users awaiting to share the spectrum
opportunistically when the primary user is not using it. This facilitates the spectrum
assignment to the opportunistic users without resulting interference to the licensed
users [1] intensity.
Figure 4.1 depicts the general architecture of Intrusion Detection system in CRs.
Input module receives some of the information like type of signals, TCP information,
agreement parameters, number of packets sent, receive and drop, time delay or arrival
etc. The monitoring module consists of previous historical data of input module and
stores in the routing table to find which kind of attack occurs in the network [2]. The
Detection module detects the attacks based on the monitoring information stored.
Some of the attacks detected are PUE, jamming, selfish attacks and flooding attacks.
Also, it focus on two operations namely finding the cooperative location and to
make reliable system against attacks. The output module enumerates the attack alert,
specify the attack type, the source of attack and finally the victim of attack [3].

4.2 Impact of Security Attacks on Cognitive Network Work

In Fig. 4.2, the vulnerable effects of the security attacks are shown. A definitive
objective of deploying CR systems is to address the spectrum under-utilization that
is caused by the current settled spectrum usage strategy. By powerfully getting to
the spectrum “openings”, the SUs can recover these generally squandered spectrum
resources. However, PUE attackers may take the spectrum “openings” from the SUs,
prompting consistent bandwidth wastage [4].
On a basic level, the secondary administrations in CR organizes naturally have no
security that they will have stable radio resource due to the idea of dynamic spectrum.
The presence of these attacks fundamentally expands the connection unreliability of
CR systems [5].
In Denial of Service, consider PUE attacks with high assaulting recurrence; at that
point the attackers may involve in occupying large number of the spectrum openings.
The SUs will have inadequate Bandwidth for their transmissions, and thus, a portion
of the SU administrations will be denied. In the worst scenario, the CR network
may even discover no channels to set up a typical control channel for conveying the
control messages [6].
4 Bio-Inspired Search Optimization for Intrusion Detection System … 35

Fig. 4.1 General scenario of

IDS in cognitive radio
networks

Fig. 4.2 Vulnerable impacts of security attacks in CRN

Figure 4.3 depicts the types of intrusion detection with respect to CRNs. They are
classified in to two types namely misuse detection and anomaly detection. Both the
cases are vulnerable in nature. Many attacks Syn flooding attacks are some examples
of misuse and anomaly detection [7, 8].

Fig. 4.3 IDS classification

in CRN
36 M. S. Vinmathi et al.

4.3 Algorithm for Search Optimization

4.3.1 Elephant Search Algorithm (ESA)

The ESA has a place with the gathering of contemporary met heuristic inquiry
enhancement calculations [9]. This calculation impersonates the conduct and quali-
ties of an elephant, and its system depends on double hunt instrument, or the pursuit
operators can be isolated into two gatherings [10]. Elephants live in gatherings which
are bunch is separated interested in a few tribes below the initiative most seasoned
primary gathering. The ESA copies the fundamental attributes and highlights the
crowd source of elephants.
Elephants are extraordinary, wherever the male elephants want to survive inside
segregation and females likes gatherings; the improvement is measured and answer-
able for the objectives of the investigation. Apparently, the algorithm has three
fundamental attributes as compelling inquiry improvement strategies;
1. The hunt procedure iteratively refines the answer for optimal ideal arrangement;
2. Head female elephants lead escalated nearby inquiries at places, wherever
superior likelihood of result the most excellent arrangement was normal
3. The male elephant encompass obligations through investigations out of the
neighbourhood.
Elephants contain a few highlights in addition to attributes which create the
motivation procedure through the natural conduct are significant [11, 12].
ESA is depicted live respectively below the authority of the most seasoned, assume
that x elephant in tribe cl be able to be portrayed by accompanying arithmetical
equations:
 
E L new,cl,x = E L cl,x + y · E L Best,cl − E L cl,x · z (4.1)

where are the new and old phase for the elephant x in clan cl. respectively and y
m[0, 1] as shown in Eq. 4.1.
The Male elephants which are grown-up disregard their family and exist in the
detached region. It circumstance be able to be reenacted through isolating adminis-
trator to take care of complex streamlining issues. To get better the pursuit capacity
people with the most noticeably awful wellness case will actualize the isolating
administrator as indicated by the accompanying Eq. 4.2:

E L wor,cl = E L Min + (E L Max − E L Min + 1) · Rnd (4.2)

where E L new,cl,x , E L cl,x are assigned to find the upper and lower bound position
of each elephant, represent the worst elephant in clan cl, and Rnd m[0, 1] is assigned
for stochastic distribution. At the last point, ESA is created with the portrayal of
tribe refreshing and isolating administrator. The Eq. 4.2 is used to find the malicious
4 Bio-Inspired Search Optimization for Intrusion Detection System … 37

Plain Text

DES Encryption Key 1

DES Encryption Key 2

DES Encryption Key 3

Cipher Text

Fig. 4.4 Data encryption in symmetric level

Table 4.1 Algorithm for encryption

Step 1: Ctext = EncryptK3(DecryptK2EncryptK1 (pltext)))
DES encrypts with K1, DES decrypt with K2,then DES encrypt with K3
Decryption is the reverse:
Step 2: Pltext = Decrypt K1 (EK2 (DecryptK3 (Ctext)))
i.e., decrypt with K3, encrypt with K2,and then decrypt with K1

node by monitoring approach [7] which is in turn utilized for developing intrusion
detection system.
The symmetric data triple data encryption algorithm (Fig. 4.4) is used for authen-
tication during a feature extraction process to transmit data in an encrypted manner.
The algorithm consists of a 64-bit plain text key. The cryptographic text is obtained
as well as the XOR is really the perfect block of plain text with triple DES operation.
There are triple keying options used as per the requirement. Table 4.1 defines the
encryption algorithm for security and protected data transmission [13, 14].
In any feature extraction method, this encryption algorithm is used to remove
attacks that are considered to be intermediate in the network.

4.4 Implementation and Results

A network scenario with an area of 1000 m * 1000 m, consists of maximum of 50

nodes (users) for simulation. Also, we considered 20 Primary Users, 20 Secondary
Users and an attacker/intruder for analysing intrusion detection in CRNs. Figure 4.5
describes about the data transmission scenario form source to destination and Fig. 4.6
depicts the intrusion detection by ESA algorithm.
Figure 4.7 shows the encrypted keys for each node. The encryption is performed to
transmit the data to provide an additional security. The efficiency of IDS is calculated
38 M. S. Vinmathi et al.

Fig. 4.5 Data transmission from source to destination through neighbouring nodes

Fig. 4.6 Intrusion detection scenario

form simulation results using network simulator. The detection rate is evaluated as
shown in Eq. 4.3.

Dr = Mean dr op /Tn (4.3)

4 Bio-Inspired Search Optimization for Intrusion Detection System … 39

Fig. 4.7 Generation of encrypted keys for data transmission

where, Dr is the detection rate, Mean drop is the drop of packets in failed sessions
and Tn is the sequence of transmission with n rounds.
Figure 4.8 shows the packet delivery ratio for 50 nodes. The results shows that the
packet loss is high as compared in normal scenario. Figure 4.9 shows the throughput
for 50 nodes. The results depicts that there is drastic reduction in throughput during
the intrusion detection.

0.8
Packet delivery raƟo

0.7
0.6 Number of
0.5 nodes(50)
0.4 100 Sec
0.3 200 Sec
0.2
0.1
0
Normal AŌer Intrusion
Time/Sec

Fig. 4.8 Packet delivery ratio

40 M. S. Vinmathi et al.

100

throughput(Mbps)
80
Number of nodes
60 (50)

40 100 Sec
200 Sec
20

0
Normal AŌer Intrusion
Ɵme(sec)

Fig. 4.9 Throughput comparison

4.5 Conclusions and Future Work

In this paper, we have discussed and implemented bio-inspired search optimization

algorithm called elephant search optimization algorithm for finding the intruders in
cognitive wireless sensor networks. Also we have discussed on the characteristics of
Primary users and secondary users of CRN. The results discussed here shows that the
packet delivery ratio and overall throughput has been slashed to 50% in the network.
Due to this adverse effect the primary users cannot receive the requested resources
and in turn data loss occurs. In future, the simulation study can be increased up to
1000 nodes with live datasets.

References

1. Elangovan, K., Subashini, S.: A survey of security issues in cognitive radio network. ARPN J.
Eng. Appl. Sci. 11(17), 10496–10500 (2016)
2. Wang, X., Ji, Y., Zhou, H., Li, J.: Auction based frameworks for secure communications in
static and dynamic cognitive radio networks. IEEE Trans. Veh. Technol. 66(3), 2658–2673
(2017)
3. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of
IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019)
4. Hemanth Kumar, G., Gireesh, N.: End-to-end communication between IoT devices to maximize
energy efficiency through optimization and localization based on the bio inspired algorithms.
Int. J. Adv. Sci. Technol. 28(16), 1444–1452 (2019)
5. Rajesh, D.: Improved distributed cooperative spectrum sensing (dcss) for cognitive radio adhoc
network. Int. J. MC Square Sci. Res. 7(1), 170–182 (2015)
6. Li, J., Feng, Z., Feng, Z., Zhang, P.: A survey of security issues in cognitive radio networks.
China Commun. 132–150 (2015)
7. Balaji, S., Sasilatha, T.: An efficient routing approach for detection of syn flooding attacks in
wireless sensor networks. EAI Endorsed Trans. Energy Web Inf. Technol. 5(20), 1–6 (2018)
8. Balaji, S., Sasilatha, T.: Detection of denial of service attacks by domination graph application
in wireless sensor networks. Clust. Comput. J. Netw. Soft. Tools Appl. 22(6), 15121–15126
(2019)
4 Bio-Inspired Search Optimization for Intrusion Detection System … 41

9. Darwish, A.: Bio-inspired computing: Algorithms review, deep analysis, and the scope of
applications. Future Comput. Inf. J. 3, 231–246 (2018)
10. Deb, S., Fong, S., Tian, Z., Wong, R.K., Mohammed, S., Fiaidhi, J.: Finding approximate
solutions to NPhard optimization and TSP problems using elephant search algorithm. J.
Supercomput. (Springer, New York, 2016)
11. Meng, X., Liu, Y., Gao, X., Zhang, H.: A new bio-inspired algorithm: chicken swarm
optimization. In: Advances in Swarm Intelligence, Springer, Berlin (pp. 86–94) (2014)
12. Wang, G.G., Deb, S., Coelho, L.D.S.: Elephant herding optimization. In: 3rd International
Symposium on Computational and Business Intelligence, IEEE (2015)
13. Elangovan, K., Subashini,S.: Particle bee optimized convolution neural network for managing
security using cross-layer design in cognitive radio network. J. Amb. Intel. Hum. Comput.
(2018)
14. Elangovan, K., Subashini, S.: Cumulative cooperative spectrum sensing scheme to defend
against selfish users. Indian J. Sci. Technol. 9(16), 1–4 (2016)
Chapter 5
Additive Manufacturing of Thin-Wall
Steel Parts by Gas Metal Arc Welding
Robot: The Surface Roughness,
Microstructures and Mechanical
Properties

Van Thao Le, Dinh Si Mai, Van Chau Tran, and Tat Khoa Doan

Abstract The Additive Manufacturing (AM), which employs the arc to fuse the
metallic wire (WAAM) is attracted much attention for producing metallic parts.
This technique shows a high rate of the material deposition and low production costs
when compared to other metallic AM technologies. In the current work, an industrial
welding robot has been employed for building ER70S6 steel walls according to the
additive manufacturing method. First, two deposition strategies were considered to
build the thin walls layer by layer—i.e., the same deposition direction and alternating
deposition direction strategies. After that, the microstructural and mechanical char-
acteristics of thin walls built with the most suitable strategy were investigated. The
results indicate that the alternating deposition direction strategy allows achieving thin
walls with more regular height. The roughness of the side surface of the thin walls
is about 0.23 mm. The microstructure of ER70S6 thin walls changes from region by
region: the upper region consists of lamellar structures; the middle region features
granular structures; and the lower region shows mixed lamellar and equiaxed struc-
tures. The hardness also varies according to these three regions. The upper region
shows the highest average value of hardness (~191 HV), followed by the lower region
(~178 HV), and the middle region (~163 HV). Finally, the ER70S6 walls built by
WAAM exhibit the anisotropy in terms of tensile strengths in the horizontal and
vertical directions.

Keywords Welding robot · Additive manufacturing · Low-carbon steel ·

Microstructures · Hardness · Tensile strengths

V. T. Le (B) · D. S. Mai · V. C. Tran · T. K. Doan

Le Quy Don Technical University, Hanoi, Vietnam
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 43

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_5
44 V. T. Le et al.

5.1 Introduction

Additive Manufacturing (AM) is considered as a new technique for manufacturing

components with highly complex geometry without using supplementary resources,
for example, cutting tools, fixture systems as in machining [1]. AM technology allows
the manufacture of a physical part from its three-dimensional computer aided design
(3D CAD) model layer by layer, which is opposite to subtractive manufacturing
processes [2]. Thanks to the layer manufacturing principle, AM uses only an amount
of materials required to build designed parts and support structures if necessary.
Therefore, the waste of materials and environmental impacts could be reduced [3, 4].
Based on the energy source employed, metallic AM is categorized into three
groups - i.e., laser-based AM (including powder bed and powder feed systems),
arc-based AM (or WAAM), and electron beam-based AM [5]. Among metallic AM
technologies, WAAM becomes a potential technology for the manufacture of large
components. This technique features a high rate of the material deposition and low
costs of investment [6]. WAAM uses an electrical arc made between the electrode
and the workpiece to melt the metal wire and forms metal parts layer by layer.
In the literature, considerable studies on the microstructures and mechanical char-
acterization of WAAM components from titanium and nicked alloys, as well as
aluminum alloys were carried out [7]. However, limited studies investigated on the
microstructures, hardness and tensile strengths of steel walls produced by a welding
robot according to the WAAM principle. Most of published studies used the stain-
less steel wire (e.g., 304 [8], 304L [9], 308L [10, 11], 316 and 316L [12, 13]) as the
feedstock in WAAM.
In the cases of low-carbon steels, in [14], the authors investigated the effect
of processing conditions on the evolution of microstructures of mild steel walls
produced by a WAAM process, in which a MIG welding torch was controlled by a
3-axis CNC machine. These authors stated that there was any significant difference
in terms of microstructures between samples built with different processing param-
eters. However, in this study, a low level of welding current (50 A) was used to build
the walls, and the voltage was varied in a narrow range from 11.7 V to 13.1 V. Such
a change in terms of processing parameters might be not sufficient enough to cause
a noticeable difference about the microstructure between the printed walls. Haden
et al. [15] also used a WAAM printer constructed at Lehigh University to build thin
walls from SS304 and ER70S mild steel. These authors focused on investigating
the tensile strengths and the wearing property of thin wall materials. In terms of the
microstructure and hardness, the authors only observed in the middle region of the
walls. Nevertheless, the microstructure in other regions, i.e. the bottom and the upper
regions of the wall was not reported. In terms of tensile strengths of mild steel walls,
these authors observed that the tensile strengths (YS and UTS) in the horizontal
direction were lower than that in the building direction. However, other authors [16]
observed an opposite trend, namely the YS and UTS in the horizontal direction were
significantly greater than those in the vertical direction.
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal … 45

In our research, the manufacture of thin-wall mild-steel components by an indus-

trial welding robot is addressed. The effect of two depositing strategies on the geom-
etry of thin walls was first investigated to select the most suitable one. After that,
the surface roughness, the microstructural and tensile strengths of the walls were
investigated to confirm the compatibility of built components with real applications.

5.2 Materials and Experimental Methods

In the experiment, the copper-coated welding wire made of mild steel (ER70S-6)
was used as the feedstock in the welding process. The wire’s diameter is equal to
1.2 (mm). The chemical elements of the wire are 0.04% C, 0.92% Si, 0.45% Mn,
0.015% S, 0.011% P, 0.2% Cu, and Fe (balance). Several plates made of SS400 steel
were also employed as the substrate. The length, width, and thickness of the plates
are 250, 150, and 15 (mm), respectively.
The robotic welding system (Panasonic TA-1400) was used to build the thin
walls on the substrates according to the AM principle (Fig. 5.1). The welding torch
movement is implemented by a 6-axis robot performed. The power source is YD-
350GR3 of Panasonic. The gas of 99.9% CO2 was used for the shielding during the
welding process. The flow rate of the shielding gas was fixed at 16 L/min.

Fig. 5.1 The welding robot TA-1400 of Panasonic

46 V. T. Le et al.

Fig. 5.2 Two deposition strategies considered for building thin-wall components

In the first stage, two strategies - i.e., the same deposition direction (Fig. 5.2a) and
alternating deposition direction (Fig. 5.2b), were used for building thin-wall samples.
In the first strategy, the welding deposits were performed in the same direction,
whereas the second one deposits two adjacent layers in two opposite directions. In
two strategies, after the deposit of a layer, the torch moves to the starting position of
the next deposit. A holding time (t dw ) of sixty seconds was also applied between two
adjacent layers to cool down the part and transfer the cumulated heat to the ambiance
of the experimental room. The current, I = 90 A, the voltage, U = 18 V, and the
welding speed, v = 300 mm/min, were used to build thin walls in both strategies.
In the second stage of this research, a thin wall was built on a substrate by using
the most suitable strategy chosen from the first stage to analyze the surface rough-
ness, microstructures, hardness, and tensile properties. The aforementioned process
parameters were used to build the wall. The height, length, and thickness of the wall
are about 100, 200, and 4.5 (mm), respectively (Fig. 5.3).
To analyze the microstructures of the wall, a specimen MS was cut from the wall in
the middle region by using a wire cutting EDM machine (Fig. 5.3). The cross section
of the wall is ground, polished, and chemically etched. Then, the microstructures
were analyzed by using an optic microscope, AXIO A2M, in different regions of the
cross section. A hardness-testing machine, Future-Tech Vicker FV-310, was used to
measure the hardness.
Three specimens in the vertical direction (vTS-1-2-3) and three specimens in
the horizontal direction (hTS-1-2-3) were extracted from the thin wall (Fig. 5.3).
All tensile strength tests were implemented on a tensile testing machine, INSTRON
3369, at room temperature.
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal … 47

Fig. 5.3 Thin-wall part with the positions of cutting the specimen (MS) for observing microstruc-
tures, the tensile specimens in vertical (vTS-1-2-3) and horizontal (hTS-1-2-3) directions for
observing tensile properties

5.3 Experimental Results and Discussion

5.3.1 Effects of Depositing Strategies on the Shape of Thin

Walls

As revealed in Fig. 5.2d, e, the alternating deposition direction allows achieving the
thin wall with more regular height (Fig. 5.2e) than the first strategy (Fig. 5.2d). This
can be explained based on the original shape of a single welding layer (Fig. 5.2c).
Normally, when the parameters of the welding process were not adjusted, the height
of single welding beads in the starting region is higher than that of the steady region,
whereas the ending region is lower than the steady region [17, 18]. In the alternating
deposition direction strategy, the arc in the current deposit/layer was struck in the
ending region and it was turned off in the starting region of the previous deposit.
Hence, the deviation of the height between the starting and the ending regions of the
previous layer was effectively compensated in the current layer (Fig. 5.2e). On the
other hand, in the same deposition direction strategy, the height of thin walls was
significantly decreased from the starting region to the ending region (Fig. 5.2d).
From this result, the alternating deposition direction strategy was considered as
the most suitable one for building thin wall with more stable shape, and it was
used to build the thin wall to investigate the surface roughness, microstructures, and
mechanical properties of the wall.
48 V. T. Le et al.

5.3.2 Surface Roughness Analysis

The appearance of the side surfaces of the thin wall (Fig. 5.3) was acquired by a
non-contact 3D digitizer (Konica Minolta Range 7). The digitized data (STL file,
Fig. 5.4a) was then transformed into the point cloud {pi }, and a plane (P) that fits
the point cloud {pi } of the side surface was estimated. The equation of the plane (P)
is described by Eq. (5.1):

(P) : α X + βY + γ Z + δ = 0 (5.1)

where α, β, γ , and δ are the coefficients of the plane and α 2 + β 2 + γ 2 = 0. The

surface roughness (SR) of the side surfaces was calculated by Eq. (5.2) [19]:
⎛ ⎞
1 Np
1 ⎝ 
Np
|α X i + βYi + γ Z i + δ|
SR = Di =  ⎠ (5.2)
N p i=1 N p i=1 α2 + β 2 + γ 2

Fig. 5.4 a Appearance of the side surface of the wall (STL file) and b distribution of distance Di
from the point pi to the fitted plane (P)
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal … 49

where N p is the number of the points in the point cloud {pi } and Di is the distance
from the point pi = (X i , Y i , Z i ) to the fitted plane (P). The distribution of the distances
(Di ) in the computation area was shown in Fig. 5.4b. The surface roughness of the
side surfaces of the walls was about 0.23 (mm), which is compared to those reported
in the study of Xiong et al. [19].

5.3.3 Microstructure Analysis

From the images of microstructures, it was firstly observed that the thin walls present
various microstructure types that can be classified in three regions of the walls: the
upper, the middle, and the lower regions. The upper region shows lamellar structures
composed of ferrite with different types: Widmanstätten, allotriomorphic, and acic-
ular morphologies (Fig. 5.5). In fact, the top layer was rapidly cooled by the air in
the experimental room, and this layer was not influenced by the heat of other layers
as in the other regions. In addition, the heat was conducted towards the bottom of
the wall. Thereby, the lamellar structure was generated in this region.
The middle/central region reveals granular structures of ferrite and little portion
of pearlite in the boundaries of grains (Fig. 5.6). It was also found that the center of a
layer, the zone ➀, features finer and denser grains in comparison to the overlapping
zone ➁ (Fig. 5.6). This is because the heat of molten pool that formed the layer (N)

Fig. 5.5 Microstructures in the top region: lamellar structures of ferrite

50 V. T. Le et al.

Fig. 5.6 Typical microstructures in the middle region composed of equiaxed grains of ferrite and
perlite in the boundaries of grains

reheated and partially remelted the previous layer (N − 1), causing coarser grains in
the overlapping zones.
The lower region exhibits a mix of equiaxed and lamellar structures of ferrite
(Fig. 5.7). The lamellar structures also distribute in the equiaxed grains of ferrite.
The lower region contacts the large substrate, whereas the middle region contacts
warmer layers. Thus, the lower region features a higher cooling rate when compared
to the middle region, and the lamellar structures of ferrite were generated in the
bottom region [8].

5.3.4 Mechanical Properties

Microhardness. The hardness of the built thin-wall material was revealed in Fig. 5.8.
The microhardness was measured at five positions in each region, which distribute
on the centerline of the cross section of MS.
It was found that the upper region of the wall is hardest, while the middle region
shows the lowest hardness. The hardness of the upper region, the middle region, and
the lower region is 191 ± 4 HV, 163 ± 4 HV, and 178 ± 6 HV, respectively. This
observation is in line with the microstructure observation. The hardness of the upper
region is greater than the middle and the bottom regions, because the upper region is
dominantly composed of Widmanstätten structures (Fig. 5.5). Due to the presence of
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal … 51

Fig. 5.7 Microstructures in the lower region: mixed lamellar and equiaxed structures

Fig. 5.8 Hardness variation

in three regions of the wall

lamellae structures (Fig. 5.7), the lower region has the hardness value that is higher
than that in the middle region characterized by granular structures (Fig. 5.6).
Tensile strengths. The ultimate tensile strength (UTS) and yield strength (YS) of the
wall were presented in Fig. 5.9. The YS and UTS in the vertical direction equal to
362 ± 8 MPa and 479 ± 7 MPa, respectively, are higher than those in the horizontal
direction (YS = 320 ± 6 MPa and UTS = 429 ± 8 MPa). The anisotropy in terms of
the microstructure might cause this difference of the tensile strengths between two
directions.
52 V. T. Le et al.

Fig. 5.9 Comparison of the

tensile strengths in the
horizontal direction (hTS)
and the vertical (vTS)
direction of the wall

5.4 Conclusions

In this paper, an industrial welding robot was employed for producing thin wall
components according to the AM method. The effect of the deposition strategy on the
shape was firstly investigated. Thereafter, the most suitable deposition strategy was
used to build a thin wall for investigating the surface roughness, the microstructures,
and the mechanical characteristics of built materials. The main outcomes of this
research can be summarized as follows:
• The alternating deposition direction strategy allows achieving thin-wall compo-
nents with more shape stability and regular height.
• The surface roughness of the thin wall is about 0.23 mm.
• The wall features the microstructure that varies from region by region. The upper
region is characterized by lamellar structures. The middle region is composed of
granular structures of ferrite. The pearlite appears in the boundaries of grains. The
lower region reveals a mix of lamellar and equiaxed structures.
• The variation of microstructures in different regions leads to the variation of
hardness from 163 to 191 HV.
• The mechanical properties of thin-wall low-carbon components built by the
welding robot (YS: 320–362 MPa and UTS: 429–479 MPa) are also comparable to
those of wrought A36 low-carbon steel, which has similar chemical compositions
in compared to ER70S-6.

Acknowledgements This research is funded by Vietnam National Foundation for Science and
Technology Development (NAFOSTED) under grant number 107.99-2019.18.
5 Additive Manufacturing of Thin-Wall Steel Parts by Gas Metal … 53

References

1. Chen, L., He, Y., Yang, Y., Niu, S., Ren, H.: The research status and development trend of
additive manufacturing technology. Int. J. Adv. Manuf. Technol. 89, 3651–3660 (2017). https://
doi.org/10.1007/s00170-016-9335-4
2. ASTM: F2792 - 10e1 Standard Terminology for Additive Manufacturing Technologies (2010).
https://doi.org/10.1520/F2792-10E01
3. Le, V.T., Paris, H., Mandil, G.: Environmental impact assessment of an innovative strategy
based on an additive and subtractive manufacturing combination. J. Clean. Prod. 164, 508–523
(2017). https://doi.org/10.1016/j.jclepro.2017.06.204
4. Le, V.T., Paris, H.: A life cycle assessment-based approach for evaluating the influence of total
build height and batch size on the environmental performance of electron beam melting. Int.
J. Adv. Manuf. Technol. 98, 275–288 (2018). https://doi.org/10.1007/s00170-018-2264-7
5. Frazier, W.E.: Metal additive manufacturing: a review. J. Mater. Eng. Perform. 23, 1917–1928
(2014). https://doi.org/10.1007/s11665-014-0958-z
6. Williams, S.W., Martina, F., Addison, A.C., Ding, J., Pardal, G., Colegrove, P.: Wire + arc
additive manufacturing. Mater. Sci. Technol. 32, 641–647 (2016). https://doi.org/10.1179/174
3284715Y.0000000073
7. Wu, B., Pan, Z., Ding, D., Cuiuri, D., Li, H., Xu, J., et al.: A review of the wire arc additive
manufacturing of metals: properties, defects and quality improvement. J. Manuf. Process. 35,
127–139 (2018). https://doi.org/10.1016/j.jmapro.2018.08.001
8. Bai, Y., Gao, Q., Chen, X., Yin, H., Fang, L., Zhao, J.: Research on microstructure and properties
of 304 stainless steel made by MIG filler additive manufacturing. IOP Conf. Ser. Earth Environ.
Sci. 237 (2019). https://doi.org/10.1088/1755-1315/237/3/032096
9. Laghi, V., Palermo, M., Tonelli, L., Gasparini, G., Ceschini, L., Trombetti, T.: Tensile properties
and microstructural features of 304L austenitic stainless steel produced by wire-and-arc additive
manufacturing. Int. J. Adv. Manuf. Technol. 106, 3693–3705 (2020). https://doi.org/10.1007/
s00170-019-04868-8
10. Le, V.T., Mai, D.S.: Microstructural and mechanical characteristics of 308L stainless steel
manufactured by gas metal arc welding-based additive manufacturing. Mater. Lett. 271, 127791
(2020). https://doi.org/10.1016/j.matlet.2020.127791
11. Foley, R.P.: Microstructural analysis of additively manufactured 308L stainless steel produced
by plasma arc welding. Master thesis, Montana Tech (2019)
12. Chakkravarthy, V., Jerome, S.: Printability of multiwalled SS 316L by wire arc additive manu-
facturing route with tunable texture. Mater. Lett. 260, 126981 (2020). https://doi.org/10.1016/
j.matlet.2019.126981
13. Wang, L., Xue, J., Wang, Q.: Correlation between arc mode, microstructure, and mechanical
properties during wire arc additive manufacturing of 316L stainless steel. Mater. Sci. Eng. A
751, 183–190 (2019). https://doi.org/10.1016/j.msea.2019.02.078
14. Liberini, M., Astarita, A., Campatelli, G., Scippa, A., Montevecchi, F., Venturini, G., et al.:
Selection of optimal process parameters for wire arc additive manufacturing. Procedia CIRP
62, 470–474 (2017). https://doi.org/10.1016/j.procir.2016.06.124
15. Haden, C.V., Zeng, G., Carter, F.M., Ruhl, C., Krick, B.A., Harlow, D.G.: Wire and arc additive
manufactured steel: tensile and wear properties. Addit. Manuf. 16, 115–123 (2017). https://
doi.org/10.1016/j.addma.2017.05.010
16. Lu, X., Zhou, Y.F., Xing, X.L., Shao, L.Y., Yang, Q.X., Gao, S.Y.: Open-source wire and arc
additive manufacturing system: formability, microstructures, and mechanical properties. Int. J.
Adv. Manuf. Technol. 93, 2145–2154 (2017). https://doi.org/10.1007/s00170-017-0636-z
17. Waqas, A., Qin, X., Xiong, J., Wang, H., Zheng, C.: Optimization of process parameters
to improve the effective area of deposition in GMAW-based additive manufacturing and its
mechanical and microstructural analysis. Metals (Basel) 9, 775 (2019). https://doi.org/10.3390/
met9070775
54 V. T. Le et al.

18. Xiong, J., Yin, Z., Zhang, W.: Forming appearance control of arc striking and extinguishing area
in multi-layer single-pass GMAW-based additive manufacturing. Int. J. Adv. Manuf. Technol.
87, 579–586 (2016). https://doi.org/10.1007/s00170-016-8543-2
19. Xiong, J., Li, Y., Li, R., Yin, Z.: Influences of process parameters on surface roughness of
multi-layer single-pass thin-walled parts in GMAW-based additive manufacturing. J. Mater.
Process. Technol. 252, 128–136 (2018). https://doi.org/10.1016/j.jmatprotec.2017.09.020
Chapter 6
An Application of Observer
Reconstruction to Estimate Actuator
Fault for DC Motor Nonlinear System
Under Effects of the Temperature
and Disturbance

Tan Van Nguyen and Xuan Vinh Ha

Abstract In this paper, we analysis the factors that impact the precisional control
process of the DC motor such as disturbance, the temperature effect on coil resistance
and the temperature effects on magnetic fluxes. From that, we suggest a method
to estimate the actuator fault which will be performed in the future to apply for
the process of eliminating fault. First, a nonlinear mathematical model of the DC
motor under the action of temperature is constructed to control the system. Second,
building the inequalities based on the reconstruction of unknown input observer
(UIO) with considering disturbance is constructed to estimate the actuator faults
using Lyapunov’s stability condition and a linear matrix inequality (LMI) optimiza-
tion algorithm to obtain the control signal error asymptotically stable. Finally, the
numerical simulation process is done to monitor the obtained result of the proposed
method.

Keywords DC motor nonlinear system · Unknown input observer · Actuator fault

estimation

6.1 Introduction

The speed simulation result of the DC motor nonlinear system is shown as Fig. 6.4.
The result shows the response speed and its estimation signal approaching to the
reference signal at the times without touch of actuator fault and opposite the response
signal do not approach to the reference signal at the time where have actuator faults.
We can see errors in the signal under the impact of the disturbance d.

T. Van Nguyen · X. V. Ha (B)

Engineering—Technique Faculty, Thu Dau Mot University, Binh Duong, Vietnam
e-mail: [email protected]
T. Van Nguyen
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 55

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_6
56 T. Van Nguyen and X. V. Ha

In the past decades, the types of DC motor have been applied in the role such
as actuators because they have the potential choices for modern industries ranging
from manipulators to accurate machine tools because of their vantages such as
compact structure, controllability, high precision accuracy, reliability, low costs, etc.
The temperature effect on coil resistance and the temperature effects on magnetic
fluxes are shown in [1–7]. Here, [6, 7] showed the effect of the motor temperature
to brushless direct current motor performance on motor and electric vehicle. The
temperature effect lead to plant parameter variation on the feedback control loop
also showed in [2]. In addition, the motor may arise faults such as Shaft: Imbalance,
misalignment, and wear or Rotor: Imbalance, rotor bar faults, loose rotor, and eccen-
tricity or Motor bearing housing: Mechanical looseness, misalignment. However,
there are many faults or failures arise from around environment, as well as from
system nonlinearities with large uncertainties become critical challenges in utilizing
motors to obtain high precision position and accurate speed control. To solve this
issue, the fault-tolerant control (FTC) technique for DC motors is proposed for actu-
ator faults. The compensation process is performed once the actuator fault estimation
obtained. There have been many researchers to study estimation of the actuator and
sensor fault using various algorithms such as unknown input observer (UIO) as
in [8–10], or using the augmented system per-formed by [11], using sliding mode
observer as shown in [12–15], etc. Especially, Actuator and sensor fault estimation
was effectively implemented using variable reconstruction shown in [16, 17].
In this paper, UIO is constructed based on the Lyapunov’s stability condi-
tion and LMI optimization algorithm to reconstruct the actuator faults that the
control error dynamic is asymptotically stable. The decoupled process of actuator
faults is performed via UIO reconstruction. Numerical simulation results show the
effectiveness of the proposed method.
The important contributions of paper can be shortened as follows:
• A UIO reconstruction is constructed to estimate the actuator fault based on set up
of control error to obtain asymptotically stable of the system.
• Inequalities to apply LMI optimization algorithm is approved by Lyapunov
stability.
• The numerical simulation results of the actuator fault compensation process show
the success of the motor’s proposed method.

6.2 DC Motor Model Formulation

The control scheme of the DC motor is implemented by the applied voltage V a

controls the angular velocity ω of the shaft. Where a simplified model of the DC
motor is shown (Fig. 6.1) with the parameters shown in Table 6.1.
The dynamics equation of the DC motor can be rewritten as [1, 17]:
6 An Application of Observer Reconstruction to Estimate Actuator … 57

Fig. 6.1 Scheme of DC

motor and control loop
design

Table 6.1 Variables and

Variables Parameters
parameters of the DC motor
Electrical i(t): Current (A) R: Resistance ()
dynamics v(t): Voltage (V) L: Inductance (H)
e(t): Back EMF (V) : Motor torque
constant (N m/A)
Mechanical T m (t): Motor torque f : Friction torque
dynamics (N m) (N m s rad−1 )
ω(t): Angle velocity J: Moment of inertia
(rad s−1 ) (kg m2 )

⎧ (t) f
⎪
⎨ ω̇(t) = i(t) − ω(t)
J J (6.1)
⎩ di = v(t) − (t) ω(t) − R(t) i(t)
⎪
dt L L L
Resistance and K(t) parameter under the effects of temperature can be written as
[1]
  
R(t) = R0 (t) + R0 (t)α t f − t0
  (6.2)
(t) = 0 (t) + 0 (t)β t f − t0

where R0 (t) and 0 (t) are resistance and torque constant of DC motor at the
temperature 20 °C.
α, β are the parameters shown in Tables 6.2 and 6.3 [7]. t0 and t f are the temperature
at 20 °C and at any time t respectively.
From (6.1) and (6.2), the DC motor dynamics equation can be represented as

Table 6.2 Temperature

Conductor material α (°C)
coefficient of metals [7]
Gold 0.0037
Silver 0.0038
Copper 0.0040
Aluminum 0.0043
58 T. Van Nguyen and X. V. Ha

Table 6.3 Permanent magnet

Magnet material β (°C) Maximum temperature
material temperature
(°C)
coefficients [7]
Cobalt nickel aluminum −0.0002 540
Samarium cobalt −0.0004 300
Neodium iron boron −0.0012 150
Ceramics −0.0020 300

⎧
⎪ 0 (t) 0 (t)   f
⎪
⎪ ω̇(t) = i(t) + β t f − t0 i(t) − ω(t)
⎪
⎪ J J J
⎪
⎨ di v(t) 0 (t) 0 (t)   0 (t)i(t)
= − ω(t) − β t f − t0 ω(t) − (6.3)
⎪
⎪ dt L L L   L
⎪
⎪
⎪
⎪ 0 (t)α t f − t0
⎩ − i(t)
L
The DC motor dynamic system (6.3) may be described in state space as

ẋ(t) = x(t) + ϒu(t) + ζ (x, t) (6.4)

where
x1 = i(t); ẋ1 = x2 = di
dt
; x3 = ω(t); x4 = ω̇(t) and
⎧
T
⎪
⎪ x = x1T x3T
⎪
⎪  
⎪
⎪ R0 (t) 0 (t)  
⎪
⎨ x2 = − L − L x1
0 (t)
x4 − Jf x3 ;
⎪
⎪ J
   
⎪
⎪ R0 (t)α (t f −t0 )  (t)
 
⎪ 1
+ −0 (t)  L  − L β t f − t0
⎪
0 x1
⎪
⎩ + L u(t)
J
β t f − t 0 0 x 3 0
 R0 (t) 0 (t)  1
− L − L
= 0 (t) ; ϒ= L ;
− Jf 0

J
   
R0 (t)α (t f −t0 ) 0 (t)
−
ζ (x, t) = 0 (t)  L  − L
β t f − t 0 x(t); x(t) = 1
x
J
β t f − t0 0 x 3

6.3 UIO Design for Nonlinear System

Considering a state-space equation of the nonlinear system subject to an unknown

input with assuming that disturbance and actuator fault as the following form:
6 An Application of Observer Reconstruction to Estimate Actuator … 59

ẋ(t) = x(t) + ϒu(t) + ζ (x, t) + Fa f a (t) + Dd(t)
(6.5)
y(t) = x(t)

where x(t) ∈ R n is the state vector, y(t) ∈ R p is the outputs vector and d(t) ∈ R rd is
the unknown input or disturbance vector, and f a (t) ∈ R f is the actuator fault. Here
, ϒ, , and D are known constant matrices with suitable dimensions. ζ (x, t) is a
nonlinear function vector ∀x(t), x(t) ∈ R n , u(t) ∈ R m .
The equation of (6.5) can be rewritten in the following form:

E ẋ(t) = x(t) + ϒu(t) + ζ (x, t) + F a f a (t) + Dd(t)
(6.6)
y(t) = x(t)

where
       
 Fa I 0 D ζ (x, t)
= ; E= n ; D= ; ζ (x, t) =
0 −I f 0 0f 0 0
     

0 x(t) ϒ
= 0 ; Fa = ; x(t) = ∈ Rn ; ϒ =
If f a (t) 0

with n = n + p, and f a (t) is the actuator fault.

By using [18], a Lipschitz constraint of the nonlinear function vector ζ (x, t) can
be expressed as:
 

ζ (x, t) ≤ χ x(t) − x̂(t) (6.7)

where
  

ζ (x, t) = ζ (x, t) − ζ x̂, t 

Based on [16], the UIO paradigm can be designed in the influences of unknown
inputs in the system (6.6) as:
⎧  

⎪
⎪ ż(t) = Mx̂(t) + Mϒu(t) + Mζ x̂, t + L y(t) − ŷ(t)
⎨
x̂(t) = z(t) + H y(t) (6.8)
⎪
⎪
⎩
ŷ(t) = x̂(t)

where x̂(t) ∈ R n , and ŷ(t) ∈ R p are state vector estimation of x(t), and measurement
output estimation vector, respectively. z(t) ∈ R n is the state vector of the observer.
60 T. Van Nguyen and X. V. Ha

M ∈ R n×n , H ∈ R n× p , L ∈ R n× p are the observer matrices and these matrices

should be designed according to the state estimation error vector.
An estimation control error can be defined as:

e(t) = x(t) − x̂(t) (6.9)

and
˙
ė(t) = ẋ(t) − x̂(t)
 
= In − H ẋ(t) − ż(t)
= M ẋ(t) − ż(t) (6.10)

where

M = In − H

Besides

e y (t) = e(t) (6.11)

Equation (6.4) can be rewritten as:

M E ẋ(t) = Mx(t) + Mϒu(t) + Mζ (x, t) + M F a f a (t) + M Dd(t) (6.12)

and from (6.6), we obtain

˙  
x̂(t) = Mx̂(t) + Mϒu(t) + L y(t) − L ŷ(t) + Mζ x̂, t + H ẋ(t)

 
= M − L x̂(t) + L y(t) + H ẋ(t) + Mϒu(t) + Mζ x̂, t (6.13)

Using (6.7)–(6.11), we have:

˙

M E + H ẋ(t) − x̂(t) = M − L x(t) − M − L x̂(t)

+ M
ζ (x, t) + M F a f a (t) + M Dd(t) (6.14)

 
where
ζ (x, t) = ζ (x, t) − ζ x̂, t .
From (6.14), the estimation error can be represented as:
 
ė(t) = M − L e(t) + M
ζ (x, t) + M F a f a (t) + M Dd(t) (6.15)

If the satisfy the following conditions

6 An Application of Observer Reconstruction to Estimate Actuator … 61

M E + H = In (6.16)

The matrix M can be chosen:

 
In 0
M= . (6.17)
− 0 f

From that, we can be calculated H as

 
0
H= (6.18)
If

Lemma [8, 16] Consider the state space equation in the following form:

ζ̇ (t) = τ (t) + λ(t) (6.19)

where  ∈ R n×n is the eigenvalues of a given matrix belong to the circular region
D(ϕ, μ)with the center ϕ + j0 and the radius μif and only if there exists a symmetric
positive definite matrix N ∈ R n×n such that the following condition holds:
 
−N N( − ϕ In )
<0 (6.20)
∗ −μ2 N

In addition, from (6.5), we have:

 

ζ (x,u)  ≤ χ e(t) (6.21)

where
 
χ In 0
χ= ∈ Rn ;
0 0f

e(t) = x(t) − x̂(t)

The matrix  can be inferred from (6.22) as:

⎡ ⎤
−χ T χ 0 0
 = ξ T ⎣ 0 In 0 ⎦ξ ≤ 0 (6.22)
0 0 0

where
 T
T
ξ = e T (t)
ζ (x, t) f aT (t) d T (t) .
62 T. Van Nguyen and X. V. Ha

Theorem [8] By applying the character ofthe control error condition, and require-
ment of the output estimation error satisfy e y (t) ≤ ρd(t), as well as there exist
symmetric matrices N ∈ R n×n , Q ∈ R n× p , and positive scalars ρand εthat the
system (6.4)exists a robust UIO model in the form of (6.6)that satisfies a prescribed
circular area D(ϕ, r )that insured the following inequalities hold:
⎡ T   T     T
⎤
M N + N M − Q T − Q + εχ T χ NM N M F a N M D
⎢ ⎥
⎢ ∗ −ε In 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −δ I f 0 0 ⎥<0
⎢ ⎥
⎣ ∗ ∗ ∗ −ρ Id 0 ⎦
∗ ∗ ∗ ∗ −ρ I p
(6.23)

and
 
−N NM − Q − ϕN
<0 (6.24)
∗ −r 2 N

where Q = NL;
Then Eq. (6.8)is asymptotically stable.

Proof of (6.23) Consider a Lyapunov function of the control error signal is:

V (t) = e T (t)Ne(t) (6.25)

Derivative Eq. (6.25), we have:

˙
V̇ (t) = ē˙ T (t)Nē(t) + ē T (t)Nē(t)
 

= M ¯ − L ¯ ē(t) + M
ζ̄ (x, t) + M F̄a f a (t) + M D̄d(t) T Nē(t)
 

+ ē T (t)N M  ¯ − L ¯ ē(t) + M
ζ̄ (x, t) + M F̄a f a (t) + M D̄d(t)
⎡ ⎤T ⎡        ⎤
ē(t) M ¯ − L ¯ T N + N M ¯ − L ¯ NM N M F̄a N M D̄
⎢
ζ̄ (x, t) ⎥ ⎢ ⎥
=⎢ ⎥ ⎢ ∗ 0 0 0 ⎥
⎣ f a (t) ⎦ ⎢⎣ ∗ ∗ 0 0
⎥
⎦
d(t) ∗ ∗ ∗ 0
⎡ ⎤
ē(t)
⎢
ζ̄ (x, t) ⎥
⎢ ⎥ (6.26)
⎣ f a (t) ⎦
d(t)

Based on (6.22) and (6.26), we obtain:

V1 = V̇ (t) − ε ≤ 0 (6.27)
6 An Application of Observer Reconstruction to Estimate Actuator … 63

Equation (6.27) can be rewritten as:

⎡    ⎤
T N + N NM N M F a N M D
⎢ ∗ 0 0 0 ⎥
V1 = ξ T ⎢
⎣
⎥ξ − ε − δ f T (t) f a (t)
⎦ (6.28)
∗ ∗ 0 0 a

∗ ∗ ∗ 0

where  T
T
ξ = e T (t)
ζ (x, t) f aT (t) d T (t) and  = M − L

(6.28) can be represented as

⎡    ⎤
T N + N − εχ T χ NM N M F a N M D
⎢ ∗ −ε In 0 0 ⎥
V1 = ξ T ⎢
⎣
⎥ξ < 0
⎦ (6.29)
∗ ∗ −δ I f 0
∗ ∗ ∗ 0
 
In addition, with limitation of the output measurement error condition e y (t) ≤
ρd(t), then the matrix V 2 is computed as:
⎡ ⎤
1 T
ρ
0 0 0
⎢ ⎥
⎢ ∗ 0 0 0 ⎥ξ ≤ 0
V2 = ξ T ⎢ ⎥ (6.30)
⎣ ∗ ∗ 0 0 ⎦
∗ ∗ ∗ −ρ Id

From (6.29) and (6.30), the matrix V 12 can be presented as:

V12 = V1 + V2 (6.31)

Equation (6.31) is represented as:

⎡    ⎤
T N + N − εχ T χ NM N M F a N M D
⎢ ∗ −ε In 0 0 ⎥
V12 = ξT⎢
⎣
⎥ξ
⎦
∗ ∗ −δ I f 0
∗ ∗ ∗ 0
⎡ T
⎤
1
00 0
⎢ρ ⎥
T⎢ ⎥
+ ξ ⎢ ∗ 0 0 0 ⎥ξ = ξ T ξ (6.32)
⎣ ∗ ∗0 0 ⎦
∗ ∗ ∗ −ρ Id

where
64 T. Van Nguyen and X. V. Ha

⎡    ⎤ ⎡ 1 T ⎤
T N + N − εχ T χ NM N M F a N M D 0 0 0
⎢ ⎥ ⎢ρ ⎥
∗ −ε In 0 0 ⎥+⎢ ∗ 0⎥
=⎢
⎣ ⎦ ⎢
0 0
⎥
∗ ∗ −δ I f 0 ⎣ ∗ ∗ 0 0⎦
∗ ∗ ∗ −ρ Id ∗ ∗ ∗ 0
(6.33)

Applying the Schur lemma [19] of Eq. (6.33) with condition   < 0, then the
inequality (6.23) is satisfied if V12 < 0 satisfies the condition e y (t) ≤ ρd(t).

Proof of (6.24) Applying (6.15) to Lemma, then (6.24) is satisfied.



Conclusion, the process of the actuator fault estimator is implemented in the

following steps:
• Step 1: Solve the matrix inequalities (6.23) and (6.24) to find out the matrices Q,
N, and L = N−1 Q.
• Step 2: Calculate actuator
fault
estimation via state vector f a (t) = a x̂(t),
respectively, with a = or ×n Ir .

6.4 Simulation Results

Results are obtained from the simulation process of the DC motor nonlinear system
with parameters shown in Table 6.4. Here, the DC motor nonlinear system in state
space is described as

ẋ(t) = x(t) + ϒu(t) + ζ (x, t) (6.34)

From the parameters of the DC motor nonlinear system in [2], item  in (6.34)
can be calculated as:
⎡ ⎤ ⎡ ⎤
−200 −40 200 200

 = ⎣ 200 −0.1 0 ⎦; ϒ = ⎣ 0 ⎦; = 0 1 1 .
0 0 −1 0

Table 6.4 The value of parameters of DC motor

Parameters Symbol Units Value
Resistance R  1
Inductance L H 5 × 10–3
Motor torque constant (N.m/Amp)  0.2
Friction torque F N m s rad−1 2.5 × 10–2
Moment of inertia J kg m2 0.001
6 An Application of Observer Reconstruction to Estimate Actuator … 65

We can choose the positive scalars χ , ρ, α, ε, and μ as χ = 0.1; ρ = 0.1, α = 0.2,

ε = 0.05, and μ = 0.1 using the design experience. By solving for N, Q, and L
using the optimal algorithm LMI, if the solution is feasible, then the results can be
obtained as: ⎡ ⎤
⎡ ⎤ ⎡ ⎤ −3.7339 × 10−10
1 0 0 0
⎢ ⎥
M = ⎣ 0 1 0 ⎦; H = ⎣ 0 ⎦; Q = ⎣ 1.7764 × 10−09 ⎦;
0 −1 0 1 1.2226 × 10−09
⎡ −12
⎤
−5.8680 × 10 −1.7313 × 10−11 −1.5908 × 10−11
N = ⎣ −1.7313 × 10−11 1.5150 1.5150 ⎦; L =
−11
−1.5908 × 10 1.5150 1.5150
⎡ ⎤ ⎡ ⎤
46.624 100
⎣ 71.055 ⎦; E = ⎣ 0 1 0 ⎦
−71.055 001
The input signal is expressed as

0 if t ≤ 2
yr (t) = (6.35)
10 i f t > 2

The actuator fault is given by equation as following


0 if t ≤ 1
f a (t) = (6.36)
50t − 50 if 1 < t ≤ 1.4

and after time t = 5 s, the temperature of DC motor is 150 °C.

6.4.1 Simulation Without Disturbance d(t) = 0

The simulation result of the speed of the DC motor nonlinear system is shown in
Fig. 6.2. The result shows the response speed and its estimation signal following
the reference signal at the times without impact of actuator fault. However, it was
affected by the impact of the actuator fault at locations 1–2 s as shown in Fig. 6.2.
The actuator fault and its fault estimation are demonstrated in Fig. 6.3.
The simulation result in Fig. 6.3 indicators the actuator fault estimation similar
to the given actuator fault signal. This proves that the actuator estimator works well.
66 T. Van Nguyen and X. V. Ha

Fig. 6.2 Simulation signals in the case without disturbance

Fig. 6.3 Actuator fault and its estimations

6.4.2 Simulation with Disturbance

d(t) = 0.025 r andom(2, t)

The simulation result of the speed of the DC motor nonlinear system is shown in
Fig. 6.4. The result shows the feedback speed and its estimation signal reach the
reference signal at the times where there is no impact of actuator fault, and opposite
the responsible signal does not follow to the reference signal at the time that has the
actuator fault. We possibly see errors in the feedback signal at the location has the
impact of the disturbance d.
The simulation result of fault estimation is shown in Fig. 6.5 for disturbance case.
6 An Application of Observer Reconstruction to Estimate Actuator … 67

Fig. 6.4 Reference and output with the actuator fault and the disturbance d(t)

Fig. 6.5 Actuator fault and its fault estimations with disturbance d(t)

6.5 Conclusions

A UIO model was reconstructed to perform the actuator fault estimation. The control
error result of the DC motor nonlinear system reaches the asymptotic stability state
by using the UIO conversion structure of the algorithm to optimize the LMI and
Lyapunov stabilization conditions, even though the system is affected by noise and
other force components impact to the actuator. Also, the simulation results of the
estimated actuator fault in Figs. 6.3 and 6.5 showed the achieved signal as desired.
The result displayed influences of control error under the impacts of the actuator
fault and in the case without and with disturbance were also compared to show their
effects in the system. The actuator fault estimation approach successfully applies
68 T. Van Nguyen and X. V. Ha

to the DC motor nonlinear system where there is a considerable influence of the

disturbance of the environment.

References

1. Richard, H.W., George, W.Y.: How temperature affects a servomotor’s electrical and mechanical
time constants (2002)
2. Musaab, A.M.S., Muawia, M.A.M.: Effect of plant parameter variation on feedback control
loop (DC motor temperature effect model). Int. J. Adv. Res. Comput. Sci. Manag. Stud. 2(9),
384–388 (2014)
3. Andresen, M., Liserre, M.: Impact of active thermal management on power electronics design.
Microelectron. Reliab. 54, 1935–1939 (2014)
4. Bum-Su, J., Joon, S.P., et al.: Temperature estimation of stator winding in permanent magnet
synchronous motors using d-axis current injection. Energies 11, 2033 (2018). https://doi.org/
10.3390/en11082033
5. Abdullah, N.F., Abbas, A.W.: Study of temperature rise of small brushed DC motor under
different load. MATEC Web Conf. 225, 03010 (2018). https://doi.org/10.1051/matecconf/201
822503010
6. https://www.motioncontroltips.com/understanding-dc-motor-curves-temperature/
7. Iwa, G., Nadya, N., Muhammad, V.N.: Analysis of the effect of the motor temperature to brush-
less direct current motor performance on KARLING electric vehicle. In: 2018 International
Conference on Engineering, Technologies, and Applied Sciences (ICETsAS 2018), Lampung,
Indonesia
8. Tan, V.N., Cheolkeun, H.: Sensor fault-tolerant control design for mini motion package electro-
hydraulic actuator. MDPI Process. 7, 89 (2019)
9. Xiaoxu, L., Zhiwei, G.: Unknown input observers for fault diagnosis in Lipschitz nonlinear
systems. In: Proceedings of 2015 IEEE International Conference on Mechatronics and
Automation, Beijing, China (2015)
10. Gao, Z., Liu, X., Chen, Q.Z.M.: Unknown input observer-based robust fault estimation for
systems corrupted by partially decoupled disturbances. IEEE Trans. Ind. Electron. 63(4), 2537–
2547 (2016)
11. Noura, H., Theilliol, D., Ponsart, J.C., Chamseddine, A.: Fault-Tolerant Control Systems
Design and Practical Applications. In: Michael, J.G., Michael, A.J. Springer, Dordrecht, Heidel-
berg, London, New York. ISBN 978-1-84882-652-6. https://doi.org/10.1007/978-1-84882-
653-3
12. Jian, Z., Akshya, K.S., Sing, K.N.: Reconstruction of actuator fault for a class of nonlinear
systems using sliding mode observer. In: 2011 American Control Conference, San Francisco,
CA, USA (2011)
13. Lu, Q., Bin, J., Pu, Y.: Actuator fault diagnosis for flight control system based on sliding mode
observer. In: 2018 Chinese Control and Decision Conference (CCDC). https://doi.org/10.1109/
CCDC.2018.8407683
14. Ali, B.B., Slim, D., Fayçal, B.H., Anis, S.: Sliding mode observers based-simultaneous actuator
and sensor faults estimation for linear parameter varying systems. Math. Probl. Eng. 2017, 13
pages. Article ID 1747095 (2017). https://doi.org/10.1155/2017/1747095
15. Liu, M., Zhang, L., Shi, P., Zhao, Y.: Fault estimation sliding-mode observer with digital
communication constraints. IEEE Trans. Autom. Control 63(10), 3434–3441 (2018). https://
doi.org/10.1109/TAC.2018.2794826
16. Jia, Q., Li, H., Zhang, Y., Chen, X.: Robust observer-based sensor fault reconstruction for
discrete-time systems via a descriptor system approach. Int. J. Control Autom. Syst. 13, 274
(2015)
6 An Application of Observer Reconstruction to Estimate Actuator … 69

17. Ayodeji, O., Oluwadamilola, O., et al.: Modeling and simulation of DC motor using
Simelectronics and Simulink. Gazi Mühendislik Bilimleri Dergisi 5(1), 91–100 (2019)
18. Liu, X., Gao, Z., Zhang, A.: Robust fault-tolerant control for discrete-time dynamic systems
with applications to aero engineering systems. IEEE Access 6, 18832–18847 (2018)
19. Boyd, S., Ghaoui, L.E., Feron, E., Balakrishnan, V.: Linear Matrix Inequalities in Systems and
Control Theory. SIAM, Philadelphia, PA, USA (1994). ISBN 0-89871-334-X
Chapter 7
Enhanced Hadoop Distribution File
System for Providing Solution to Big
Data Challenges

A. Essakimuthu, R. Karthik Ganesh, R. Santhana Krishnan,

and Y. Harold Robinson

Abstract Big Data is utilized for large amount of data in the format of information
to the media. The information technology based organizations are developed Big
Data for huge volume of data processing like Google, Oracle, Yahoo and IBM. Big
data is also used to gather the piece of information without losing any data in the
organization. The dynamic and responsive business will utilize the content of the
large amount of data. This paper analyzes the problems occurred for maintaining big
data processing and maintaining it. Hadoop Distribution File System is developed
to produce the solutions for the big data challenges with the concept of Acquire,
organize, analyze and decide using Analytic application.

Keywords Clouds · Big data · Business intelligence · Data warehouse · Data over
the internet

7.1 Introduction

Data processing for diverse types of data on big cluster delivers more proficient
performance as associated with various database management systems [1]. The enor-
mous amount of data comes from different sources in various kinds of patterns such

A. Essakimuthu · R. Karthik Ganesh

Department of Computer Science and Engineering, SCAD College of Engineering and
Technology, Cheranmahadevi, India
e-mail: [email protected]
R. Karthik Ganesh
e-mail: [email protected]
R. Santhana Krishnan
Department of Electronics and Communication Engineering, SCAD College of Engineering and
Technology, Cheranmahadevi, India
e-mail: [email protected]
Y. Harold Robinson (B)
School of Information Technology and Engineering, Vellore Institute of Technology, Vellore, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 71

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_7
72 A. Essakimuthu et al.

as videos, pictures, social media sites and sensors, RFID, purchased and transactions,
cellular phone and satellites signals [2]. If business needs to know what’s happening
in their business, they need complete and thorough level of information [3]. The
source of information coming from sensors, devices, web applications and social
media which is never been exploit for Business Intelligence (BI) purposes before
[4]. The development of information is coming from internet, sensors and images
[5]. The information which is thousands times higher data rates those IT systems
managing currently which is causing the business losing valuable information [6].
The companies are demanding for more improved and detailed value of information
[7]. Information is the only assets that companies earn and run their business on, if
that information is out dated or lost it will bring the whole organisation [8]. The idea
of big data has been implemented as huge dataset which comes from conventional
and non-conventional sources [9]. The Financial industry is delicate to data concerns
because it always needs reliable and precise information about financial data and
customer information [10]. As discussed that big data is becoming the need of every
industry in the world, new technologies are retrieving and collecting more data than
ever before [11]. Each industry wants to utilize every piece if it’s valuable informa-
tion to strive and excel with its competitor better than others [12]. The big data is
more than about volumes which involves variety and velocity of organisation’s data
[13]. Oracle and IBM also share the same concepts about three but oracle goes one
point ahead with four with the addition of value in its v definitions [14]. When it
comes to data, every organisation is different from others and they have different
requirements to store and present data to business [15].
• Volume: This sort of non-traditional data is generated by machine in enormous
quantities.
• Velocity: This sort of non-traditional data is generated by social media.
• Variety: This type of data is traditional data and modified gradually.
• Value: There is always valuable information is concealed among huge form of
non-traditional data; the task is to find that data for analysis.
Figure 7.1 demonstrates that the volume, velocity and variety in details. Volume
will contain physical existence of the records in the form of tables, records and
transactions. Velocity will involve, real time and streams sort of data. Variety will
have all sorts of data including structured and unstructured.
The Global institute described that data volume is increasing 40% per year and it
will rise forty four times [16]. This will not only the matter of managing volumes of
data but the value of the data will also matters [17]. The data is the value and main
assets of the organisations [18]. Big Data will make easy access to the applicable
data for different stake holders in well-timed which can produce incredible value for
the organisation [19]. By making data availability at time and easy accessibility will
drop search and processing time which will create value and customer satisfaction
for the organisations [20]. The big data impact on the health care system where the
false claim can be easily identified by individuals [21]. Figure 7.2 demonstrates the
health care system that the air flow, skin conductance, pulse meter, temperature and
position. It is connected with the time analysis based database connectivity through
7 Enhanced Hadoop Distribution File System for Providing Solution … 73

Fig. 7.1 Three parameters of big data

Fig. 7.2 Health care system

network and it is connected with the network router with cloud applications. The
user can view the health care information system and it is communicated with the
doctors for online transmission [22].
Big data will make organisation’s transactional data available in a digital form
which will allow them to collect precise and detailed information of data on real
time [23]. The financial intuition grieves from inefficiencies around data; it will save
time and human efforts of manual processing financial applications [24]. It will also
provide the whole picture of customer relationship with other institutions to regulators
74 A. Essakimuthu et al.

authorities to analyse risk associated with it [25]. The big data will be delivered the
great openings to the financial institutions to improve the customer satisfaction with
the support to establish the vigorous customer communication management program
[26]. Big data will permits organisations to generate segmentation of their data which
will permit them to shape their products and services more accurately. Big data
practices will allow real time micro-segmentation of customers to aim advertising
and promotions [27].

7.2 Solution for Big Data Challenges

Big data allows organisations to improve its decision making and display data that is
hidden by computerized algorithms. It will help tax collecting departments and other
financial institutions to banner candidates for further detailed analysis. With the help
of big data organisations can analyse the entire data sets which will provide them
better decision making against the real time data. The big data will provide insurers
organisation to analyse whole data sets to enhance business process around customers
and products to improve customer service. Through big data companies will be able
to produce different products and services, improve the existing one as well and
develop innovative business models. It will allow the manufacturers companies to
use the actual data to develop next generation’s products and to produce after sales
inventive contributions. Hamm described that companies wants to understand what
sort of products customer really want and at the end of the day what they will buy
all linked to bid data. The old-style methods to EIA are obstructed by big data that
EIA move its emphasis from optimization and standardization to different practices
like coordination and externalization in enterprise life cycle. Big data shift the focus
from data warehousing to further towards data pooling.
Oracle discussed about to bring big data into enterprise architecture, so the
problem is find the relevant information and use it in real time. Organisations are
looking for to make most of big data and use it within the existing infrastructure.
Oracle come up with the big data model where it can deployed efficiently in organi-
zation’s existing infrastructure. The big data states to data streams of higher velocity
and higher variety, this information should be capture with no latency with simple
quires to manage huge volumes of data. NoSQL databases are used quite often to
obtain and to accumulate big data. The data kept in NoSQL database is normally of
a great diversity because the systems are planned to purely capture all data without
classifying and analysing the data. The method for utilizing NoSQL data storage
engines to exploit the storing data the same way it is stored. Some big companies
still preferred to use the relational database and to measure data for all sort of kinds.
Google considered Big Table and Amazon come with Dynamo to utilize all data
form huge relational database.
The data is stored on its original location, so it will save both time and money
by not transferring huge volumes of data. This phase is more like integration of data
in data warehousing terms. Apache Hadoop is the modem technology which will
7 Enhanced Hadoop Distribution File System for Providing Solution … 75

permits large data volume to be organised and managed while keeping the data into
its original data storage cluster. All analysis taken place on the data that the analysis of
big data is essential to support crucial detailed statistical analysis and data mining on
massive and different varieties of data types. It will provide organisation to not only
analyse new data but also keep the old analysis to deliver an innovative viewpoint
on old problems and it is demonstrated in Fig. 7.3.
The customer data is the precious assets for the organisations which required
vigilant governance and protection. Managing big data is not easy and it will require
lot of precaution and architectural framework to make sure the quality and security
of the underlying data. The data governance issues related to more towards security.
Big data provide the detailed level of analysis and sensitive information to trade
secret, financial archives and knowledgeable belongings. Make information available
centrally will provide informal and treasured target for attackers or hackers which
can harm the status of the organisations. The newly framed big data issues like data
governance problem in public and private organisations. To address these sorts of
issues organisations needs to re-design its data policy which supports and simplify
data inventions. As explained by the experts that data governance is great worry
for records which is coming from multiple sources. Data gathered from different
geographical locations where user needs to share the locality with the service provider
which will cause the privacy issue. There is still great concern in terms of big data,
how to restrict access to information and what types of data which will end up with
a lesser amount of detailed information. In practical world data is not still and it is
always changes, its volume, velocity and variety, so fundamental procedures will not
gone work in this scenario.

7.3 Results and Discussion

The big data technologies are still developing which will lead the companies to
revisit, its privacy and security policies. Organisations are struggling to combine
data in new ways which is not only causing the great security concerns but also how
the data should be processed and where to keep that data? If companies will decide to
use third parties tools to integrated the data whether it will be feasible and safe to do
that or not? Data governance is still the main problem, who owns the data and who
will held accountable if information is breached or how to track where information
is stolen and by whom? The massive obstacle in the implementation of the bid data
is accountability of the data. Organisations should make resolutions about who will
utilize big data and data sets. Organisations need successful governance strategy to
implement data governance; otherwise it will not be durable to link unrelated data
or information back to the client, customers and production tables to make right
decisions at right time. The enormous challenge face big data in term of governance
is the lack of budget, resources and awareness from the top management. Companies
need proper data governance policy to make successful execution of big data, lack
of procedures and direction principal around data causes governance issues for big
76 A. Essakimuthu et al.

Fig. 7.3 Steps for providing solution

7 Enhanced Hadoop Distribution File System for Providing Solution … 77

data. In short data governance issues revolve around, data privacy, data ownership,
governance policies, budget and interest from the executive level. Data governance
is not a project or program; it is an ongoing process which cannot be stop at any
time, even after the successful implementation of big data.
Big data is becoming the hottest topic and flash discussion point in any format
of information media today. Most of the largest organisations linked to information
technology investing in and advertising about it, like Oracle, IBM, Google, Amazon,
Microsoft, Yahoo and SAS. Big data is about data which can be in whichever format
and size, utilize by the organisations without losing a single piece of information. Big
Data is not simply about the size but also about the content and data which will make
organisations business more responsive and dynamic. Every day data is growing
and expanding in different formats, to accomplish and proficiently process rising
thousands of petabytes data each day, current technologies and method cannot fulfil
the necessities and needs new trends of technologies like ‘big data’. The data volumes
are increasing daily which only involves transactions data and other conventional data
types. The next generation operation technology is part of that data which involves
audio, videos, streaming data and social network. Storing massive amount of data is
the issue but doing the analysis on it also complex.
According to Fig. 7.4 show different dimensions of big data, in quantification it
will look for volume of data, variety of data formats and velocity of data. In access

Fig. 7.4 Dimensions of big data

78 A. Essakimuthu et al.

enablement and control, it will look for classifications, contracts and technology
enablement. In qualification and assurance, it will look for fidelity, linked data and
validation of the data.
The described about the survey which held under TDWI shows different organi-
sation’s commitment and prospective growth. According to Fig. 7.5, different level
of commitment and potential growth shown by different participants, the way data
is handled and process.
Figure 7.6 shows the big data spectrum with NoSQL and SQL perspective, NoSQL
is required on larger data sets to acquire all of the data in timely manner on the other
hand SQL is used under more secure and administered environment against relational
database systems.

Fig. 7.5 Potential growth versus commitment

7 Enhanced Hadoop Distribution File System for Providing Solution … 79

Fig. 7.6 Big data solution for spectrum

Figure 7.7 which shows the complete oracle big data solutions, where Hadoop
Distribution File System (HDFS) is used to collect and organise data and theses
accumulated outcomes later loaded into relational DBMS system.
The solution is provided to minimize the lack of technical people staff with
profound knowledge of analytical skills. There are some risks attached to big data
such as regulatory and legal challenges which needs proper education and training
to the organisations staff. The data quality and governance plans need to be revisited
and modified when implementing big data in organisations. Traditional data quality
will not work in context of big data; organisations will be seeing the whole data sets
for wholeness, correctness and reliability of the data. With big data sets, data profiling
will be time consuming and inefficient, so different practices which involves limit
the scope of the data to accomplish data quality processes.

Some of the data flow processes happens outside the organization’s control
which needs to be handle through by creating quality controls procedures otherwise
80 A. Essakimuthu et al.

Fig. 7.7 Hadoop

Distribution File System
7 Enhanced Hadoop Distribution File System for Providing Solution … 81

there will be harmful for internal processes. In relation to big data implementation
Company A will need extra hardware and applications given below:
• Big Data Appliances
• Licensing for Oracle Data Integrator (ODI), Hadoop, MapReduce and NoSQL
Applications
• Transferring all the code from O WB to ODI because OWB will no longer
supported and obsolete in next two years
• ODI has adaptor for big data connectivity and support heterogeneous sources
• Needs Big Data Scientists
• Staff training
• Master Data Management for Big Data
• Data Steward
• Redesign the Governance Policy Framework.

7.4 Conclusion

Big data plays an important role in modem organizations in today’s world. Organi-
sations are managing huge volume of data, at great velocities, from multiple sources
which includes structured, unstructured, conventional, non-conventional, inside and
third party data and information because of big data. Complied and gathered detailed
level information about customer information which was not possible before. It
provides organisation ability to classify and handle all transactions, their hierarchies
and association through variety of views. Big data is leading edge for the business
growth and cost reduction for the organisations and delivers new openings for the
CIO to become the strategic partner with different business divisions. Company
A needs big data technology for its strategic objectives and to gain advantage of
its competitors which will give Company A more standardise approach for business
operation good data governance, detailed data analysis and information for improved
decision making, fraud detection in finance and accounts subject areas, mobile assets
management, fixed plant equipment, statistical analysis for correlation, regression
and business trends, superior visualisation of the data and towards enhanced single
approach for global KPI’s and business solutions. EMC explained that big data will
bring new functioning competences, fresh profits streams, modest benefits and new
business model and processes. Now it is the time for the business frontrunners to
reflect tactically to prepare their organisations for the big data because technology
wise it is mature now and lot of well renowned business names already deployed it
successfully.
82 A. Essakimuthu et al.

References

1. Shinde, G., Deshmukh, S.N.: Sentiment TFIDF feature selection approach. Int. J. Comput.
Commun. Eng. (2016)
2. Pandey, A.C., Rajpoot, D.S., Saraswat, M.: Twitter sentiment analysis using hybrid cuckoo
search method. Inf. Process. Manag. 53(4), 764–779 (2017)
3. Harold Robinson, Y., Santhana Krishnan, R., Golden Julie, E., Kumar, R., Son, L.H., Thong,
P.H.: Neighbor knowledge-based rebroadcast algorithm for minimizing the routing overhead
in mobile ad-hoc networks. Ad Hoc Netw 93:1–13 (2019)
4. Huang, C.-M., Shao, C.-H., Xu, S.-Z., Zhou, H.: The social Internet of Thing (S-IOT)-based
mobile group handoff architecture and schemes for proximity service. IEEE Trans. Emerg.
Top. Comput. 5(3), 425–437 (2017)
5. Zhou, Z., Jin, X.L., Vogel, D.R., Fang, Y., Chen, X.: Individual motivations and demographic
differences in social virtual world uses: an exploratory investigation in second life. Int. J. Inf.
Manag. 31(3), 261–271 (2011)
6. Harold Robinson, Y., Golden Julie, E.: MTPKM: multipart trust based public key management
technique to reduce security vulnerability in mobile ad-hoc networks. Wireless Pers. Commun.
109, 739–760 (2019)
7. Wang, G., Gunasekaran, A., Ngai, E.W., Papadopoulos, T.: Big data analytics in logistics and
supply chain management: certain investigations for research and applications. Int. J. Prod.
Econ. 176, 98–110 (2016)
8. Liang, P.-W., Dai, B.-R.: Opinion mining on social media data. In: 2013 IEEE 14th International
Conference on Mobile Data Management (MDM), vol. 2. IEEE (2013)
9. Balaji, S., Robinson, Y.H., Julie, E.G.: GBMS: a new centralized graph based mirror system
approach to prevent evaders for data handling with arithmetic coding in wireless sensor
networks. Ingénierie Syst. d’Inform. 24(5), 481–490 (2019)
10. Shvachko, K., Kuang, H., Radia, S., Chansler, R.: The hadoop distributed file system. In: 2010
IEEE 26th Symposium on Mass Storage Systems and Technologies, MSST, pp. 1–10. IEEE
(2010)
11. Riquelme, F., González-Cantergiani, P.: Measuring user influence on Twitter: a survey. Inf.
Process. Manage. 52(5), 949–975 (2016)
12. Harold Robinson, Y., Rajaram, M.: Energy-aware multipath routing scheme based on particle
swarm optimization in mobile ad hoc networks. Sci. World J. 1–9 (2015)
13. Chard, K., Caton, S., Rana, O., Bubendorfer, K.: Social cloud: cloud computing in social
networks. In: The IEEE 3rd International Conference on Cloud Computing, Miami, USA,
5–10 July 2010
14. Iqbal, R., Grzywaczewski, A., Halloran, J., Doctor, F., Iqbal, K.: Design implications for task-
specific search utilities for retrieval and reengineering of code. Int. J. Enterp. Inf. Syst. 11(5),
738–757 (2017) (Taylor and Francis, pp. 1751–7575)
15. Santhana Krishnan, R., Golden Julie, E., Harold Robinson, Y., Raja, S., Kumar, R., Thong,
P.H., Son, L.H.: Fuzzy logic based smart irrigation system using internet of things. J. Clean.
Prod. 252, 119902 (2020)
16. Han, G., Zhou, L., Wang, H., Zhang, W., Chan, S.: A source location protection protocol based
on dynamic routing in WSNs for the social internet of things. Future Generat. Comput. Syst.
82, 689–697 (2018)
17. Cheng, X., Meng, B., Chen, Y., Zhao, P., Li, H., Wang, T., Yang, D.: Dynamic table: a layered and
configurable storage structure in the cloud. In: Bao, Z., et al. (eds.) WAIM 2012. LNCS, vol.
7419, pp. 204–215. Springer, Heidelberg (2012). https://doi.org/10.1007/978-3-642-33050-
6_21
18. Lakshminarayanan, K., Santhana Krishnan, R., Golden Julie, E., Harold Robinson, Y., Kumar,
R., Son, L.H., Hung, T.X., Samui, P., Ngo, P.T.T., Tien Bui, D.: A new integrated approach based
on the iterative super-resolution algorithm and expectation maximization for face hallucination.
Appl. Sci. 10, 718 (2020). https://doi.org/10.3390/app10020718
7 Enhanced Hadoop Distribution File System for Providing Solution … 83

19. Zhang, X., Ding, S., Xue, Y.: An improved multiple birth support vector machine for pattern
classification. NeuroComputing 225, 119128 (2017)
20. Salloum, S.A., Al-Emran, M., Monem, A.A., Shaalan, K.: A survey of text mining in social
media: Facebook and twitter perspectives. Adv. Sci. Technol. Eng. Syst. J. 2(1), 127–133 (2017)
21. Balaji, S., Golden Julie, E., Harold Robinson, Y., Kumar, R., Thong, P.H., Son, L.H.: Design
of a security-aware routing scheme in mobile ad-hoc network using repeated game model.
Comput. Stand. Inter. 66 (2019)
22. Hogenboom, A., Heerschop, B., Frasincar, F., Kaymak, U., de Jong, F.: Multi-lingual support
for lexicon-based sentiment analysis guided by semantics. Decis. Support Syst. 62, 43–53
(2014)
23. Rasooli, A., Down, D.G.: A hybrid scheduling approach for scalable heterogeneous hadoop
systems. In: Proceedings of SC Companion High Performance Computing, Networking,
Storage Analysis (SCC), Nov 2012, pp. 1284–1291
24. Santhana Krishnan, R., Golden Julie, E., Harold Robinson, Y., Kumar, R., Son, L.H., Tuan,
T.A., Long, H.V.: Modified zone based intrusion detection system for security enhancement in
mobile ad-hoc networks. Wireless Netw. 1–15 (2019)
25. Ahmad, A., et al.: Toward modeling and optimization of features selection in big data based
social internet of things. Future Generat. Comput. Syst. 82, 715726 (2017)
26. Samariya, D., Matariya, A., Raval, D., Babu, L.D., Raj, E.D., Vekariya, B.: A hybrid approach
for big data analysis of cricket fan sentiments in twitter. Paper presented at the Proceedings of
International Conference on ICT for Sustainable Development (2016)
27. Harold Robinson, Y., Jeena Jacob, I., Golden Julie, E., Ebby Darney, P.: Hadoop MapRe-
duce and dynamic intelligent splitter for efficient and speed transmission of cloud-based
video transforming. In: IEEE 3rd International Conference on Computing Methodologies and
Communication (ICCMC), pp. 400–404. IEEE (2019)
 Part II
Intelligent System Design and Applications
Chapter 8
Electronic Public Distribution System
in Electronic Government
Israa M. Hayder, Dalshad J. Hussein, Hussain A. Younis,
and Hameed Abdul-Kareem Younis

Abstract This paper which entitles “Electronic Open Dissemination Framework

in Electronic Government” may be a web based on online application expecting
for citizen. The most destinations of this application are to create it intuitively and
makes facilities to the client for tackling every day issues. Citizens can utilize this
framework to rummage around for data and send essential documents within the
electronic government for the Service of Commerce in client side for understanding
their issues. Moreover, it is utilized to total their data within the database framework
by utilizing full title and proportion card number of family to enter the framework.
Citizen can take after this case and can discover the result for this ask from the
system. In Citizen to Government (C2G), client can send their ask to Electronic Open
Dissemination Framework directly and E-center of the framework can take activity
for these cases by staff part who contains a extraordinary client title and watchword,
staff can do overhaul and increment within the database records such as individual,
and operator. Electronic Open Dissemination Framework can take activity for these
cases by admin who incorporates a uncommon admin title and password, admin can
overhaul and embed and erase within the database records. This framework has two
clients (citizen) portion framework and admin (government) portion framework. The
framework comprises of Microsoft Office Get to, Tomcats server, JSP and HTML as
Web browser as the front-end client.

Keywords Electronic government · Citizen government · Ministry of Commerce

I. M. Hayder (B)
Department of Computer Systems Techniques, Technical Institute/Qurna, Southern Technical
University, Basrah, Iraq
e-mail: [email protected]
D. J. Hussein
Slemani Polytechnic University Computer Science Institute, Slemani, Iraq
H. A. Younis
College of Education for Women, University of Basrah, Basrah, Iraq
H. A.-K. Younis
Department of Computer Science, College of Computer Science and I.T, University of Basrah,
Basrah, Iraq

© Springer Nature Switzerland AG 2021 87

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_8
88 I. M. Hayder et al.

8.1 Introduction

Information and Communication Technologies (ICT)s comprising telecommunica-

tion, computer, optical fibers, communication satellites and the internet are to infor-
mation societies, those technologies of information, generation of knowledge and its
use to improve the efficiency and economy for all human in any societies.
In the past few decades, government has come to organize and regulate not only the
traditional activities of defense law and order, and justice but also social welfare and
human resource development governments are the single largest takers of people’s
earning and the providers of variety of services to citizen, including dispensing
information and welfare.
It is therefore natural ICTs are increasingly used to improve government services.
This activity is called e-government just the commerce, education and advocacy
have come to be called e-commerce-education, and e-advocacy.
In every country, the first steps for e-government were taken by establishing
the national informatics’ Center (NIC); it has done a remarkable job of connecting
thousands of government offices which are the repositories of a wealth of information
required for planning and execution of numerous development schemes and also for
governance.
NIC be the largest repository of comprehensive information about our country,
its people, its wealth, its projects and trade and commerce and many other areas.
It has been assigned the task of computerizing work in different department of
the union as well as state government.
In the process, NIC developed varied and commendable competence for comput-
erizing and networking government offices.

8.2 E-Government

E-governance is a form of e-business in governance comprising of tactics and systems

worried in deliverance of digital offerings to the public citizens. In military applica-
tion [1], Toward the social inclusive digital society [2]. It also entails participating
with commercial enterprise partners of government via conducting electronic trans-
actions with them. Beside it entails enabling most people to engage with the govern-
ments, thru electronic approach for purchasing the favored offerings. In different
word, E-governance method software of digital method inside the interaction among.
1. Government (G) and residents (C), each method (i.e. G2C and C2G).
2. Authorities or corporations both methods.
3. Internal authorities’ operation.
As the winds of E-governance and e-government beneath extensively thru public
organizations across the phrase increasingly more authorities in exclusive countries
varied nature as follows:
8 Electronic Public Distribution System in Electronic Government 89

Fig. 8.1 A broad schematic system for E-government model

• G-a-G (govt. to govt. inside and over the govt.)

• G-a-C (services by way of the govt. to Citizen)
• C-a-G (interaction of Citizen with the govt.)
• G-a-B (offerings of the govt. business)
• B-a-G (Govt interplay with commercial enterprise) (Fig. 8.1).

8.3 Literature Review

We as folk, impost digress city limit, are insightful of the astuteness of ICT (Infor-
mation and Message Technology) in our resolutely dynamism No it is abrogation of
splendid immigrant shoal, buy, quilt, suavity, iatrical, allowing, certitude, adventure
distribution, flawless estate deals, purchase of sharers, marriages or entertainment,
explanation of IT has become a necessity in the private sector. Inflation use of ATM’s
for abandonment of prime underpinning be cited as example [3].
90 I. M. Hayder et al.

Directing impacts of administration on data foundation and e-government

improvement. Appealing extraordinary the assets complementarily constrain of
the resource-based exhortation of a changeless, this examination looks at. Cate-
gorize obligation of administration dimensions namely, acknowledgment and obli-
gation, factional grapple, specialist combat, administrative haughtiness, adminis-
tration of embody, As well as control of corruption—on the occurrence between
a nation’s notion camp and its e-government assistance. Based on straight get-
at-able documented truths outlander 178 nations, our comes about fit back for
the hypothesized demonstrate. Uncommonly, sick civic attach, government deed,
and head up of conduct oneself quieted the occurrence of tip modest all
over it loan in a unadulterated conveyance, voice and responsibility and
control of debasement directed the relationship contrarily. Face, the relation-
ship between data foundation and E-Government move ahead was not unex-
pected on administrative quality. Our percipience abdicate to the supernatural apply
oneself to on E-Government prepayment by highlighting the paratactic commit-
ment of coordinating and give recommendations for application in overseeing e-
government improvement by appropriate directorate, subsequently leveraging
the total of data framework on e-government advancement [4].

8.4 Materials and Methods

In this study a new E-government is presented which is based on the Java, Html and
JSP as interface, MS access has been used as a backend of the database.
Java server page is the web Application tool of java to design the website is called
JSP (java Server page) [1]. The website is the collection of web pages in the form
of static and dynamic webpage, the static page is the collection of items like text,
graphics, animation, audio, video etc. while the dynamic page is a type of data entry
form as UI form (user interface), this has been made by the various UI controls like
text, radio, checkbox, combo, text area, password, button etc. [5, 6].

8.4.1 Application Servers (Tomcat Apache)

Apache Tomcat is a easy outset fall on plate and servlet maintain again excellent
through “the Apache Software Foundation (ASF)”. Tomcat paraphernalia the Java
Servlet and the Java Server Pages (JSP) specs outlander Diviner Question, and affords
a internet server surroundings for Java Apache Tomcat is code to [7].
Microsoft place of business get entry to be a Microsoft database management
laptop that blends Microsoft’s relational Jet Database Engine with a graphical one
consumer interface and software-development equipment. MS get admission to shops
statistics in its private plan primarily based at the get admission to Database Engine.
It might also moreover import or hyperlink at as soon as to statistics saved in different
8 Electronic Public Distribution System in Electronic Government 91

programs and databases. Software builders and files architects can use Microsoft get
admission to enlarge software program and can use it to construct software program
software programs. Like different workplace programs [8].
access is supported through visible simple for applications, an object-orientated
programming language that can reference a diffusion of items along with DAO (statis-
tics access objects).The original idea of get right of entry to was for quit users so
that it will “get right of entry to” information from any source. Different capabilities
consist of the import and export of records to some codecs which includes Microsoft,
square Server, Oracle, and ODBC [9].

8.4.2 Proposed E-Government Schemes

See Fig. 8.2.

Fig. 8.2 E-R diagram E-government schemes

92 I. M. Hayder et al.

8.5 Result and Discussion

Work system in Ministry of Trade in Iraq generally consists of two categories,

depending on electronic system and manual system. This system operating is not
working on net services such as e-mail but it can be applied by a user who is oper-
ating it. The user duty is entering all data. Another aspect is that each province
consists of different areas and each area has its own computer system.

8.5.1 E-Governments in the Middle East

Recently, in Middle East countries and specially in Iraq, electronic government

appeared but this system is not being used as its real function. It appears like a
propaganda for announcement and news. For instance, when you look at the site
ministry of trade it is shown the activities of this ministry and the amounts of money
for buying food and so on. Therefore, the citizen doesn’t make any relation with the
site.

8.5.2 Futures This Project

In this project, throughout our site we tried to make a real connection between the
citizen and the government. For example, the citizen can enter the site and interact
with it by sending email and asking for any request which related to him. There is
an icon which is written as “result” by clicking it he\she can follow the next steps
and result.
The government part of this application consists of admin and staff who have
special password and username. Both can arrange all the duties of the system.

8.5.3 Running the Project

While testing this project the first step that should be taken is running the TOMCAT
program. Next step we call the electrical site on the address bar of the internet browser,
by writing https://localhost:8080/trade/main.jsp.

8.5.4 Main Page

We can find the main page of our project as below.

8 Electronic Public Distribution System in Electronic Government 93

Fig. 8.3 First page that describe the system

Fig. 8.4 User page interface

And we can enter to the system as User, staff or admin with special password for
each of these three (Fig. 8.3).

8.5.5 User Page

User can enter to the system as User with special user name and password which
include of full user name and family number (Fig. 8.4).

8.5.6 Main Site for Use

The user must enter the user name and password for opening the site and for making
any request such as searching, updating, requesting for inserting new registration
(Fig. 8.5).
94 I. M. Hayder et al.

Fig. 8.5 Main page of health care system

Fig. 8.6 Staff page for updating information

8.5.7 Staff Page

The staff must take permission from admin to enter the system by staff user name
and password for making update such as insert new information and updating of
information and it cannot make any deletion (Fig. 8.6).
By giving special Personnel with usernames and passwords can open staff page,
can make updating only inserting and correction for data.

8.5.8 Admin Main Page

By giving special admin name and password, admin can open admin page and can
make all updating like inserting, correction, and deletion for data (Fig. 8.7).
The main function of this site is searching for data which depend on the data in the
database’s named persons in the ACCESS program. The HTML database together
8 Electronic Public Distribution System in Electronic Government 95

Fig. 8.7 Main page of admin for updating his information

Table 8.1 Structure of

Field name Data type Field size
family table
RC_NO Text 3
FAM_NO Text 6
FIRST_NAME Text 10
SECOND_NAMAE Text 10
LAST_NAME Text 10
PER_NO Number Byte
INF_NO Number Byte
FOOD_AGENT Text 4
FLOWER_AGENT Text 4
ID_OF_HEAD Text 4

with the Access database (PERSON) is connected via ODBS and JDBS to the site
for accessing the database.
These three file are 1-Agent’s1 file, 2-Family’s2 file and 3-Per’s3 file will be use
as the main database. Family is the first file, which contain names of the head of the
family with their size and their numbers. Also contains the numbers of flour agents
and numbers of the food agent and sector number RC_NO (Table 8.1).
Second file is Person which contains (personal information) such as names of the
head of the families and names of all members of the family with birth date for all
members of the family and ID card number and record number and page number and
mother name of each member and sector number RC_NO (Table 8.2).
Third file is Agent which contains information about food agent and flour agent
like food agent names and flour agent names and address for food and flour agent,
also contain total number of family and total number of persons which receive food
and flour on monthly bases from these agents (Table 8.3).
When we running the project in client we get the first page as below (Fig. 8.8).
From this page we can search by ID card number, ration card number and ask
for some services for example including new born, transfer from agent to agent,
reprinting ration cards, new marriage cases, also we can ask for some information

1 https://localhost:8080/trade/index.html.
2 https://localhost:8080/trade/index.html.
3 https://localhost:8080/trade/index.html.
96 I. M. Hayder et al.

Table 8.2 Structure of

Field name Data type Field size
person table
RC_NO Text 3
FAM_NO Text 6
PFIRST_NAME Text 10
PSECOND_NAMAE Text 10
PLAST_NAME Text 10
BIRTH Text 4
REL Text 5
ID_NO Text 6
PAGE Text 3
REC Text 5
MOTHER Text 10

Table 8.3 Structure of agent

Field name Data type Field size
table
AGENT_NO Text 4
AGENT_NAME Text 30
AGENT_ADDRESS Text 50
AGENT_TYPE Text 6
CITY Text 10
IDICATED POSISTION Text 50

Fig. 8.8 Shows startup web page for user

8 Electronic Public Distribution System in Electronic Government 97

Table 8.4 Response the

Field name Data type Field size
result
IDP Text 6
F_FIRST Text 10
F_FATHER Text 10
F_GRAND Text 10
ACTION Text 10
TYPE OF PROBLEM Text 30

about food agent and flour agent, also we can search for the status of cases by clicking
result button, also we can find some information from clicking home and contact us
bottoms and we can take some help by help bottom. And there is also a file that will
be created in the data center dedicated for the paperwork progression and to know if
the application was successful or not this fie name is (result). The citizen can follow
up or any inquiry for their transaction in the E–G center they must contact E–G
database via searching in the result file to know the status for their cases (Table 8.4).
As far as the families with newly born children concerned, they must scan their
ration card and another document such as ID card and birth certification and up load
them to the center via the E–G site and follow its progression. On completion of
their application they will have the right of receiving share of milk and soap and
detergents.
In regard to newly families (marriage case) they will have to up load their original
families ration cards their id cards and their marriage certification so as their names
can be removed from their original families, and the new family will be created, so
they can receive their share of food and flour and oil (which is 600 letters per year)
as separate family.
We can use this website to solve the problems that is existing in the system of food
distribution and flour distribution, and because of an increase in birth and death rates
and the dynamic change in the Databank information such as address of families,
leading to change numbers of food and flour agent, also the presence of thousands
of new families every month Because of marriages is made the citizens to visit
government establishments within the cities, districts, and rural areas continuously.
How we can ask for including new infant to E–G database by using this site: First
step: ID card and birth certification must scan. Second step: After first step must call
the site by https://localhost:8080/trade/index.html.
Third step: Must press new born bottom and follow the instruction.

8.5.9 Searching Operation

How we can search in the E–G governorates by Ration card number: First step: must
call the site by https://localhost:8080/trade/index.html.
98 I. M. Hayder et al.

Fig. 8.9 Search information windows

Fig. 8.10 Displaying information after search operation

Second step: Must press family number bottom and follow the instruction as
showed below (Fig. 8.9).
Third step: if the family number is available in the database we can see the family
information as below (Fig. 8.10).

8.6 Conclusions

Nowadays, developing of electronic health care system very required from varies
government and non-government organization. The best suite system should have the
ability to better interaction between 3 parties (patient, physician, and administrator of
the system). The main useful point behind using health care systems is to keep track
on each and every patient case in which it will help to provide better perception for to
predict and expect the relation among patience that belong to same committee. In our
proposed system we provide a model for designing e-health system that able to find
the correlation among health cases by analysis the food and the patience behavior
8 Electronic Public Distribution System in Electronic Government 99

which could produce better comprehensive study to correlate between the food and
patience and also to manage the public health issues by government.

References

1. Sharma, N., Gupta, S.: An investigation of IT-intervention adoption in public distribution system:
a stakeholder and agency theory perspective. 35(2), 203–219 (2019). https://doi.org/10.1177/
0266666917736714.
2. Tikhomirova, O.: E-governance and social inclusion of entrepreneurship and businesses: toward
the social inclusive digital society. Int. J. E-Entrepreneur. Innovation (IJEEI) 10(2) (2020).
https://doi.org/10.4018/IJEEI.2020070101
3. Akemi, T.C., Omar, A.: A cross-country comparative analysis of e-government service delivery
among Arab countries. Inf. Technol. Deve. Spec. Issue: e-Govern. Initiat. Dev. World: Challenges
Opportunities 15(3), 151–170 (2015)
4. Satish K.: Moderating effects of governance on information infrastructure and e-government
Development, 2- Journal of the American Society for Information Science and Technology. The
complete reference JSP Phil Hanna (2012)
5. Graphics J.N.: Learning Web Design A Beginner’s Guide to (x)HTML, Style Sheets, and Web
(2007)
6. Hussain, A.Y., Ahmed, S.A., Israa, M.H.: Supposed techniques to overcome institute corruption
that result from bad use of database management system 7(4), 4980–4983
7. Lemuria, C.: Electronic government, or e-government, increases the convenience and accessi-
bility of government services and information to citizens. Inf. Syst. J. 15(1), 5–25 (2005)
8. Claudio C., Navarra, D.D.: Good governance, development theory, and aid policy: Risks and
challenges of e-government in Jordan, Article first published online: 5 APR 2005. Inf. Technol.
Dev. 11(2), 141–159 (2005). https://doi.org/10.1002/itdj.20008
9. Caldow J.: The Quest for electronic government: a defining vision, company report. IBM
Corporation (1997).
Chapter 9
Combining Syntax Features and Word
Embeddings in Bidirectional LSTM
for Vietnamese Named Entity
Recognition

Bui Thanh Hung

Abstract Named Entity Recognition (NER) is the task to classify words into prede-
fined target entity classes. This task is now considered a core of many natural language
processing tasks such as document classification, machine translation, search engine,
information extraction,… Distributed representation of words (word embeddings) is
an important technique in current natural language processing. Word embeddings is a
good feature but if we combine both word embeddings and syntax features can bring
better effect. This paper proposes a new approach by combining automatic syntactic
features with pre-trained word embedding in Bidirectional Long Short Term Memory
(BiLSTM) deep learning method for Vietnamese Named Entity Recognition. The
proposed system has achieved good results for the Vietnamese NER problem on the
VLSP 2016 dataset.

Keywords Named entity recognition · Deep learning · Bi-LSTM · Word

embeddings · Syntax features

9.1 Introduction

In natural language processing, Named Entity Recognition (NER) is an importance

task. The main role of this task is to identify the noun phrases in the text and clas-
sify them into predefined groups such as name of organization, location, person,
time and monetary value. Towards those goals, NER can provide useful informa-
tion for other natural language processing applications such as search engines, text
summaries, machine translation, information extraction… [1, 2]. Identification of
proper names has been studied and applied in many languages such as English,
Japanese, Chinese, etc. Recently, a systematic comparison contest of NER systems

B. T. Hung (B)
Engineering—Technology School, Data Analytics & Artificial Intelligence Laboratory, Thu Dau
Mot University, 6 Tran Van On Street, Phu Hoa District, Thu Dau Mot City, Binh Duong
Province, Vietnam
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 101

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_9
102 B. T. Hung

for Vietnamese language has been conducted and launched by the Vietnamese Speech
and Language Processing (VLSP) community [3].
In this research, we propose to use the deep learning approach for Vietnamese
Named Entity Recognition by combining automatic syntax features with pre-trained
word embeddings as the input of deep learning model - Bidirectional Long Short
Term Memory (Bi-LSTM).
This paper consists of 5 sections, Sect. 9.1 introduces about Named Entity Recog-
nition problem. The related works are reviewed in Sect. 9.2. Section 9.3 describes
the proposed model. Our experiments are discussed in Sects. 9.4 and 9.5 concludes
about our research.

9.2 Related Work

In generally, we could divide the existing machine learning methods in NER into two
groups. The methodology in the first group applied manual extraction techniques
and combined with a sequential labeling algorithm such as Conditional Random
Field (CRF), Hidden Markov Model (HMM), Support Vector Machine (SVM), or
Maximum Entropy Markov Model (MEMM) [1]. The second group is related to the
deep learning approach. Several neural network architectures have been proposed
to solve the problems of string labeling. The outstanding advantage of deep neural
network architectures is the end-to-end learning capability—which is the ability to
learn the rules of string labeling from a pre-labeled dataset without the need for any
human intervention [4–7]. Typically, deep neural network architectures require quite
large datasets to achieve high accuracy. For languages with few labeled dataset such as
Vietnamese, the most effective solution is to use a well-trained word embedding from
a large unlabeled dataset. In the research [8], the author used a combination of Bi-
LSTM and CRF to identify Vietnamese names with data taken from 720,000 articles
of the online newspaper. This system has a high rate of 74% on test dataset. However,
the author only created the features that the word vectors are separated from the
word embedding of GloVe or FastText [9, 10]. In this paper, we will present the most
appropriate neural network architecture for the NER problem. This architecture is
based on the combination of a Bi-LSTM model, word embeddings and automatically
generated syntactic features: Chunk, POS and Regex.

9.3 Proposed Method

Figure 9.1 shows our proposed model which is described more detail in this section.
9 Combining Syntax Features and Word Embeddings … 103

Vietnamese NER

Output
Vietnamese Documents

Word, Chunk, POS LSTM forward LSTM backward

Input

Vector Features

Fig. 9.1 The proposed method

9.3.1 Word Embeddings

Word embeddings is a technique which converts it to vectors. A large external data

with various models is used to train those vectors in advance. We use the Word
embeddings to map words into real number vectors. The word embeddings has the
same number of columns as the words in the vocabulary dictionary; the number of
dimensions of the vector representing the word is the number of lines. Thus, when
converting a word to a real vector, we only need to access the word embeddings and
retrieve the corresponding column. The values in the matrix are achieved by applying
algorithms such as word2vec (CBOW, Skip-gram [11]), GloVe [9] or FastText [10]
on unlabeled text datasets. In addition, we could also initialize these values randomly
and gain general updates during the training of the labeling model.
Continuous Bag of Words (CBOW), one kind of word embeddings developed by
Mikolov et al. [11]. In the CBOW model, for a given target words multiple words
represent as context. Figure 9.2 shows CBOW model.
Although the word embeddings may contain the semantics of a word on several
levels, there may be data sparseness problems; for example, words with very low
frequency of representation presented by vectors with a majority of components with
values close to 0, words not in the dictionary list, words with misspellings. To solve
this problem, it could be done by increasing the embedding word information for
word level characters; then combining this vector with the word embeddings vector
to create a unique vector representing a word.
We use the pre-trained Word embedding—FastText [10] in this research. To
improve system performance, we combined some syntax features with Word embed-
dings as input to the Bi-LSTM model. These syntax features are automatically
generated by specific toolkits, so the actual input of our system is just plain text.
104 B. T. Hung

Fig. 9.2 CBOW—Continuous bag of words model

Table 9.1 An example of generating features automatically

Word POS CHUNK NER
Ở E O O

phi trường N B-NP B-LOC

quốc tế N I-NP B-LOC

Kuala Lumpur NP I-NP I-LOC

, CH O O
người NC B-NP O

tài xế N B-NP O

Taxi FW B-NP O

These additional features include POS, CHUNK embedded in the dataset, and regular
expressions. Table 9.1 presents the features that are automatically generated.

9.3.2 Bidirectional Long Short Term Memory (Bi-LSTM)

Long Short Term Memory [12] is a modification of the Recurrent Neural Networks
(RNN). The advantage of the LSTM is which can remember the knowledge about
previous outputs in the long-term.
Figure 9.3 shows a diagram of a simple LSTM cell with five elements: a forget
gate f , an input gate i, an output gate o, a recurring cell state c and hidden state output
9 Combining Syntax Features and Word Embeddings … 105

Fig. 9.3 Standard LSTM

h. Given a sequence of vectors (x1, x2, …, xn), σ is the logistic sigmoid function, the
following formulas describes how to calculate the hidden state ht of LSTM at time t.

f t = σ (W x f xt + W h f h t−1 + W c f ct−1 + b f ) (9.1)

i t = σ (W xi xt + W h i h t−1 + W ci ct−1 + bi ) (9.2)

i t = σ (W xi xt + W h i h t−1 + W ci ct−1 + bi ) (9.3)

ct = f t ∗ ct−1 + i t ∗ tan h(W xc xt + W h c h t−1 + bc ) (9.4)

h t = ot ∗ tan h(ct ) (9.5)

In many task of natural language processing [13–17], Bidirectional LSTM is

suitable model. The Bidirectional LSTM is made by combining forward LSTM and
backward LSTM. For Vietnamese Named Entity Recognition, we use two Bi-LSTM
layers with softmax layer at the top to detect entities named after the sequence.
Inputs are a combination of word embedding and syntactic features, and the output
is a probability distribution on named entity tags. Figure 9.4 describes in details our
model.
106 B. T. Hung

Yt-1 Yt Yt+1

Ct-1 Ct Ct+1

Forward
LSTM

Rt-1 Rt Rt+1

Backward
LSTM

Lt-1 Lt Lt+1

Vector Vector Vector

Features Features Features

Fig. 9.4 Bi-LSTM layers in our model

9.4 Experiments

9.4.1 VLSP Dataset

The NER of VSLP 2016 dataset is used to conduct our experiments [3]. This dataset
includes four types of named entities such as organizations (ORG), people (PER),
places (LOC), and other named entities (MISC). Named entities are labeled with
the IOB notation as in the common CoNLL 2003 tasks [2] with 9 labels including
B-ORG, I-ORG, B-LOC, I-LOC, B-PER, I-PER, B-MISC, I-MISC and O for other
factors [3]. The number of annotated entities for training and testing are shown in
Table 9.2. Table 9.3 presents the training, developing and testing datasets.

Table 9.2 Number of named

Entity Training Testing
entity in the VLSP 2016
dataset ORG 1213 274
LOC 6245 1379
PER 7480 1294
MISC 282 49

Table 9.3 Statistics of the

Dataset Training Developing Testing
dataset
Number of sentence 16.861 2000 2831
9 Combining Syntax Features and Word Embeddings … 107

9.4.2 Experiment Results

We used Pyvi to tokenize our dataset, chunking of NLTK toolkits (3.3), pre-train word
embeddings by Bojanowski et al. [10]. For our training models, we use Keras deep
learning tools which provide many useful layer and parameter. We used following
parameters for Bi-LSTM: Training iteration (epochs): 500, Number of hidden nodes:
128, Adam Optimization function, Batch size: 500, Activate function in the output
layer: Softmax, Loss function: Categorical cross entropy.
We use the F1 score measure to evaluate the effectiveness of the system.

2 ∗ pr ecission ∗ r ecall
F1 =
pr ecission + r ecall

Precision (P) and Recall (R) are percentages of correctly named entities defined
by the learning system and percentages of identified entities with corresponding
names contained in the dataset. To calculate the F1 score of NER system we use the
evaluation tool used for the previous systems [18].
We did experiments in 1 and 2 Bi-LSTM layers. The effectiveness of the Bi-LSTM
model combined with word embedding is shown in Table 9.4. This Table shows that
using 2 Bi-LSTM layers gives better results than 1 Bi-LSTM layer. And Fig. 9.5
describes the F1 score of 1 Bi-LSTM and 2 Bi-LSTM layers.
Dropout helps reduce “over-matching” in the model that why we did experiments
in different dropouts. Table 9.5 presents the experimental results of the model with
different dropout ratios and Fig. 9.6 shows the F1 score of the different dropouts. By
these experiments, we chose dropout as 0.2 for the best results.
To evaluate the result of combining syntax features and word embeddings, we did
experiments with different combinations of word embeddings with syntax features:
Chunk, POS and Regex. Table 9.6 is the result of the experimental process and
Fig. 9.7 shows the F1 score of the different combining of syntax features with word
embeddings.
The above experiments show that, if only the word embeddings is used as input for
the model, the optimal result is only 74.04%. Adding each syntax feature improved
the efficiency of the model. To increase the accuracy of the model, we integrated

Table 9.4 The results in 1 and 2 Bi-LSTM layers

Model ORG PER LOC MISC ALL
1 Bi-LSTM layer P 44.10 72.70 82.22 85.15 74.83
R 40.88 62.15 80.64 74.29 68.91
F1 42.39 66.91 81.41 79.32 71.74
2 Bi-LSTM layer P 49.85 72.77 83.63 84.14 75.90
R 50.51 65.73 82.48 78.37 72.27
F1 50.07 69.06 83.05 81.07 74.04
108 B. T. Hung

Fig. 9.5 The F1 score of 1 Bi-LSTM and 2 Bi-LSTM layers

Table 9.5 The results in using different dropouts

Entity Dropout = 0 Dropout = 0.2 Dropout = 0.5
P R F1 P R F1 P R F1
ORG 41.09 32.92 36.43 83.94 83.64 83.79 49.85 50.51 50.07
PER 67.35 59.23 62.97 92.74 92.88 92.81 72.77 65.73 69.06
LOC 80.98 76.79 78.79 93.69 92.21 92.94 83.63 82.48 83.05
MISC 84.09 64.49 72.73 87.76 100.00 93.48 84.14 78.37 81.07
ALL 71.97 72.26 68.27 92.29 91.83 92.06 75.88 72.26 74.02

Fig. 9.6 The F1 score of the different dropouts

9 Combining Syntax Features and Word Embeddings … 109

Table 9.6 The results of using different syntax features

Syntax feature P R F1
Word 75.88 72.26 74.04
Word + Chunk 90.73 83.18 86.79
Word + POS 84.23 87.64 85.90
Word + Chunk + POS + Regex 91.09 93.03 92.06

Fig. 9.7 The F1 score of the different combining of syntax features with word embeddings

syntax features such as Chunk, POS, regular expressions as input to the NER model.
The model combines all the syntax features: Chunk, POS, Regex for the highest F1
accuracy of 92.06%.

9.5 Conclusion

In this research, we proposed Bi-LSTM deep learning method which is less compli-
cated than the previous models such as the combination model LSTM-CNN-CRF, etc.
and incorporated the automatic syntax features appropriately such as Chunk, POS,
Regex with pre-trained word embeddings for Vietnamese Named Entity Recogni-
tion. The result has shown the effectiveness in combining all the syntax features:
Chunk, POS, Regex with word embeddings in identifying Vietnamese proper names
with the highest accuracy of 92.06% on NER VLSP 2016 dataset. In future work, we
would like to train in large data and apply the result to other tasks of natural language
processing.
110 B. T. Hung

References

1. Sharnagat, R.: Named Entity Recognition: A Literature Survey. Center for Indian Language
Technology, Mumbai, India https://doi.org/10.1023/A:1007558221122
2. Tjong, E.F., Sang, K., De Meulder, F.: Introduction to the CoNLL-2003 shared task: language-
independent named entity recognition. In: CoNLL (2003). https://doi.org/10.3115/111917611
19195
3. Nguyen, T.M.H., Vu, X.L.: VLSP 2016 shared task: named entity recognition. In: Proceedings
of Vietnamese Speech and Language Processing (VLSP) (2016). https://doi.org/10.15625/
1813-9663/34/4/13161
4. Lample, G., Ballesteros, M., Subramanian, S., Awakami, K., Dyer, C.: Neural architectures for
named entity recognition (2016). arXiv:1603.01360
5. Young, T., Hazarika, D., Poria, S., Cambria, E.: Recent trends in deep learning based natural
language processing. IEEE Comput. Intell. Mag. (2018). https://doi.org/10.1109/MCI.2018.
2840738
6. Sundermeyer, M., Ney, H., Schluter, R.: From feedforward to recurrent LSTM neural networks
for language modelling. J. IEEE/ACM Trans. Audio Speech Lang Process 3, 517–529 (2015).
https://doi.org/10.1109/TASLP.2015.2400218
7. Yadav, V., Bethard, S.: A survey on recent advances in Named Entity Recognition from
deep learning models. In: Proceedings of the 27th International Conference on Computational
Linguistics, pp. 2145–2158 (2018). arXiv:1910.11470
8. Pham, T.-H., Le-Hong, P.: End-to-end recurrent neural network models for Vietnamese named
entity recognition: word-level versus character-level (2017). arXiv:1705.04044
9. Pennington, J., Socher, R., Manning, C.D.: Glove: global vectors for word representation. In:
Empirical Methods in Natural Language Processing (EMNLP), pp. 1532–1543 (2014). https://
doi.org/10.3115/v1/D14-1162
10. Bojanowski, P., Grave, E., Joulin, A., Mikolov, T.: Enriching word vectors with subword infor-
mation. Trans. Assoc. Comput. Linguist. 5, 135–146 (2017). https://doi.org/10.1162/tacl_a_
00051
11. Mikolov, T., Sutskever, I., Chen, K., Corrado, G.S., Dean, J.: Efficient estimation of word
representations in vector space. In: Proceedings of International Conference on Learning
Representations (ICLR-13): Workshop Track (2013). arXiv:1301.378
12. Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9, 1735–1780
(1997). https://doi.org/10.1162/neco.1997.9.8.1735
13. Wang, P., Qian, Y., Soong, F.K., He, L., Zhao, H.: Part-of-Speech Tagging with Bidirectional
Long Short-Term Memory Recurrent Neural Network. Cornell University (2015). arXiv:1510.
06168
14. Huang, Z., Xu, W., Yu, K.: Bidirectional LSTM-CRF models for sequence tagging (2015).
arXiv:1508.019910
15. Hung, B.T.: Domain-specific versus general-purpose word representations in sentiment analysis
for deep learning models. In: Frontiers in Intelligent Computing: Theory and Applications,
pp. 252–264, Springer (2019). https://doi.org/10.1007/978-981-32-9186-7_27
16. Hung, B.T.: Integrating diacritics restoration and question classification into vietnamese ques-
tion answering system. Spec Issue Adv. Eng. Comput. Sci. J. (ASTESJ), 4(5), 207–212 (2019).
ISSN 2415–6698. https://doi.org/10.25046/aj040526
17. Hung, B.T.: Vietnamese question classification based on deep learning for educational
support system. In: The 19th International Symposium on Communications and Information
Technologies, ISCIT, vol. 9 (2019) https://doi.org/10.1109/ISCIT.2019.8905237
18. Sundheim, B.M.: Overview of results of the MUC-6 evaluation. In: Proceedings of the 6th
Conference on Message Understanding, MUC6’95, pp. 13–31. Association for Computational
Linguistics, Stroudsburg, PA, USA (1995). https://doi.org/10.3115/1119018.1119073
Chapter 10
Human Activity Recognition by Deep
Convolution Neural Networks
and Principal Component Analysis

Amir A. Aljarrah and Ali H. Ali

Abstract Human activity recognition (HAR) is a domain that has a great deal
of interest in recent years. Because it’s used in various application domains like
(critical situations for the elderly or Firefighter, detect about injured in different
sported and other application in domain Medicinal, Military, and Technological).
Moreover, HAR are categorized by video, images or sensors. The multi-sensors
that stabilized on the human’s body provides raw data. The raw data consist
of multi-channel in a pattern time series signals and activities that are prede-
fined. There are so many types of sensors which are utilized in HAR such as
accelerometers, gyroscopes, magnetometer or electrocardiogram. For automatic
feature extraction from this data to the classification of the activities, we propose new
design of convolutional neural networks (CNNs) and Principal Component Analysis
(PCA). The proposed system is estimated using dataset five datasets; (Mhearth),
(Sensors-Activity-Recognition-Dataset-SHOAIB), (REALDISP), (REALWORLD)
and (Activity Recognition Dataset), where the accuracy equals 99.8, 99.44, 99.85,
96.90, 98.68%, Sequentially.

Keywords Classification · Human activity recognition (HAR) · Deep learning ·

Wearable sensor · Machine learning · Convolutional neural networks (CNNs)

10.1 Introduction

HAR is a very significant region of research in filed Artificial intelligence in general.

In order to analysis human behavior and human interaction with the computer.
Research in these areas used different algorithms for machine learning for recognize

A. A. Aljarrah (B)
Department of Mathematics, Faculty of Computer Science and Mathematics, University of Kufa,
Najaf, Iraq
e-mail: [email protected]
A. H. Ali
ECE, University of Kufa, Najaf, Iraq
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 111

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_10
112 A. A. Aljarrah and A. H. Ali

complex and simple physical actives such as standing still, lying down, walking,
jogging, jumping front and back and etc.
Human activity recognition is done in two ways: First way vision-based activity
recognition, where recognition of the activity by using angles and distances between
joints [1]. The second way is the wearable sensors activity recognition, where recog-
nition of activity by using various sensors the attachment to the body [2]. To recognize
the human activities correctly, this process requires the feature extraction from raw
data. The feature extraction is done in two ways: First way is hand crafted from
the raw data like statistical features such that standard deviation, medium absolute
deviation median, mean, min smallest, and max largest in the vector in time domain
[3]. Covariance features, number of peaks, magnitude, spectral power features, and
spectral peak features in frequency domain [4]. Depend on feature extraction from
time and frequency domain [5, 6]. When use the feature extraction by hand crafted
that requires knowledge about the signals. On the hand, the second way is based
on approach to learning, like deep learning. This a way could feature extraction by
automatically [7].
The proposed use principal component analysis and convolution neural networks
with two-dimensional kernel to recognize on the different physical activity of the
public life of humanity. Datasets used focus on sensor to human activity recognition,
where those datasets consist of multiple sensors. This paper is orderly as follows: In
Sect. 10.2, review related work of all worked in datasets. Section 10.3, more desires
of the dataset. After that Sect. 10.4 for Deep Learning Architectures. Then, Sect. 10.5,
is architecture of the method.

10.2 Related Works

This section displays the works that connected of this paper. This related work will
be subdivided into two parts: machine learning, human activity recognition datasets
standard. Machine learning is one of the most important field of AI that provides the
systems with the ability to automatically learn. The process of learning begins from
data or observations, such as samples or examples, in order to search for patterns to
make better decisions in the future. ML technology powers many aspects of society
such as: financial services, media, service, social media, creative arts and healthcare,
with many applications such as recommendations on e-commerce websites, to iden-
tify objects in images, select the relevant results of search and convert speech into
text. These applications use a technique that is called deep learning (DL) [8]. DL is
an umbrella term where many architectures resides such as: deep convolutional nets
and recurrent nets. Deep convolutional nets have achieved excellent results in audio,
video and image recognition [9], while recurrent nets have achieved excellent results
with sequential data such as speech and text [10]. The convolution neural networks
is types of deep learning, most commonly applied to analyzing visual imagery. The
year 2012 has witnessed the introduction of AlexNet that consists of 5 convolution
10 Human Activity Recognition by Deep Convolution Neural Networks … 113

layers, Max Pooling layers (ReLU), dropout layers for reducing the problem of over-
fitting, and three fully connected layers with size of filters in first layer equal 11 ×
11. The training of AlexNet model on a dataset containing over 15 million images
from over 22,000 categories has achieves 15.4% error rate [9]. Chowdhury et al.
present a method that consists of 22 layers. This method has parameters 12× fewer
parameters than AlexNet. They use nine inception modules in the whole architecture
with over 100 layers in total. This model is achieved a 6.7% error rate [11].
A. K. Chowdhury et al. present a method that consists from three steps of activity
recognition are sensors placement, data pre-processing and data classification. Using
two sensors from dataset right wrist and left ankle with pre-processing and feature
extraction from each axis 45 features in time domain, and frequency domain depends
on two second for each 100 samples without overlap. After that normalization all
feature by zero mean for classification one state have n-model and m-class depend on
voted. Finally, this method has achieved high accuracy by Support Vector Machines
(SVM) where F1 -score equal 91.64 [5]. Rueda et al. have presented a two methods
CNN-pf and CNN-pff. In CNN-pff, they exploit three convolution kernel group in
the first layer, one accelerometer and two gyroscopes. CNN-pf only two different
convolution kernels are used. The size of input equals 60 samples with overlap of 50%
Get accuracy of 91.94% [12]. Yun et al. have presented a CNN method that consists of
input layer, 2 convolutional layers, first 3 × 3 with 32 filters and second 3 × 3 with 64
filters, 2 subsampling layers, 2 fully-connected layers. The segment of dataset to 60
samples with 50% overlap. In Experiment, they divide the dataset into data training
90% and data testing 10%. The accuracy equals 98.2% [13]. Lyu et al. have proposed
LSTM-CNN method comprising RNN with 6-time steps, convolutional layer that is
divided into (4 filters, 3 filters and 2 filter), one Max Pooling and Fully Connected
layer. The divided of dataset into data training 70%, data validation 15% and data
test 15%. After that split of data into the window with size 128 samples from 2.56 s
with overlap 50%, from each window extract 561 feature in frequency domain and
time domain. The result equal 95.5% [14].
Chen et al. have proposed a method that consists of many steps. First step removes
noise by many filters, divided dataset by one second with overlap 50%. The second
step is to extract features from time and frequency domain (89) features. The third
step is normalization by Z-score. Forth step is classification by three algorithms
(SVM, Random forest and K-NN). The fifth step is an evaluation method by used
tenfold cross validation. The best accuracy equal 96.26% [15]. Ehatisham et al. have
proposed method that consists of many steps, first used average smoothing filter 1
× 3 after that split data to five second with 50% overlap then extract 16 features
from the time domain. Finally, classification by used SVM. the technique used in
divided dataset is K-fold cross validation, where the input data from magnetometer,
gyroscope, and accelerometer sensors of a smartphone for classification six activities.
The accuracy equal 97.95% [16]. Shoaib et al. presented a method used accelerometer
and gyroscope for classification seven activities. The divided dataset with two second
and 50% overlap by used tenfold cross validation. The classification used modes in
WEKA are K-NN, decision tree, and SVM, the best result above 97% [17].
114 A. A. Aljarrah and A. H. Ali

Mario et al. have presented a CNN method comprising 15 layers. In the first step,
divided dataset by using fivefold cross-validation technique (data training 80% and
data test 20%). After that, splitting the data into sliding window with 5 s and 50%
overlap to classification the climbing stairs down and up, jumping, running/jogging
and walking only. The result F1 score value is 94% [18]. Munoz et al. have presented
a deep recurrent neural networks (DRNNs) that consist of 50 hidden units in layer
LSTM. The input data are divided into 60-sample windows. This method used only
four activities (running, walking, climbing down, and up) from dataset. The best
result equal 97% [19].
Al-Fatlawi et al. have presented reducing the number of sensors to six necessary
sensors. The feature selection by using data mining algorithm and classification by
algorithm deep belief network. In the end, the sensors S1, S4 and S5 are low necessary.
The result equal 96.52% [20]. Damas et al. presented meta-classifier including three
levels (laser classification, N node and fusion of the decision) for classification. The
extract of five features from each six second in dataset. The accuracy equal 93% [21].
Nabian and Mohsen have presented a method to classification of twelve activities
in dataset. The dataset from left ankle, chest and right wrist. After that, applied
normalization on the data, where the result between zero and one, then divided
dataset into training 80% and testing 20%. Finally, used PCA to preprocessing and
K-NN algorithm for classification. The accuracy is equal 99.4% [22].
Shoaib et al. have presented a method to select feature, divide data into small
segments (depend on 2 s sliding window with overlap 50%) and add magnitude
for each sensor, then extracted two features from time domain: standard deviation,
and mean for all dimensions of each sensors. From all sensors, number of features
equals 24. Using tenfold (cross validation technique) to evaluate different classifiers.
Imbalance dataset reduces all data from all sensors to low data from sensors [23].

10.3 Dataset

There are many well-known repositories of datasets specifically acquired for the
purpose of extracting knowledge and/or machine learning purposes. Examples of
such repositories include kaggle, UCI Machine Learning Repository and Amazon
Datasets.
Mhealth Dataset [24]: This dataset have twelve different physical activities such
as (cycling, jogging, running, standing still, etc.). It is collected from ten volunteers
by used three sensors position on right wrist, left ankle, and chest. The sample rate
equals 50 Hz, where provides 23 features.
Sensors-Activity-Recognition-SHOAIB Dataset [25]: This dataset has seven
different physical activities such as (Walking, standing, jogging, biking, Etc.). It
is collected from ten volunteers (ages between 25 and 30 years) by using five smart-
phones positions in their left jean’s pocket, their right jean’s pocket, the right upper
arm, belt position towards the right leg, and the right wrist.
10 Human Activity Recognition by Deep Convolution Neural Networks … 115

REALWORLD Dataset [26]: This dataset has eight different physical activities
such as (lying, jogging, running, standing, Etc.). It is collected from fifteen volunteers
by using seven accelerations on body in positions forearm, chest, thigh, upper arm,
head, shin, and waist with sample rate equals 50 Hz.
REALDISP Dataset [21]: This dataset has 33 different physical activities such as
(jump up, frontal hand claps, rowing, cycling, Etc.). It is collected from seventeen
subjects by using nine sensor positions on left thigh, left calf, right thigh, right calf,
left lower arm, right lower arm, right upper arm, left upper arm and back with sample
rate equals 50 Hz.
Activity Recognition Dataset [27]: This dataset has seven different physical activ-
ities such as (walking upstairs, downstairs, running, walking, standing, Etc.). It is
collected by using four Samsung Galaxy S2 smartphones on body positions of the
participants in belt, right jeans pocket, wrist and arm for four participants with sample
rate equals 50 Hz.

10.4 Deep Learning Architectures

Representation learning (RL) is a group of methods in AI, that allows for a machine
(algorithm) to be fed with raw data and detect automatic for the representations that
needed for classification or other. Deep learning methods are (RL) methods with
multiple levels of representation. It is obtained through structure simple, but non-
linear. Works each one transforms the representation in one level into higher level
and slightly more abstract. For example, in classification tasks, input image consists
from an array of pixel values. First layer of representation is the absence or presence
of edges at orientations and particular locations in image. In second layer, try to
detects motifs through spotting specific arrangements of edges with don’t focus on
small change in the positions of edge. Third layer collect these motifs into group
that correspond to parts of objects. Fourthly, layer detects objects as combinations of
these parts. The advantage of DL is that these layers used for extract features from
raw data with not designed by human engineers [8].

10.4.1 Convolutional Neural Network (CNN)

Convolutional networks (CNNs) is a specialized type of NN used to processing the

data has a grid, examples of data like image data, which can be thought of as a two
dimension grid, time-series data, which can be one dimension grid taking samples
at time intervals. Its consist from group of neurons that have biases and weights;
each neuron receives inputs (data) and then performs a product with optionally a
non-linearity [28].
116 A. A. Aljarrah and A. H. Ali

10.4.2 Convolutional Layer (CL)

The task of CL is to detect local conjunctions of features from past layer. CL will
compute the result of neurons that are connected to native regions in the past layer,
each computing a dot product between a small region they are connected to in the
input volume and their filters (weights). Its convert input feature maps (IN) into
output feature maps (OUT), each containing multiple units, each unit in the output
feature map (N × N × F) is connected to local patches of the units (T × T × C) in
the input through filter W [F](T × T × C × 1) in filter bank W (T × T × C × F).
A formal mathematical description in Eq. (10.1([29]:
⎛ ⎞

c 
T 
T
Out(f, x, y) = α⎝ IN(Sx + i, Sy + j, c) × W(i, j, c, f) + B(f)⎠
c=0 i=0 j=0
(10.1)

where S is astride, α is activation function such as ReLU, c is the number of

channels in the input, T is size of width and height in filter plane, B is bias, N is the
height and width of the output feature map, F equal number of filters per layer, and
x, y, f are bounded by: y, x ∈ (0,1, ..., N) and f ∈ [0,1, ..., F) [29],as show in Fig. 10.1.
The three parameters is responsible of size the output volume: zero-padding,
depth and stride. (1) Zero-padding is added zeros volume around the border, (2)
Depth corresponds the number of filters in layers. (3) Stride corresponds moves the
filters on image, when the stride is equal 1, this mean move the filters one pixel at a
time; when the stride is equal 2, this mean filters jump 2 pixels at a time, can compute
the spatial size of output as a function of the input volume size (X), receptive field
size of the convolution layer neurons (F), stride (S), and the amount of zero padding
used (P) on the border. The correct formula for calculating how many neurons “fit”
is given by Eq. (10.2) [30]:

X − F +2× P
H=W = +1 (10.2)
S
Height (H) and width (W), C is the depth is equal number of filters. The all
parameters in a convolutional layer show in Eq. (10.3):

Fig. 10.1 Example of a multi-layer CNN [29]

10 Human Activity Recognition by Deep Convolution Neural Networks … 117

ρ = (H × W × C + 1) × NF (10.3)

where 1 is the bias, and NF Equal number of filters.

10.4.3 Batch Normalization Layer

This layer is increase CNN training speed and decrease of the sensitivity to initial
parameter network. It is used between CN and non-linearities. It normalizes the
inputs xi by find variance σ B2 and mean μ B over a mini-batch and all input channel,
also compute normalized activations as explained in Eq. (10.4) [31]:

xi − μ B
Bi =  (10.4)
σ B+
2

where Epsilon (ε) enhancement numerical stability that mini-batch variance is

very few; where inputs with zero mean and unit variance are not optimized of layer
which batch normalization layer, it can shifts and scales of the activations as shown
in Eq. (10.5) [31]:

yi = γ Bi + β (10.5)

where β, γ are offset and scale features, they learnable parameters which updated
by training of network. This layer compute variance and mean during stores and
training set [31].

10.4.4 Rectified Linear Unit (ReLU) Layer

The activation function is most widely used in deep learning applications. The ReLU
activation function performs depending on threshold operation to each input element,
it outputs 0 when input values S < 0, and conversely, it result of outputs a linear
function if input values S ≥ 0, It is given by Eq. (10.6) [30]:

Si , i f Si ≥ 0
f (S) = max(0, S) = (10.6)
0, i f Si < 0

where S is input value and f (S) is output.

118 A. A. Aljarrah and A. H. Ali

10.4.5 Pooling Layer

It is reducing the size of the features maps that combine neighboring value of a specific
area of the feature map into a single representative value. This layer is used after
multiple stages of other layers (convolutional, non-linearity layers and normalization
layer) in order to minimize the calculations; size result from this layer depend on
two parameters are extent of the filter (size of filters) F (L) and stride S (L) . It takes
input size from previous layer W (L−1) × H (L−1) × D (L−1) and provides an output
with size W (L) × H (L) × D (L) , See Eqs. (10.7)–(10.9) [32]:

D(L) = D(L−1) (10.7)

H(L−1) − F(L)
H(L) = +1 (10.8)
S(L)
W(L−1) − F(L)
W(L) = +1 (10.9)
S(L)

where D1(L−1) is number of 2-dimensional feature maps in convolutional layer

(L − 1), each one have a size H (L−1) ×W (L−1) . Two methods for reduced dimensions
in this layer, average pooling and max pooling. First used mean of values within a
determinant window region and result one value. The second finding highest value
within the determinant window region [33].

10.4.6 Fully-Connected Layer

There are many layers like (convolution, pooling and etc.); will convert data into
vector and input to this layer, all neurons in this layer content full connections to all
nodes in the previous layer, as shown in Eq. (10.10) [34].

Fc(l+1)
p = clj W lj p + αlp (10.10)
j

where W lj p is matrix weight from node j of L layer to the p node of layer (L +

1) and clj is the data of node j in the layer L. The advantage of this layer combines
all features from the previous layer together to create a more complex model. It
gives network more power for prediction, for example, predict of the human face, in
convolutional layer and another layer might be able to find features like nose, ears,
faces, and etc.; but do not know where these features should be in the face. Fully
connected layer is result output vector consist of n dimensions, each value in this
vector contain the probabilities of class, where n number of classes in network.
10 Human Activity Recognition by Deep Convolution Neural Networks … 119

10.4.7 Softmax Layer

The Softmax layer is often used to predict probabilities associated with a multi-
class distribution. It resizes the outputs of each unit to be between 0 and 1, where
summation of outputs equal one. Output of softmax layer is categorical probability
distribution, where give probability of any true class [33] as shown in Eq. (10.11).

eS j
so f t Max(S) j = n (10.11)
i=1 eSi

n
where 0 ≤ so f t Max(S) j ≤ 1 and so f t Max(S) j = 1.
i=1

10.4.8 Classification Layer

it takes an input value from Softmax layer and assigns each input into one class from
the M by using the cross entropy function depend on the Eq. (10.12) [29]:


C
L(b, y) = − bi log(yi ) (10.12)
i=1

where C number of classes, bi equal one of target class and zero others and yi is
the predicted output value from Softmax layer.

10.5 Architecture of the Method

The design of system consists of six steps: data acquisition, feature engineering,
model design, training, performance analysis and fine tuning step. Figure 10.2.
describe structure of the proposed system.

10.5.1 Data Acquisition

The raw data is obtained from multi-sensors that stabilized on the human’s body,
such as gyroscopes, magnetometer, accelerometers, and electrocardiogram. Feature
engineering: Feature engineering is one of the most important processes that occur
prior entering the data into a method. Data raw has been often inconsistent, lacking
in certain behaviors, and is likely to contain many errors like classless of data. These
problems have an effect on predict of this method. In this section, explain steps of
120 A. A. Aljarrah and A. H. Ali

Start

Data Acquisition

Feature Engineering

Model Design

Training

Fine Tuning

Performance Analysis

No
High

Yes

End

Fig. 10.2 Workflow of Deep Learning system

feature engineering (dataset collection, classless data, feature selection, a division of

the dataset, normalization of data and principle component analysis). Data collection
is collection and measuring of the information on summation targeted variables, then
enables one to answer relevant questions and evaluate results. Data collection is a
component of the research in all fields such as data mining, statistical methods and
machine learning. In many cases, the received data from sensors is collected in
isolated files for each person or each sensor or each activity and it has different
extensions. To process and divide the dataset, data must be collected in one file and
pre-processing the procedure, this requires perception on how the data is collected
from sensors and stored in isolated files. Classless Data: Several datasets contain
samples that do not contain a label (Null), because sensors might not determine the
type of label during the registration of the data or may be the result of transitions
between activities. This type of samples is treated similar to the noise because of the
10 Human Activity Recognition by Deep Convolution Neural Networks … 121

overlap between samples and data that contain labels. It is difficult to deal with such
a data in the training of particular algorithms; for example, a set of features is given
to a particular label for the training and then the same features that given different
labels. For removing data contains class Null that is depend on the Eq. (10.13):


n
D= Dai |(yi ! = 0) (10.13)
i=1

where D is new dataset, Dai is dataset,yi is physical activity in the sample i

and n is the number of samples in the dataset. Feature selection: This part selects
important features form the training process, where other features can remove for
every sample because such features may adversely influence the training process like
name of subjects, ID of person, name, and the ID of dataset and time. Division of
Data: All methods in machine learning examine from examples (instances). Groups
of instances called datasets, to determine response or accuracy of this method; dataset
should be divided into subsets. The divided process varies is depending on the size of
the dataset or the type of problem. The dataset is dividing into (depend on Hold-Out
Cross-Validation): (1) Training dataset that equal to 60%, this instance is feed into
our model for training. (2) Validation dataset equals to 20%, such instance is used
to adjust and update for the best possible weights. (3) Testing dataset that equal to
20%, where this instance is applied to validate the accuracy of the current model.
More advantages of the current technique are (Avoid overfitting of the model
by using only testing dataset and applying the result from the validation dataset to
compare between several datasets by using the same model).
The disadvantage is when dataset is never enough to train our method. It is known
that all structures of CNN deal with two dimensions or three dimensions of dataset,
but the gained dataset from sensors is one dimension (series theme). This problem
can be solved by collect each group of samples as two dimensions form which is
depending on the sample rate; for example, when sample rate equal to 50, this means
each sensor gives 50 samples per second. Practically, to determine any activity related
to the body of human, we do not need 50 results per second, but we can suffice of
one result per second Mostly. After various tests, 50 samples can be determined
for each block which is a suitable value for many datasets (as shown in chapter
four). This means converting the dataset from time series to two dimensions (like
an image). Dividing the dataset into blocks that have 50 rows, and then choosing
the blocks randomly for each subset (training, validation, and testing) according
implementation steps (1).

Algorithm (1): Division of Dataset into (Training, Testing, and Validation)

Input: Features of dataset
Output: Features of dataset split into validation, training, and testing
Begin:
Step 1: Remove irrelevant data:
(continued)
122 A. A. Aljarrah and A. H. Ali

(continued)
Algorithm (1): Division of Dataset into (Training, Testing, and Validation)

m
Dnew = D|( j! = 4, 5)
j=1

Where M is number of features and D is dataset

Step 2: Select random index without repeating as the following:
X = random(1 : h : Dnew )
Where height(h) = 50.
Step 3: Dividing 60% training, 20% validation, and 20% testing as follows:
|T r | = (|X |/100) × 60
|V a| = (|X |/100) × 20
|T e| = (|X |/100) × 20
Step 4: Select index of training, validation, and testing as follows:
IT r = X (1 : |T r |)
I V a = X (|T r | + 1 : |T r | + |V a|)
IT e = X (|T r | + |V a| + 1 : |T r | + |V a| + |T e|)
5. Find training data as follows:
DT r = Dnew (IT r : IT r + h − 1, :)
(5.14)
6. Find validation data as follows:
DV a = Dnew (I V a : I V a + h − 1, :)
(5.15)
7. Find testing data as follows:
DT e = Dnew (indT e : indT e + h − 1, :)
(5.16)
End

After that select only the blocks has one label by used follows:
10 Human Activity Recognition by Deep Convolution Neural Networks … 123

10.5.2 Normalization of Data

124 A. A. Aljarrah and A. H. Ali

The value of features predicts from using z-score have zero mean and one stan-
dard deviation. So, using a z-score must calculate the mean and also the population
standard deviation. Using the Eq. (10.18) for calculating the mean of all features
separate. After that calculated standard deviation depends on the Eq. (10.19). Then
minus mean from each features amount mean and divided on standard deviation, see
the Eq. (10.17):

eF
i − MF
ZF
i = (10.17)
δF

where eiF is element i in feature F, δ is standard deviation and MF is mean of

feature F.

1
n
M= xi (10.18)
n i=1

where x is an entire set of samples and M indicate to mean of the set , n is the
number of samples in the set ,xi is the sample from set .


n
− M)2
i=1 (x i
Sd = (10.19)
n−1

where Sd is the standard deviation. Each element of a Z-score has a certain

meaning, for example, a value of elements which is less than zero represents an
element less than mean and vice versa. If value equal to zero represents an element
equal to mean or equal to one represents an element that is one standard deviation
greater than mean [35].

10.5.3 Principle Component Analysis

PCA is technically used for reducing dimensions in dataset that depends on the
maximum variance between all features. It is has more advantages that remove shared
data which do not help in the training process and less time in training. This is
based on the creation of new directions that maximize contrast without change in
dataset original. This direction is obtained by using Singular Value Decomposition
(SVD). PCA is explained in the following steps [35, 36]:

Step 1: Mean
The first step for PCA is a subtract mean from each feature in the dataset. This
operation produces sets of features whose mean is zero as shown in Eq. (10.18) [35].
The advantage of minus mean from all features is to back to origin a point and easy
determinate the diagonal more important.
10 Human Activity Recognition by Deep Convolution Neural Networks … 125

Step 2: Standard Deviation

The aim of using the standard deviation is to show the average distance from the
mean of dataset to each point. Average distance from the mean of the data set to a
point as shown in Eq. (10.19) [35].

Step 3: Covariance Matrix

The aim of the statistical analysis of these data sets is usually to see if there is
any relationship between the dimensions. The calculated covariance to one dimen-
sion with itself, this operation gives the variance. If it is measured based on two or
more dimensions this give covariance between there dimensions. For example, if we
have three dimensional like (x, y, z), we find covariance between dimensions (x, y),
dimensions (x, z), and dimensions (y, z), as show in Eq. (10.20), which explains the
covariance [35]:
n
i=1 (ai − M)(li − M)
con(a, l) = (10.20)
(n − 1)

where n is the number of samples, M is mean.

Step 4: Calculating Singular Value Decomposition (SVD)

The SVD of any matrix consists of three matrixes, for example A matrix with size
m × n can be written in the formal A = U SV T , Where U, S and V are matrixes with
different size [36]. U is an orthogonal matrix with size m × m, V is an orthogonal
matrix with size n × n and S is a matrix with size m × n. S matrix consists of z
diagonal entries are the σ1 , σ2 , σ3 , . . . , σz non-zero singular values of matrix A and
other entries are all zeros. The expression U SV T is known as the singular value
decomposition of matrix A [37] as shown in Fig. 10.3.
For more understand work of SVD, the example is used; Let Am×n =
Um×m Sm×n Vn×n , Where U T U = I, V T V = I . The columns of matrix U are
orthonormal eigenvectors of A A T and the columns of matrix V are orthonormal
eigenvectors of A T A, and S is a diagonal matrix containing the square roots of
eigenvalues from U or V in decreasing order[37], shown in Algorithm.

Fig. 10.3 Content of Matrix

A
VT
A U S
=
126 A. A. Aljarrah and A. H. Ali

Algorithm (3): Singular Value Decomposition

Input: matrix
Output: U, S and V are matrixes with different size
1. To find matrix U, compute A A T
2. Compute the eigenvalues of A A T depending on the equation Ax = γx, solve γ by setting the
determinant of the coefficient matrix to zero. From γ, we will get on eigenvectors
3. Sort the eigenvector, where the largest eigenvalue is column one and column two is the next
largest eigenvalue and so on to until the smallest of eigenvalue as the last column in the our
matrix
4. Apply the Gram-Schmidt to convert a matrix into orthogonal matrix, where the
gram-Schmidt orthogonalization process combines orthogonalization with normalization,
where compute (w1 , w2 , w3 ,… wn ) that represented an orthonormal as follows [38]:
• v1 = x1 andw1 = v1
v1
v2
• v2 = x2 − (x2 , w1 )w1 andw2 = ,
v2
v3
• v3 = x3 − (x3 , w1 )w1 − (x3 , w2 )andw3 = v3 ,
vn
• vn = xn − (xn , w1 )w1 , . . . . . . ..(xn, wn−1 )wn−1 andwn = 
vn
1. Calculate V by similar calculations of U , but finds the eigenvalues of A T A
2. Find S by taking the square roots of the eigenvalues and populate the diagonal with them,
laying the largest in S11 , Then, the next largest in S22 and so on until the smallest value in the
Smm

Step 5: Choosing the Optimal Number of Principal Components Depending

on Variance
A number of eigenvectors are chosen from all eigenvector in U orthogonal matrix
using high eigenvalue in S. Highest eigenvalue of eigenvector is the principle
component of the data set, where it depends on the Eq. (10.21) [39]:
q
Si
κ = i=1
n < variance (10.21)
i=1 Si

where Si is diagonal value, n is number of all eigenvalue and q is the number of

eigenvalue.

Step 6: Deriving the New Dataset

The final step of PCA is choosing eigenvectors with high eigenvalues, then transpose
this vector with multiply on left of data after subtract mean and divided by standard
deviation.
Proposed design of Human Activity Recognition-Convolutional Neural Network
(HAR-CNN) system. In this section, explaining the architecture of HAR-CNN
design, where it contains of five learning layers-Four convolutional and one fully-
connected. Each layer will be explained with the number of parameters, input and
output size that showing in the Algorithm 4 (Table 10.1).
10 Human Activity Recognition by Deep Convolution Neural Networks … 127

Algorithm (4): Architecture HAR-CNN

Input: Features of dataset
Output: Classification of dataset
Begin:
Step 1: Layer One (Input Layer)
Data input consists of three dimensions namely H , W and D, where H is the height, W is the
width and D is the depth. Applying the normalization by using ‘zerocenter’ technique, the input
size equals the output size [40]
Step 2: Layer Two (Convolutional Layer)
Convolution layer content 8 filter with size 2 × 2 and padding 1. See Eq. (10.1)
Step 3: Layer Three (Batch Normalization Layer)
Batch Normalization layer content of epsilon () equals 0.00001, the offset β equal to zero and
scale factor γ property is equal to one
Step 4: Layer Four (ReLU Layer)
This layer removes negative values by converting them to zero values
Step 5: Layer Five (Max Pooling Layer)
Max pooling layer with the settings: the size of max pooling equals 2 × 2, and stride 2
Step 6: Layer Six (Convolutional Layer)
Convolution layer content 16 filter with size 3 × 3 and padding 1
Step 7: Layer Seven (Batch Normalization Layer)
Batch Normalization layer content of epsilon () equals 0.00001, the offset β = 0 and scale
factor γ property is equal to one
Step 8: Layer Eight (ReLU Layer)
This layer removes negative values by converting them to zero values. See the Eq. (10.6)
Step 9: Layer Nine (Convolutional Layer)
Convolution layer content 32 filter with size 2 × 2 and padding 1
Step 10: Layer Ten (Batch Normalization Layer)
Batch Normalization layer content of epsilon () equals 0.00001, the offset β = 0 and scale
factor γ property is equal to one
Step 11: Layer 11 (ReLU layer)
This layer removes negative values by converting them to zero values. See the Eq. (10.6)
Step 12: Layer 12 ( Max Pooling Layer)
Max pooling layer with the settings: the size of max pooling equals 2 × 2, and stride 2
Step 13: Layer 13 (Convolutional Layer)
Convolution layer content 64 filter with size 2 × 2 and padding 1
Step 14: Layer 14 (Batch Normalization Layer)
Batch Normalization layer content of epsilon () equals 0.00001, the offset β = 0 and scale
factor γ property is equal to one
Step 15: Layer 15 (ReLU Layer)
This layer removes negative values by converting them to zero values. See the Eq. (10.6)
Step 16: Layer 16 (Fully Connected)
Fully connected layer is content 12 neural. (See Eq. (10.10))
Step 17: Layer 17 (Softmax Layer)
The aim of this layer is predicting probabilities that associated with multi-classes distribution
Step 18: Layer 18 (Classification Layer)
Classification layer in CNN receives the input value from the previous layer (Softmax function)
and assigns each of the input into one class from the classes
End
128 A. A. Aljarrah and A. H. Ali

Table 10.1 Layers of HAR_CNN Method

No Type layer Input size Output size Parameters Explanation
1 Input (17, 50, 1) (17, 50, 1) 0 Normalization
‘zerocenter’
2 Convolution (17, 50, 1) (18, 51, 8) 6808 Filter = 8, size = 2 × 2,
padding = 1
3 Normalization (18, 51, 8) (18, 51, 8) 0 ε = 0, β = 0,
γ =1
4 ReLU (18, 51, 8) (18, 51, 8) 0
5 Max Pooling (18, 51, 8) (9, 25, 8) 0 size = 2 × 2, stride =
2×2
6 Convolution (9, 25, 8) (9, 25, 16) 28,816 Filter = 16, size = 3 × 3,
padding = 1
7 Normalization (9, 25, 16) (9, 25, 16) 0 ε = 0, β = 0,γ = 1
8 ReLU (9, 25, 16) (9, 25, 16) 0
9 Convolution (9, 25, 16) (10, 26, 32) 115,232 Filter = 32, size = 2 × 2,
padding = 1
10 Normalization (10, 26, 32) (10, 26, 32) 0 ε = 0, β = 0,γ = 1
11 ReLU (10, 26, 32) (10, 26, 32) 0
12 Max Pooling (10, 26, 32) (5, 13, 32) 0 Size = 2 × 2, stride =
2×2
13 Convolution (5, 13, 32) (5, 13, 64) 133,184 Filter = 64, size = 3 × 3,
padding = 1
14 Normalization (5, 13, 64) (5, 13, 64) 0 ε = 0, β = 0,
γ =1
15 ReLU (5, 13, 64) (5, 13, 64) 0
16 Fully Connected (5, 13, 64) (1, 1, 12) 49,932 Class = 12
17 Softmax (1, 1, 12) (1, 1, 12) 0
18 Classification (1, 1, 12) (1, 1, 12) 0
10 Human Activity Recognition by Deep Convolution Neural Networks … 129

Algorithm of HAR-CNN Learning Method

This section summarizes HAR-CNN method learning process:

Algorithm (5): HAR-CNN Learning Method

Input: Raw dataset
Output: Trained model
Begin:
Stage 1: Collection of dataset
Stage 2: Remove samples that do not contain a label by Eq. (10.13)
Stage 3: Select important features for training process
Stage 4: Divide dataset into training, validation and testing by Eqs. (5.14)– (5.16)
Stage 5: Selection the blocks have one label by implementation steps (2)
Stage 6: Normalization of training, validation and testing by Eq. (10.17)
Stage 7: Determinate the value of variance
Stage 8: Calculate the covariance matrix by using training data in Eq. (10.20)
Stage 9: Calculate the SVD by using training data to determinate the number of eigenvector and
eigenvalue
Stage 10: Determination the eigenvector from left to right that depending on Eq. (10.21)
Stage 11: Multiply the eigenvector with features of training
Stage 12: Multiply the eigenvector with features of validation
Stage 13: Multiply the eigenvector with features of testing
Stage 14: Input the training data to the HAR-CNN model for training
Stage 15: Input the validation data to the HAR-CNN model for adjusting the parameters
Stage 16: Input the testing data (features) to the HAR-CNN model for evaluating our model
End

10.6 Experiments and Results

This section will present result of training HAR-CNN model to classify five datasets
by used several measures for evaluating this method. Firstly, calculating the confusion
matrix. Then, analyzing and evaluating the performance of the method, such as; recall
by Eq. (10.22), precision by Eq. (10.23), accuracy in the Eq. (10.25) and F1-score by
Eq. (10.24) [41]. Dataset divided into training data 60%, validation data 20%, and
test data 20% with random sampling, see Fig. 10.4. Used training data into train the
model, validation data to decide on when to stop training. Finally, test set is used to
find accuracy.

Fig. 10.4 Proportion of validating, training, and testing sets

130 A. A. Aljarrah and A. H. Ali


n
TP j
Recall = 1/n × × 100% (10.22)
j=1
TPj + FN j


n
TPj
Pr ecision = 1/n × × 100% (10.23)
j=1
TP j + FP j

Recall × Pr ecision
F1 scor e = 2 × × 100% (10.24)
Pr ecision + Recall
n
j=1 TP j + TN j
Accuracy = × 100% (10.25)
T otal
In MATLAB train model have (max epochs = 50, mini batch size = 500, gradient
decay = 0.9000 (ADAM), squared gradient decay = 0.999 (ADAM), epsilon =
0.00000001 (ADAM), initial learn rate = 0.001 (ADAM), l2 regularization = 0.0001,
gradient threshold method: ‘l2 norm’, gradient threshold = 1, shuffle = ‘never’,
execution environment: ‘GPU’, sequence length = longest.
In order to compare performance of proposed model. Trained several standard
machine learning models to predict the activities in datasets, namely: kNN, Decision
trees, Naive Bayes and SVM. Their results is shown in Table 10.2.
In order to compare performance of proposed model. Trained several standard
machine learning models to predict the activities in datasets, namely: kNN, Decision
trees, Naive Bayes and SVM. Their results is shown in Table 10.2.

10.7 Conclusions

Deep learning Convolutional Neural Networks algorithm used of HAR tasks. It

has six steps: data acquisition, feature engineering, model design, training, perfor-
mance analysis and fine-tuning step. The feature engineering consists of dataset
collection, classless data, feature selection, a division of the dataset, normalization
of data and principle component analysis (PCA) to reduce number of features. In
addition, used five datasets which consists of different sensors placed on subjects.
Each dataset has different physical activities recorded from subjects such as standing
still, walking, climbing stairs, lying down, running, cycling etc. Registered an accu-
racy of five datasets; (Mhearth), (Sensors-Activity-Recognition-Dataset-SHOAIB),
(REALDISP), (REALWORLD) and (Activity Recognition Dataset), where the
accuracy equals 99.8%,99.44%,99.85%, 96.90%, 98.68%, Sequentially by using
proposed model. The propose model is greater than more standard machine learning
models such as KNN, Decision Tree, Naive Bayes and SVM. Performance results
of proposed model prove feasibility of this study. Further work should include
experimentation using CNN for feature extraction and RNN for classification.
Table 10.2 Comparison of proposed methods with other standard methods
Dataset Performance K-NN (%) Decision tree (%) Naive Bayes (%) SVM (%) HAR-CNN (%)
Mhearth dataset Accuracy 87.54 94.78 84.12 64.67 99.8
F1 -score 86.53 94.29 81.59 62.67 99.8
Sensors-Activity-Recognition-Dataset-SHOAIB dataset Accuracy 98.41 93.91 87.28 73.69 99.44
F1- score 98.50 94.02 87.71 73.91 99.44
REALDISP dataset Accuracy 98.09 94.61 83.61 93.12 99.85
F1 -score 97.35 92.99 80.66 93.66 99.78
REALWORLD dataset Accuracy 94.65 87.49 66.89 Long time 96.90
F1 -score 93.93 85.44 61.77 Long time 96.73
Activity Recognition dataset Accuracy 98.69 94.40 90.65 73.97 98.68
F1 -score 98.35 93.21 89.12 70.09 98.46
10 Human Activity Recognition by Deep Convolution Neural Networks …
131
132 A. A. Aljarrah and A. H. Ali

References

1. Hsu W.C., et al.: Multiple-wearable-sensor-based gait classification and analysis in patients

with neurological disorders. Sensors (Switzerland) 18(10) (2018)
2. Lo Presti, L., Cascia La M., 3D skeleton-based human action classification: a survey. Pattern
Recognit. 53 130–147 (2016)
3. Khatun S., Morshed, B.I.: Fully-automated human activity recognition with transition
awareness from wearable sensor data for mHealth. In: IEEE International Conference on
Electro/Information Technology, vol. 2018-May, pp. 934–938 (2018)
4. Khowaja, S.A., Yahya, B.N. Evaluation of features for classifying low-level human activities
on larger datasets (2019)
5. Chowdhury, A.K., Tjondronegoro, D., Chandran, V., Trost, S.G.: Physical activity recognition
using posterior-adapted class-based fusion of multiaccelerometer data. IEEE J. Biomed. Heal.
Inf. 22(3), 678–685 (2018)
6. Souza, W.T.D., Kavitha, R.: Human activity recognition using accelerometer and gyroscope
sensors. Int. J. Eng. Technol. 9(2), 1171–1179 (K.R)
7. Ignatov, A.: Real-time human activity recognition from accelerometer data using Convolutional
Neural Networks. Appl. Soft Comput. J. 62, 915–922 (2018)
8. Lecun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521(7553), 436–444 (2015)
9. Krizhevsky A., et al.:ImageNet classification with deep convolutional neural networks alex. In:
Proceedings of 31st International Conference Machine Learning, vol. 3, no. 9, p. 322 (2012)
10. Wang, J., Chen, Y., Hao, S., Peng, X., Hu, L.: Deep learning for sensor-based activity
recognition: A survey. Pattern Recognit. Lett. 119, 3–11 (2019)
11. Zeng, G., He, Y., Yu, Z., Yang, X., Yang, R., Zhang, L.: InceptionNet/GoogLeNet—Going
deeper with convolutions. Cvpr 91(8), 2322–2330 (2016)
12. Rueda, F.M., Grzeszick, R., Fink, G.A., Feldhorst, S., Hompel, M.T.: Convolutional neural
networks for human activity recognition using body-worn sensors. Informatics 5(2), 381–388
(2018)
13. Ha, s., Yun, J.M., Choi, S.: Multi-modal convolutional neural networks for activity recognition.
In: Proceedings of 2015 IEEE International Conference Systems, Man and Cybernetics. SMC
2015, pp. 3017–3022 (2016)
14. Lyu, L., He, X., Law, Y.W., Palaniswami, M.: Privacy-preserving collaborative deep learning
with application to human activity recognition. In: International Conference on Information
and Knowledge Management, Proceedings, vol. Part F1318, pp. 1219–1228 (2017)
15. Chen, Y., Shen, C.: Performance analysis of smartphone-sensor behavior for human activity
recognition. IEEE Access 5(c), 3095–3110 (2017)
16. Ehatisham-ul-Haq, M., Awais Azam, M., Naeem, U., Amin, Y., Loo, J.: Continuous authenti-
cation of smartphone users based on activity pattern recognition using passive mobile sensing.
J. Netw. Comput. Appl. 109(March), 24–35 (2018)
17. Shoaib, M., Bosch, S., Scholten, H., Havinga, P.J.M., Incel, O.D.: Towards detection of bad
habits by fusing smartphone and smartwatch sensors. In: 2015 IEEE International Conference
Pervasive Computing and Communication Work. PerCom Work, pp. 591–596 (2015)
18. Mario, M.O.: Human activity recognition based on single sensor square HV acceleration images
and convolutional neural networks. IEEE Sens. J. 19(4), 1487–1498 (2019)
19. Munoz-Organero M.: Outlier detection in wearable sensor data for human activity recognition
(HAR) based on DRNNs. IEEE Access 7 74422–74436 (2019)
20. Al-Fatlawi, A.H., Fatlawi, H.K., Ling, S.H.: Recognition physical activities with optimal
number of wearable sensors using data mining algorithms and deep belief network. In: Proceed-
ings Annual International Conference of the IEEE Engineering in Medical and Biology Society
EMBS, pp. 2871–2874 (2017)
21. Damas, M., Rojas, I., Amft, O.: A benchmark dataset to evaluate sensor displacement in activity
recognition. A benchmark dataset to evaluate sensor displacement in activity recognition. no.
May (2014)
10 Human Activity Recognition by Deep Convolution Neural Networks … 133

22. Nabian, M.: A comparative study on machine learning classification models for activity
recognition. J. Inf. Technol. Softw. Eng. 07(04), 4–8 (2017)
23. Shoaib, M., Scholten, H., Havinga, P.J.M.: Towards physical activity recognition using smart-
phone sensors. In: Proceedings of the IEEE 10th International Conference on Ubiquitous Intel-
ligence & Computing. UIC 2013 IEEE 10th International Conference on Autonomic & Trusted
Computing. ATC 2013, pp. 80–87 (2013)
24. Banos, O., Garcia, R., Holgado-terriza, J.A., Damas, M.: In: Ambient Assisted Living and
Daily Activities, vol. 8868 (2014)
25. Shoaib, M., Bosch, S., Durmaz Incel, O., Scholten, H., Havinga, P.J.M.: Fusion of smartphone
motion sensors for physical activity recognition, vol. 14, no. 6 (2014)
26. Sztyler, T., Stuckenschmidt, H.: On-body localization of wearable devices: an investigation
of position-aware activity recognition. In: 2016 IEEE International Conference on Pervasive
Computing and Communications. Commun PerCom 2016 (2016)
27. Shoaib, M., P. S. Group: Human activity recognition using heterogeneous sensors. In: 2013
ACM International Joint Conference on Pervasive and Ubiquitous (UbiComp 2013), pp. 1–4
(2013).
28. Gad, A.F.: Practical computer vision applications using deep learning with CNNs. In: Moons,
B., Bankman, D., Verhelst, M. (eds.). Embedded Deep Learning (2018)
29. Wang, S.H., Muhammad, K., Hong, J., Sangaiah, A.K., Zhang, Y.D.: Alcoholism identification
via convolutional neural network based on parametric ReLU, dropout, and batch normalization.
Neural Comput. Appl. 0 (2018)
30. Ioffe, S., Szegedy, C.: Batch normalization: accelerating deep network training by reducing
internal covariate shift. In: 32nd International Conference on Machine Learning ICML 2015,
vol. 1, pp. 448–456 (2015)
31. Agarap, A.F.: Deep Learning using Rectified Linear Units (ReLU), no. 1, pp. 2–8 (2018)
32. LeCun, Y., Kavukcuoglu, K., Farabet, C.: Convolutional Networks in Vision, pp. 253–256
(2010)
33. Uddin, M.Z., Hassan, M.M.: Activity recognition for cognitive assistance using body sensors
data and deep convolutional neural network. IEEE Sens. J. c, 1 (2018)
34. Vargas: Face recognition using principal component analysis, OpenCV and EmguCV. Int. J.
Adv. Res. Comput. Eng. Technol. 1(9) 135–139 (2012)
35. Smith, L.: A tutorial on principal components analysis. Commun. Stat. - Theory Methods 17(9),
3157–3175 (2014)
36. Jolliffe, I.T., Cadima, J.: Principal component analysis : a review and recent developments
subject areas : author for correspondence. Philos. Trans. R. Soc. A 374 20150202 (2016)
37. Baker, K.: Singular value decomposition tutorial, p. 24. The Ohio State University (2005)
38. Zhang, Y., Xu, G.: Singular value decomposition. Encycl. Database Syst. 1–3 (2016)
39. Cohen, I., Tian, Q., Zhou, X.S., Huang, T.S.: Feature selection using principal feature analysis.
In: Proceeding of IEEE International Conference of Image Processing, pp. 301–304 (2002)
40. Olah, C.: Understanding LSTM Networks. Google Research. https://research.google/pubs/pub
45500/. Accessed 26 May 2020 (2020)
41. Aljarrah A.A., Ali, A.H.: Human activity recognition using PCA and BiLSTM recurrent neural
networks. In: 2019 2nd International Conference on Engineering Technology and its Appli-
cations (IICETA), Al-Najef, Iraq, pp. 156–160 (2019). https://doi.org/10.1109/IICETA47481.
2019.9012979
Chapter 11
Development of a New 6 DOFs Welding
Robotic System for a Specialized
Application

Truong Trong Toai, Duc-Hoang Chu, and Chu Anh My

Abstract This paper reports and summarizes the main research results of a robot
design and implementation project. The project is aimed to design and implement a
novel robotic system which has ability of replacing workers to weld complex parts
mounted on a fixture in series. In the paper, the overall and detailed design of the
robot are presented. Also, some main research results relevant to the implementation
and pilot applications of the robot are shown. It was demonstrated that the robot
prototype was effectively implemented and tested with the use of TIG/MIG/MAG
welding methods. The use of the robot in a welding cell reduces the production
cost, improves the product quality, and optimizes the manpower used in the welding
process.

Keywords Robot arm design · Industrial automation · Welding system

11.1 Introduction

A robot is a controlled mechanical structure that performs automated operations,

ensuring a function on behalf of a human. In the world today there are many manu-
facturers of robots such as ABB, Fanuc, Kuka, Staubli, Motoman, Nachi, Kawasaki,
… Sales of industrial robots globally increased by 27% in 2014, figures from the new
report [1]. Most of the International Federation of Robotics (IFR) said. According to
IFR, in 2014 about 225,000 robots were sold worldwide, with nearly two thirds of
them being sold in Asia. In general, what we think about industrial robots are their

T. T. Toai (B)
3C Manufacturing Machinery Ltd., Ho Chi Minh, Vietnam
e-mail: [email protected]
D.-H. Chu
National Technology Innovation Fund, Hanoi, Vietnam
e-mail: [email protected]
C. A. My
Le Quy Don Technical University, Hanoi, Vietnam
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 135

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_11
136 T. T. Toai et al.

size, complexity, and high price. Even though many people are enthused about the
industrial robot, until now, it has been difficult to own a quality one at home [2].
Today that is not correct with 6-axis industrial robot arm. This robot arm is suitable
for home, school, and other professional purposes.
In addition, the face of a milling operation is been changing by industrial robots.
Up to the present, milling has been completed with special CNC machines and
milling. The robots provide the below advantages over the common milling methods:
Flexibility: In general, the 6-axis articulated robot provides more flexible moving
actions than an ordinary milling machine. From multiple angles, the robot has ability
to mill a complex part; Affordability: Milling is highly economical by using a robotic
arm. Unlike milling machines, robots have ability to reassign for performing other
tasks in a shop—material handling, arc welding, etc. The robots have ability to save
valuable time, improve throughput, and increase production. Moreover, a robotic
arm has possibility to handle more of the milling tasks without human intercession.
Throughput: the robotic arm can rise overall throughput of milling tasks. Besides,
the robots are more accurate and consistent than the milling machines. A robotic arm
can mill faster than a milling machine with less time spent repositioning and fewer
mistakes; Right Touch: We employ many materials for molds and prototyping such
as foam, soft—clay, and REN board. In these cases, a robotic arm is suitable for all
types of mediums.
The use of robots in the workpiece level helps to improve many technical issues
such as increasing productivity, improving the quality and uniformity of products,
stabilizing production, reducing the risk of labor accidents, however It is necessary to
have an appropriate and economical solution for domestic SMEs that can be applied
to their production process. In a mechanical processing workshop at Truong Hai
Auto Company alone, the number of stamping machines is 150 and employs one
worker to operate one. The stamping products are materials and accessories in the
automotive industry, with specific types of product types, limited quantities in each
production cycle, so it is only suitable for the model of loose cast ingots.
In recent decades, as compared with other industrial robots, the welding and
cutting robots have been playing the most important role in several industries [3–6].
We have been using the welding robots for many welding and cutting systems in
current industrial manufacturing over the world. The main welding methods that
have been integrated with a welding robot include (1) TIG (tungsten inert gas),
(2) MIG/MAG (gas metal arc welding), (3) PAW (plasma arc welding), (4) SAW
(submerged-arc welding) and (5) LBW (laser beam welding).

11.1.1 TIG Welding Method

GTAW stands for Gas tungsten arc welding. Nowadays, GTAW is also called TIG
(tungsten inert gas) welding. This is an arc welding process (AWP) using a non-
consumable tungsten electrode for producing the weld. The electrode and the weld
areas are protected from atmospheric contaminations such as oxidation by an inert
11 Development of a New 6 DOFs Welding Robotic System … 137

Fig. 11.1 Tungsten inert gas

(TIG) welding

shielding gas (helium or argon). In general, a filler metal is employed, though some
welds do not require it. When we use helium, this weld is recognized as heliarc
welding. The electrical energy is provided by a constant-current welding power
supply. The electrical energy is controlled across the arc via a column of metal
vapors and highly ionized gas called a plasma [7] (Fig. 11.1).

11.1.2 MIG/MAG Welding Method

GMAW stands for Gas metal arc welding. Sometimes, GMAW is called by its
subtypes metal active gas (MAG) welding or metal inert gas (MIG) welding. This is
a AWP that an electric arc forms between the workpiece metal(s) and a consumable
MIG wire electrode in order to melt and join them together. To shield the process
for avoiding atmospheric contamination, a shielding gas is charged via the welding
gun along with the wire electrode. This process can be run automatically or semi-
automatically. We often employ a constant voltage such as the direct current power
source with GMAW. However, the constant current systems or replacing current
systems are also employed. In GMAW, globular, short-circuiting, pulsed-spray, and
spray are four main methods of metal transfer. Each method has some advantages
and disadvantages properties. In addition, GMAW is used as a low-cost method for
printing 3D metal objects. For the past few years, many researchers and companies
have been developed open-source 3D printers for using GMAW [3]. Such products or
components fabricated from aluminum contend with more commonly manufactured
products or components on mechanical strength. By forming a not good weld on
the first layer, GMAW 3D printed parts can be eliminated from the substrate using a
hammer.
138 T. T. Toai et al.

11.1.3 Plasma Arc Welding (PAW)

PAW is another AWP that is similar to TIG one. PAW is formed between the work-
piece and a pointed tungsten electrode. Nevertheless, by navigating the electrode
within the body of the torch, the plasma arc can be detached from the shielding gas
envelope. After that, we force plasma via a fine-bore copper nozzle that constricts
the arc. By varying the rate of plasma gas flow and the bore diameter, we can produce
three operating modes. We employ microplasma welding (0.1–15 A) to weld thin
sheets (0.1 mm thickness), wire, and mesh sections. In addition, the needle-like, stiff
arc reduces arc distortion and wander. Medium current welding is from 15 to 200 A.
This welding is employed in the melt mode as a replacement option for conven-
tional TIG. PAW has three advantages. The first one is greater tolerance to surface
contamination integrating coatings. Here, the electrode is within the body of the
torch. The second one is deeper penetration (from higher plasma gas flow). The third
one is good tolerance to variations in the electrode to workpiece distance, which has
a small change in heat input.

11.1.4 Submerged-Arc Welding (SAW)

SAW stands for submerged-arc welding. SAW is a normal AWP that implies the
formation of an arc between the workpiece and a continuously fed electrode. A
blanket of powdered flux creates a slag and a protective gas shield (and may also
be employed to add alloying components to the weld pool) for protecting the weld
zone.

11.1.5 Laser Beam Welding (LBW)

LBW is another welding technique for joining pieces of thermoplastics or metal via
a laser beam. The laser beam produces a centered heat source for deep and narrow
welds. In addition, the centered heat source also has high welding rates. This process
is often employed in high volume systems using automation. One good example is
the automotive industry. LBW relies on penetration mode welding or keyhole. In
addition, LBW is a flexible process. Besides, LBW is capable of welding HSLA
steels, stainless steel carbon steels, titanium, and aluminum. When we weld high-
carbon steels, cracking is a concern because of high cooling rates. The weld quality
of LBW is high and it is similar to the quality of the welding using an electron
beam. Moreover, the speed of welding is changed corresponding to the amount of
power supplied. The speed is also dependent on the thickness and type of workpieces.
High power capability is contained in the gas lasers. The gas lasers are suitable for
11 Development of a New 6 DOFs Welding Robotic System … 139

high volume applications. The LBW is particularly dominant [7] in the automotive
industry.

11.1.6 Industrial Robots

In our previous studies [5, 6], we presented the rough designs of the robotic systems.
These systems concentrate on industrial manipulators for performing several tasks
including painting, palletizing, and welding. In the literature, we divide the researches
into two main groups concentrate on the design and integration of the robotic systems
to produce cells. The first group focuses on the general purpose of robots’ applica-
tions which are called commercial robots. On the other hand, the second category
investigates about designing and implementing of the robotic systems.
We selected and used the commercial robots for performing the required tasks
in the first approach. Related to this issue, several studies analyzed cost-effective
solutions when using commercial robots in real production lines. Several authors
proposed the optimal robots selection for robotic cells. For the robot-human inter-
action, the study [6] presented that the automotive industry and this industry’s
supply chains are dominant clients of the robotic systems because of their ability
for improving quality and repeatability such as assembly or inspection. In these
applications, the human–robot collaboration technique for a welding or assembly
cell is quite important where the robot replaces passive and active behaviors during
assembly. Therefore, we can reduce the human workload and decrease the risk of
strain injuries.
In the second approach, many researchers try to design and implement many
robot models for a large number of applications. In order to handle the workpiece
for a random forging press cell, My et al. [6, 8] investigated the designing and
implementating of a new robot. We design a new robotic mechanism with two main
features. The first feature is the addition of the parallel links among the serial links.
Next, the second feature is the hydraulic actuators’ applications for controlling robot’s
joints. Other aspects related to the robot design and analysis as well as the integration
of an industrial robot in manufacturing systems were presented in [5, 6, 8–16].
Notice that, through a variety of the general purpose robots employed for
producing systems. Moreover, there are many studies related to the optimization
of the robotic cell performance. These studies show the advantages of the commer-
cial industrial and the general purpose robots. The robots can best implement a group
of given tasks but they cannot guarantee the optimal work execution. My et al. [6]
remarked that a considerable gap is still existed between the requirements rising from
specific applications and the capabilities of the commercial robotic systems. This gap
has been motivating the developments of the particular robotic systems for guaran-
teeing the work completed effectively. Besides, current advances in robotic area are
also inspired for designing and implementating the particular robots for satisfying
the increasing request in industrial manufacturing. Relied on the previous results of
the manipulators’ applications and the division of required tasks, novel concepts of
140 T. T. Toai et al.

robotic system are continuously investigated. Many innovative robotic architectures

and designs have been providing. In this study, we propose one of the innovative
and novel robotic systems. We create a special welding robot arm for a specialized
application in Vietnam industry.

11.2 Design of a 6DOFs Welding Robot

11.2.1 Functional Requirements

Currently industrial robots are applied in many fields including welding, to replace
workers, improve product quality, increase productivity, as well as improve working
conditions for workers in toxic environments. harm. Welding robots are commonly
used in industrial manufacturing industries such as chassis welding robots, bed-
bed welding robots, scaffold-type welding robots, tables and chairs welding robots,
automobile welding robots, motorcycles, welding racks. industry, … when mass-
producing products in large quantities, welding robots are the most productive
method.
For the welding process, the application of robots in the manufacturing process
has some outstanding advantages such as: High automation; Increasing productivity
and economic efficiency; Stable weld size and quality; Stress and deformation after
small welding; Perform welding joints with high accuracy; Working in an unfavorable
environment that replaces people [4–6, 8] (Fig. 11.2).

11.2.2 The Overall Structure of the Welding System

Welding robot is the main part of an automated welding system using robots. The
robot is responsible for performing the welding head movement along the welding
line to create a weld. To achieve the goal of this study, we designed an experimental
device in order to gain the dynamical data which is supplied by the robot, workpiece,
or tool systems. Several experiments were implemented on a 6-arm robot center as
shown in Fig. 11.3 and Fig. 11.4. We used 6 kW of power for the spindle motor.
The maximum rotation speed was 20,000 rpm. We recorded the necessary data of
cutting forces signals and vibrations at the same time with rotational speed during
the experiment. On the robot’s head, we fixed a 3D PCB piezoelectric accelerometer.
Besides, a laser sensor tachometer is set up on the spindle holder. We obtained the
rotation speed via this laser sensor. For the three-linear directions of the robot in the
X-axis, Y-axis, and Z-axis, we make them following the 3D axis of the accelerometer.
Finally, for cutting forces measurement, we positioned a 6D forces dynamometer at
the end of the robot including (1) Welding source and welding parameter adjustment
unit; (2) Robot controller; (3) Welding robot; (4) Feeder wire and welding wire; (5)
11 Development of a New 6 DOFs Welding Robotic System … 141

Fig. 11.2. 6-Arm robot block diagram

Fig. 11.3. 6-arm (degrees)

of freedom in robot design

Welding head; (6) Detailed attachment parts; (7) Welding process monitoring unit;
(8) Toxic exhaust device; (9) Protection barrier.
142 T. T. Toai et al.

Fig. 11.4. 3D drawing of

the robot 6-arm of 3C
Machinery Manufacturing
Co., Ltd

11.2.3 The Welding Arm

For any given point in space to determine, there must be six degrees of freedom.
Therefore, in order to locate all points in space and perform motion in a flexible way,
the robot used in the automatic welding system is usually a six-degrees robot (Six
rotary joints of type 5) [6].
In which, three degrees of freedom are used to determine position; The remaining
three steps are used to determine the working direction of the tool. Welding
robots with less than six degrees of freedom are usually specialized-oriented robots
optimized for application groups to reduce robotic costs.

11.2.4 The Workspace of the Robot

The reach or workspace is the most likely reach of the robot or the workspace volume
of the robot. The working range affects the ability of the robot to work at the size
limit of the part. If the robot has a bigger range, the better, but the price is high (Table
11.1).
The load of the robot can carry additional mass without affecting the workability
of the robot. It consists of two components: Load: The weight of the tool mounted
on the head of the robot. It is usually valid for 3–16 kg; Extra load: The additional
load on the robot arm. Often it is the wire feeder in MIG/MAG welding [17].
The process of checking the ability to operate, calibrate robots is defined in ISO
9283: 1998. This standard establishes various operating criteria for industrial robots
and recommends test cycles to control the value of special parameters. One of the
goals of this standard is to provide technical information to help users choose the
11 Development of a New 6 DOFs Welding Robotic System … 143

Table 11.1 Robot’s working

No. Working area
area
Working range affects Unit Value
1 Stitch length 1 (a2) mm 300
2 Stitch length 2 (a3) mm 350
3 Starting point of orbit mm <|450|
4 Orbital endpoint mm <|550|
5 Orbit length mm 1550
6 Motion time s 4
7 Acceleration characteristics Shape Sine

robot that best suits their purpose. The most important, and also the most commonly
used, criterion is AP-Accuracy of Pose and RP-Repeatability of Pose.
To assess the average of this test, the ISO9283 standard requires that the test be
carried out at a number of points in the test plane where the data (range of actual
locations achieved) is mixed. Since the direction of the ellipse is different at each test
site, the mix of the regions can be approximately spherical. The distance between
the desired position and the barycenter represents the systematic error and is clearly
shown in the error of the dynamic model. The distribution of the actual positions may
be caused by noise and friction. The absolute accuracy and repeatability accuracy
are as follows:
The actual location coordinates are reached: X ai , Yai , Z ai ; The required location
coordinates: X c , Yc , Z c ; Location coordinates actually achieved:

1N 1N 1N
X= X ai , Y = Yai , Z = Z ai (11.1)
N i=1 N i=1 N i=1

 2  2  2
Ii = X ai − X + Yai − Y + Z ai − Z ,
  2
1 N
1N N i=1 i
I −I
I = Ii , Si = (11.2)
N i=1 N −1

Pose accuracy (AP) and Pose Repeatability (RP):


 2  2  2
AP = X − Xc + Y − Yc + Z − Z c ; P = I + 3Si (11.3)

N: Number of measurement points achieved; I: Distance between the attached pose

and the barycentre of the attained poses; S: Standard deviation; AP: Pose accuracy;
RP: Pose Repeatability.
144 T. T. Toai et al.

11.2.5 The Control Software of the Robot

Software used on robots is usually specialized software written on Windows or Linux

platform, the common feature of these software packages is real-time software (real
time) [18]. The time delay in these software packages is very small to ensure the
synchronized movement of necessary joints, giving high control quality, in addition
to the fastest response to what the operator controls to avoid collisions [19]. Touching
may occur during controls [20].
For the model of the robot feeding the press machine, the impact dynamics
problem of the joints is very small, so the control can be attributed to independent
joint control, the basic PID controllers meet get these systems. The data is entered into
the controller via the console in combination with the graphic signal processing unit
on the computer. At the central processing unit, the data will be processed according
to preprogrammed programs to produce a sequence of data about the robot’s motion
parameters (Fig. 11.5).
This data series will be the input of the digital signal processor. Here, the data will
be processed in combination with the current position and speed data of each axis to
produce the desired position and speed of each axis. This result will be sent to each
axis controller, this controller combined with the central power unit will control each
axis to give the desired result.
The applications of PID controller have been being employed from small to high
technology industries. The main objective of this paper is to develop a position
controller of a robot arm by selecting the PID parameters employing a genetic algo-
rithm. We consider the model of the robot arm as a third order system. In addition, we
also compare two kinds of tuning methods of the parameter for PID controller. Under
PID controller, the step response tuned by GA is quite better in terms of minimizing

Fig. 11.5 DCS3T-27: DC

Servo motor driver 500 W
programming interface
11 Development of a New 6 DOFs Welding Robotic System … 145

Fig. 11.6 PID controller in two cases with and without integral components

the max overshoot, the increasing time and the processing time. Finally, the mean
square error associated with PID controller is smaller than that with conventional
approach (Fig. 11.6).
The objective of this work is to prove that robots can perform many dynamic
behaviors. From the simulation results, calculate the maximum torque and driving
force of the joint angles, then calculate the appropriate power source and transmission
power. The calculation results are quite suitable for systems with similar models. We
position-control the DC-Motor using a PID controller for locating the link at the
required angle; we measure the feedback of the angle using rotary-encoders. This
project introduces intelligence to industries to reduce human errors and increase the
quality and mass production of industries.

11.3 Implementation and Testing for the Robot

11.3.1 Implementation of the Robot

The use of a multi-function robot for a task like the aforementioned press is not cost
effective because there is a redundancy of many functions. Typically, robot manufac-
turers often tend to create a standardized, multi-function robot that can perform any
required task. For each requirement, the user chooses an extra structure (for welding,
for gripping or assembling). This forces enterprises to invest large amounts of tens
of thousands of dollars for a specific automation task. In a time of high competi-
tion, which requires fast production speed, high productivity, and high labor safety,
the creation of robots and manuals is not too complicated and highly competitive.
Human replacement is an urgent need. It is also a land for designers and manufac-
turers in developing countries—creating new products, manufacturing specialized
robots for specific application purposes. Thus, the robot structure is more simple, the
accuracy depends on the task performed. Therefore, the cost of robots will decrease
significantly (Fig. 11.7).
146 T. T. Toai et al.

Fig. 11.7 Operate the 6-arm

robot of 3C Machinery
Manufacturing Co., Ltd in
reality

Some lines of CNC plasma cutting machines are designed to ensure high mobility,
the system can be easily disassembled and moved quickly to new working positions.
For example, the Portable CNC Cutting Machine series of Steel Tailor, China, the
console beam type transmission structure, single-sided drive, using the gear rack
transmission for all 3 axes X, Y and Z.
Usually driven by stepper motors, because the working speed is not high. Fast
travel speed is 6000 mm/min, working speed from 300 to 3000 mm/min. Location
accuracy of ±0.2 mm. This type of structure when the tip of the cutting head reaches
far beyond the shoulder support, the greater the deflection. With the scale of industrial
cutting with high intensity, this model can not satisfy the needs of users. Using
specialized specialized robots can reduce the investment cost for robots by more
than 50%. This is because the robot does not require very high accuracy (up to
±0.1 mm) to use many gear boxes costing thousands of USD/unit. This robot also
does not require high speed, complex trajectories. The use of a specific robot—hand
machine allows a significant reduction in investment in the renewal of production
structures for a multi-machine system, such as stamping, welding, cutting, bending,
turning, CNC milling, etc.
The purpose and mission of the project is to research, design and manufacture
industrial robotic systems (industrial robotic systems) which are high-tech products
with localization rate of over 70%, fast and flexible response. manufacturing automa-
tion, easy installation, quick configuration and reduced costs for small and medium
mechanical enterprises in Vietnam.
11 Development of a New 6 DOFs Welding Robotic System … 147

11.3.2 Accuracy Analysis

The ability to repeat exactly what the robot has learned or the ability to do exactly
the programming line. The smaller the accuracy, the better the ability to meet the
requirements, but the higher the price. In fact, we have to balance accuracy and
cost. Usually, the smaller the robot, the smaller the accuracy and vice versa. This
number varies from 0.06 to 0.1 mm. This accuracy also depends on the technology
capabilities of each company.
The objective of the envelope method which is employed for the robot behaviors
for the milling process is reached by using consistent in a high accuracy transform
named synchronous FFT, frequency-domain processing, resonance band filter of
workpiece and tool, Inverse Fast Fourier Transform (IFFT), and Hilbert transform
[16]. After that, the FFT checks of the envelope guarantee high precision definition
of the milling tool for identifying the amplitude as well as the type of wear and
asymmetry [21]. We assumed that structural impacts usually affect one resonance
mode or several resonance modes of the structure. We also assumed that structural
impacts usually create impulsive vibrations in both periodic and repetitive ways. The
structural impacts are caused by a localized defect. We often place the frequencies
related to the robot-borne vibrations in lower frequency areas than those induced
by the resonance modes. Besides, these frequencies are characterized by an energy
centralization. The energy is concentrated in a relatively narrow band located at a
harmonic of the resonance frequency. For diagnosis purposes, we can separate the
defect-induced vibration features from the interference and background noise by
using the effect of mechanical amplification. In that case, the structural resonances
supply this effect.
In the process of manufacturing and manufacturing holes to install rotary joint
bearings, the transmission error of reducer. Robust simulation, robustness, and accu-
racy of the robot help us evaluate the influence of the design model on the technical
requirements and function of the robot models, thereby making model adjustments.
designed to meet the technical and functional requirements of the task’s robots.
The application potential of the project’s products is huge, the products of the
project can create a new domestic market, especially serving the needs of small and
medium mechanical enterprises in Vietnam.

11.3.3 Testing and Application

Communication protocols have always been a major challenge for the robotics
industry, especially for robots for digital transformation. Communication of compo-
nents of the robot system should be built on the basis of modern standards,
simple connection, convenience, easy configuration and installation. The connection
protocol is not only between the robot and the external peripherals, the components
themselves connected inside the robot also need to communicate and communicate
148 T. T. Toai et al.

with each other. Some common communication protocols commonly used by indus-
trial robots are EtherNet/IP, Modbus-TCP, Device Net, CANOpen, EtherCAT and
Modbus-RTU.
Equipping and integrating communication protocols into task robots should be
considered based on the following basic features and content:
• At least one basic communication standard according to industry standards to
connect with related industrial equipment in the task’s robot.
• Integrate an advanced optional protocol for current applications and complexity
in some cases.
• The transmission distance and transmission speed of the data in the robots of the
task.
• Influence of cross-channel noise (CrossTalk) and electromagnetic interference
(EMI).
• Cost-effective production so that the robot products of the task are suitable for
customers of small and medium enterprises.
• Depends on the overall plan of the central controller.
• Robots are designed in a specialized direction, optimized for application, so the
number of components and equipment is less than a multi-purpose robot of the
firms.
The use of a robotic arm during testing has many of the following outstanding
advantages: Completely replace the role of workers in a specific stage (higher oper-
ating accuracy than workers do, faster execution speed, easy to control and correct
errors); Ability to synchronize control with other control systems, integrated with the
information management system of the factory; More economical than cooperating
with workers to perform some operations in the production line.

11.4 Conclusion

In this paper, we presented the design and implementation of a novel task-oriented

robot. The robot product has been integrated in a real welding cell of a small and
medium-sized enterprises (SME) in Vietnam. It was shown that the robot is capable
of working well with the manufacturing cell so that the use of the manpower is opti-
mized, and the productivity and the welding quality are increased. It was also shown
that a new robot product was effectively designed following the task – oriented design
approach. The success of this research project opened a chance for development of
a new family of robot products that could be used suitably for welding industry in
Vietnam and other developing countries. The use of these robot products will help
SMEs in developing countries to improve their competitiveness in the regional and
global markets.

Acknowledgements This paper was supported by National Technology Innovation Fund (NATIF),
Ministry of Science and Technology (Most), Vietnam, project NATIF.TT.01.DT/2018.
11 Development of a New 6 DOFs Welding Robotic System … 149

References

1. McMorran, D., Chung, D.C.K., Li, J., Muradoglu, M., Liew, O., Ng, T.W.: Adapting a low-
cost selective compliant articulated robotic arm for spillage avoidance. J. Lab. Autom. 21(6),
799–805. https://doi.org/10.1177/2211068216630742. ISSN 2211–0682. PMID 26882923
2. Oña, E.D., Cano-de la Cuerda, R., Sánchez-Herrera, P., Balaguer, C., Jardón, A.: A review of
robotics in neurorehabilitation: towards an automated process for upper limb. J. Healthcare
Eng. 2018, 9758939 (2018). https://doi.org/10.1155/2018/9758939
3. Carvalho, M.C., Eyre, B.D.: A low cost, easy to build, portable, and universal autosampler for
liquids. Methods Oceanogr. 8, 23–32. https://doi.org/10.1016/j.mio.2014.06.001
4. My, C.A., Integration of CAM systems into multi-axes computerized numerical control
machines. In: Second International Conference on Knowledge and Systems Engineering,
pp. 119–124 (2010)
5. My, C.A., Bohez, E.L., Makhanov, S.S., Munlinb, M., Phien, H.N., Tabucanon, M.T.: On 5-axis
freeform surface machining optimization: vector field clustering approach (2005)
6. My, C.A., Le, C.H., Packianather, M.S., Bohez, E.L.: Novel robot arm design and implemen-
tation for hot forging press automation (2019)
7. Chen, G., Gu, X., Bi, J.: Numerical analysis of thermal effect in aluminum alloy by repetition
frequency pulsed laser. Optik Int. J. Light Electron. Opt. 127(20), 10115–10121 (2016). https://
doi.org/10.1016/j.ijleo.2016.08.010
8. My, C.A., Parnichkun, M.: Kinematics performance and structural analysis for the design of a
serial-parallel manipulator transferring a billet for a hot extrusion forging process. Int. J. Adv.
Rob. Syst. 12(12), 186 (2015)
9. My, C.A., Bien, D.X., Le, C.H., Packianather, M.S.: An efficient finite element formulation of
dynamics for a flexible robot with different type of joints (2019)
10. My, C.A.: Mechanical design and dynamics modelling of RoPC robot. In: Proceedings of
International Symposium on Robotics and Mechatronics, Hanoi, Vietnam, pp. 92–96 (2009)
11. My, C.A.: Inverse dynamic of a N-links manipulator mounted on a wheeled mobile robot. In:
2013 International Conference on Control, Automation and Information Sciences (ICCAIS),
pp. 164–170. IEEE (2013)
12. My, C.A., Bohez, E.L.: A novel differential kinematics model to compare the kinematic
performances of 5-axis CNC machines. Int. J. Mech. Sci. 163, 105117 (2019)
13. My, C.A., Cong, N.V., Hong, N.M., Bohez, E.L.: Transformation of CAM data for 5-axis CNC
machine spinner U5–620. Int. J. Mech. Eng. Robot. Res. 9(2) (2020)
14. My, C.A.: Inverse kinematics of a serial-parallel robot used in hot forging process. Vietnam J.
Mech. 38(2), 81–88 (2016)
15. My, C.A., Bien, D.X., Tung, B.H., Hieu, L.C., Cong, N.V., Hieu, T.V.: Inverse kinematic
control algorithm for a welding robot-positioner system to trace a 3D complex curve. In: 2019
International Conference on Advanced Technologies for Communications (ATC), pp. 319–323.
IEEE (2019)
16. My, C.A., Makhanov, S.S., Van, N.A., Duc, V.M.: Modeling and computation of real-time
applied torques and non-holonomic constraint forces/moment, and optimal design of wheels
for an autonomous security robot tracking a moving target. Math. Comput. Simul. 170, 300–315
(2020)
17. Mick, S., Lapeyre, M., Rouanet, P., Halgand, C., Benois-Pineau, J., Paclet, F., Cattaert, D.,
Oudeyer, P.Y., de Rugy, A.: Reachy, a 3D-printed human-like robotic arm as a testbed for
human-robot control strategies. Front. Neurorobot. 13, 65 (2019). https://doi.org/10.3389/fnbot.
2019.00065
18. Anzalone, G.C., Zhang, C., Wijnen, B., Sanders, P.G., Pearce, J.M.: Low-cost open-source
3-D metal printing. IEEE Access 1, 803–810 (2013). https://doi.org/10.1109/ACCESS.2013.
2293018
19. Santello, M., Bianchi, M., Gabiccini, M., Ricciardi, E., Salvietti, G., Prattichizzo, D., Ernst, M.,
Moscatelli, A., Jörntell, H., Kappers, A.M., Kyriakopoulos, K., Albu-Schäffer, A., Castellini,
150 T. T. Toai et al.

C., Bicchi, A.: Hand synergies: Integration of robotics and neuroscience for understanding the
control of biological and artificial hands. Phys. Life Rev. 17, 1–23 (2016). https://doi.org/10.
1016/j.plrev.2016.02.001
20. Kim, Y.J., Nam, H.S., Lee, W.H., Seo, H.G., Leigh, J.H., Oh, B.M., Bang, M.S., Kim, S.: Vision-
aided brain-machine interface training system for robotic arm control and clinical application
on two patients with cervical spinal cord injury. Biomed. Eng. Online 18(1), 14 (2019)
21. Rahman, Md., Khan, A., Ahmed, T., Sajjad, Md.: Design, analysis and implementation of a
robotic arm-the animator. Int. J. Eng. Res. 02, 298 (2013)
Chapter 12
Low Power Reversible Parallel and Serial
Binary Adder/Subtractor

N. Bhuvaneswary, S. Prabu, S. Karthikeyan, R. Kathirvel, and T. Saraswathi

Abstract Each and every day new technology is being established which is minia-
turized and challenging the low power than existing design. All the gates used in the
design is reversible to achieve low power and less complexity. In past few decades
reversible enlarges its applications in modern computing environment. Reversible
gates can be executed using any Boolean function. In a modern computing envi-
ronment adders plays important role for addition process as well as for subtrac-
tion. This paper proposes a modified reversible parallel and serial adder/subtraction
circuit using Dual Key Gate (DKG) and SG gate. The performance of parallel
adder/subtractor circuit design using dual key gate and serial adder/subtractor using
dual key gate with SG gate is simulated and synthesized using Xilinx. The perfor-
mance of this circuit is compared with existing design using Feynman gate and toffoli
gate based on complexity, low power and Garbage input/outputs.

N. Bhuvaneswary (B)
Department of Electronics and Communication Engineering, Kalasalingam Academy of Research
and Education, Krishnankoil, Tamilnadu, India
e-mail: [email protected]
S. Prabu · R. Kathirvel
Department of Electronics and Communication Engineering, Mahendra Institute of Technology,
Namakkal, Tamilnadu, India
e-mail: [email protected]
R. Kathirvel
e-mail: [email protected]
S. Karthikeyan
Department of Electronics and Instrumentation Engineering, M.Kumarasamy College of
Engineering, Karur, Tamilnadu 639113, India
e-mail: [email protected]
T. Saraswathi
Department of Electronics and Communication Engineering, Mahendra Engineering College,
Namakkal, Tamilnadu, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 151

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_12
152 N. Bhuvaneswary et al.

Keywords Low power · Garbage input/output · Reversible logic circuit ·

Parallel/serial adder/subtractor

12.1 Introduction

A reversible gate input is recovered back from output. A Reversible gate can be
generate distinctive output from apiece of input. Reversible gate is a one to one
communication between the input and output patterns. In reversible gates the number
of inputs and outputs are same but in irreversible it is not same except NOT gate.
The circuit designed using logic gates dissipates more heat and lost the information
for each and every bit. But in reversible gate design, the loss of information can be
reduced and recovered back [1]. One of the main advantage of reversible logic is
power reduction. According to Landauer’s [2] principle irreversible logic dissipates
more heat of K * Tln2 for every one bit of information lost, where K denoted as
Boltzmann’s constant and T as absolute temperature. As per Bennett [3] concept the
energy loss can be reduced to zero by reversible logic design. The significant strategy
restraints of reversible logic is as follows, 1. The reversible logic gates has no fan-
out. 2. Nielson and Chuang, Knill et al. [4, 5] the quantum cost of circuit is very
low. 3. The number of garbage output is reduced. 4. The number of unceasing inputs
is low. This paper proposes a parallel and serial adder/subtractor using reversible
logic. This newly proposed design is compared with existing design based on power
consumption, garbage inputs/outputs and complexity. This proposed design is proved
to be better compared to existing.
This paper is structured as follows. The basic reversible logic gates employed with
this work is discussed in Sect. 12.2. In Sect. 12.3 describes the concept of proposed
reversible adder/subtractor. In Sect. 12.4 discussed the proposed reversible parallel
add/sub using dual key gates. The serial adder and subtractor circuit design described
in Sect. 12.5. In Sect. 12.5 the results of this designed are compared with existing.
In Sect. 12.6 the design are simulated and in Sect. 12.7 the work is concluded.

12.2 Realization of Basic Reversible Logic

The basic reversible logic gates employed with this design is Feynman gate, peres
gates, dual key gates and sayem gate. The existing design comprised with Feynman
and peres gate [6] but in proposed design uses dual key gate and sayem gate.
Figure 12.1 displays a 2 * 2 Feynman gate [7]. The inputs and outputs are corre-
sponds to each other. It is also called as Complementary NOT gate. By using Feynman
gate the fan out can be minimized and the quantum cost is 1.
Figure 12.2 displays a 3 * 3 Peres gate [7]. The inputs and outputs are corresponds
to each other. Using this gate many Boolean functions can be realized. The existing
work design a adder circuit with Feynman gate.
12 Low Power Reversible Parallel and Serial Binary Adder/Subtractor 153

Fig. 12.1 Feynman gate

Fig. 12.2 Peres gate

Figure 12.3 displays a 4 * 4 dual key gate. This gate can be used as adder/subtractor
circuit. When input A is 0 then the circuit trail plays an addition process and when
the input A is 1 then the circuit trail plays a subtraction process. Boolean functions
like NOT, AND, OR, NOR etc. can be realized using reversible DKG gate. The 4
input DKG gate has quantum cost of 5 (Fig. 12.4).
Figure 12.5 represents a 4 * 4 sayem gate. This gate used to design a serial adder
with the combination of dual key gate.

Fig. 12.3 Reversible dual

key gate

Fig. 12.4 Dual key gate

quantum logic circuit
154 N. Bhuvaneswary et al.

Fig. 12.5 Sayem gate (SG)

Fig. 12.6 Full adder using

DKG

Fig. 12.7 Full subtractor

using DKG

Many sequential and combinational circuits can be realized using SG gate [8, 9].
It has quantum cost as 6.

12.3 Design of Reversible Parallel Adder/subtractor

1. Adder logic realized using DKG gate

When the first input of dual key gate is zero then the circuit perform addition
operation. In Fig. 12.6 4 * 4 Dual key gate is used to appliance a full adder. The
four input correspondence to four output. The output R and S is carry and sum
respectively and remaining output acts as garbage output which is given as input
to the parallel adder.
2. Subtractor logic realized using DKG gate
When the first input of dual key gate is one then the circuit perform subtraction
operation. In Fig. 12.7 4 * 4 Dual key gate is used to appliance a full subtractor.
The four input correspondence to four output. The output R and S is borrow
and difference respectively and remaining output acts as garbage output which
is given as input to the parallel subtractor circuit.
3. Design of reversible four bit parallel adder
12 Low Power Reversible Parallel and Serial Binary Adder/Subtractor 155

Fig. 12.8 Reversible four

bit parallel adder/subtractor.
a Reversible four bit parallel
adder when input PS = 0. b
Reversible four bit parallel
subtractor when input PS = 1

The proposed parallel adder can be realized using DKG Full adder. In the
Fig. 12.8a the internal input pin A, B, C and D is mapped with external input
pins PS, X, Y and Cn correspondingly. Similarly the internal output pins R and
S is mapped with external output pin Sn and Cn. The remaining output P and
Q is garbage output. When the input PS is zero the addition operation performs
and generates S3 and Cout. When input PS = 0 then the circuit realized as
parallel adder. Enable the input pin PS to zero then it adds the remaining input
and produce sum and carry for each adder then the first adder output bit carry
C1 is given as third input to second adder and it generates sum and carry S1 and
C2. This process rotates till last adder and the final sum and carry is generated
parallelly.

4. Design of reversible four bit parallel subtractor

The proposed parallel subtractor can be realized using DKG Full subtractor. In
the Fig. 12.8b the internal input pin A, B, C and D is mapped with external input
pins PS, X, Y and Cn correspondingly. Similarly the internal output pins R and
S is mapped with external output pin Bn and Dn respectively. The remaining
output P and Q is garbage output. When the input PS is one then the design
performs subtraction operation and generates D3 and Bout. When input PS =
1 then the circuit realized as parallel subtractor. Enable the input pin PS to one
then it subtracts the remaining input and produce borrow and difference for each
full subtractor then the first full subtractor output bit borrow B1 is given as third
input to second full subtractor and it generates borrow and difference B2 and D1.
This process rotates till last full subtractor and the final borrow and difference is
generated parallelly.

12.4 Design of Reversible Serial Adder/Subtractor

1. Serial adder/subtractor
Serial adder/subtractor can be designed using DKG [10] gate and SG gate [9].
The DKG gate performances by way of full adder/subtractor and the Sayem gate
performances by way of D flipflop (Fig. 12.9).
2. Design of Reversible serial adder
The proposed reversible serial adder can be realized using DKG and SG gates.
The SG gate is realized as D flip flop which send the input serially and the DKG
156 N. Bhuvaneswary et al.

Fig. 12.9 Reversible serial adder/subtractor. a Reversible serial adder (ctrl = 0). b Reversible serial
subtractor (ctrl = 1)

reversible full adder is enabled when control input ctrl = 0. Then adds the binary
inputs received. The number of inputs and outputs are correspondence to each
other [11]. When control input is zero the Dual key gate changed its mode to
reversible full adder and input E in SG gate is one to enable the SG gate as D
flip-flop which sends the input as feedback to DKG gate and then it starts serial
addition process. Then the sum bit S is generated in DKG full adder and carry
bit Cout is generated in SG flip-flop. Then the carry bit is given as input to next
full adder to add serially. Reversible Full adder gate [10] is a 4 * 4 reversible gate
that acts as a full adder once ctrl = 0 and E = 1 respectively. In serial adder the
number of constant input is reduced to three and the garbage output reduced to
four.

3. Design of reversible serial subtractor

The proposed reversible serial subtractor can be realized using DKG and SG
gates. The SG gate is realized as D flip flop which send the input serially and
the DKG reversible full subtractor is enabled when control input ctrl = 0. Then
12 Low Power Reversible Parallel and Serial Binary Adder/Subtractor 157

subtracts the binary inputs received. The number of inputs and outputs are corre-
spondence to each other. When control input is one the Dual key gate changed
its mode to reversible full subtractor and input E in SG gate is one to enable
the SG gate as D flipflop which sends the input as feedback to DKG gate and
then it starts serial subtraction process. Serial Subtraction is usually obtained
by 2 s complement addition. To obtain serial subtraction, first NOT the binary
number and then add the 1 to the most significant bit. Then the difference bit D is
generated in DKG full subtractor and borrow bit B is generated in SG flip-flop.
Then the borrow bit is given as input to next full subtractor to subtract serially.
Reversible Full subtractor gate [10] may be a 4 * 4 reversible gate that acts as
a full subtractor once ctrl = 1 and E = 1 respectively. In serial subtractor the
number of constant input is reduced to three and the garbage output reduced to
four.

12.5 Results

See Table 12.1.

12.6 Simulation Results

The proposed design performance is analyzed using Xilinx. The simulation results
for Parallel adder/subtractor and serial adder/subtractor are shown in the Fig. 12.10,
12.11, 12.12, and 12.13 respectively. In Fig. 12.14 the power consumption is
compared with existing design using Feynman gate and proposed design using DKG
gate. From the comparison Table 12.1 it is clear that existing design consumes more
power 2.722 W than the proposed design.

Fig. 12.10 Simulation result of parallel adder

Fig. 12.11 Simulation result of parallel subtractor

158 N. Bhuvaneswary et al.

Fig. 12.12 Simulation result of serial adder

Fig. 12.13 Simulation result of serial subtractor

Fig. 12.14 Power consumption graph comparison between existing and proposed method
12 Low Power Reversible Parallel and Serial Binary Adder/Subtractor 159

Table 12.1 Represents the comparison between existing and proposed method
Design Number of Type of gate Garbage Garbage Power consumption
gate used outputs inputs (W)
Existing work [12] 8 FG and PG 12 1 2.722
Design of 4 DKG 8 1 1.478
reversible parallel
adder/subtractor
Design of 3 DKG and 8 1 1.372
reversible serial SG
adder/subtractor

12.7 Conclusion

In this paper we proposed a four bit reversible parallel and serial adder/subtractor
using 4 * 4 DKG gates that work individually as a reversible full adder/subtractor.
With that full adder design 4 bit parallel adder/subtractor is designed and serial
adder/subtractor design is implemented with DKG gate as adder/subtractor and SG
gate as D flip-flop. A similar style will work as both adder as well as subtractor. When
compared to existing design the reversible gates usage is reduced and therefore the
garbage outputs created planned style are optimum. The reversible adder/subtractor
mentioned in this paper can be wont to build higher order arithmetic circuits
and complicated designs of quantum computers. The control circuit will be more
elaborately designed in future.

References

1. Schrom, G.: Ultra-Low-Power CMOS Technology, PhD thesis, technschen Universitat Wien
(June, 1998).
2. Landauer, R.: Irreversibility and heat generation in the computing process. IBM J. Res. Dev.
5(3), 183–191 (1961)
3. Bennett, C.H.: Logical reversibility of computation. IBM J. Res. Dev. 17(6), 525–532 (1973)
4. Nielson, M., Chuang, I.: Quantum Computation and Quantum Information. Cambridge Univ,
Press (2000)
5. Knill, E., Laflamme, R., Milbun, G.J.: A scheme for efficient quantum computation with linear
optics. Nature 409(6816), 46–52 (2001)
6. Fredkin, E., Toffoli, T.: Conservative logic. Int. J. Theor. Phys. 21(3–4), 219–253 (1982)
7. Feynman, R.P.: Quantum Mechanical Computers. Optics News 11(2), 11–20 (1985)
8. Merkle, R.C.: Two types of mechanical reversible logic. Nanotechnology 4(2), 114 (1993)
9. Sayem, A.S.M., Ueda, M.: Optimization of reversible sequential circuits. J. Comput. (2010)
10. Krishnaveni, D., GeethaPriya, M.: Design of an Efficient Reversible 8 x 8 Wallace Tree
Multiplier. Submitted for Review to Circuits and Systems-I Journal, IEEE (2012)
11. Srinivasa Rao, N.: Serial adder using reversible gates. Int. J. Adv. Res. Comput. Commun. Eng.
4(5) (2015)
12. Rangaraju, H.G., Venugopal, U., Muralidhara, K.N. and Raja, K.B.: Low Power Reversible
Parallel Binary Adder/Subtractor. arXiv preprint arXiv:1009.6218 (2010)
Chapter 13
Computational Fluid Dynamics Analysis
of Reduction of Temperature Difference
for a Bundle of Steel Pipes Inside
Annealing Furnace

Lam Hai Dinh, Tan Van Nguyen, and Tu Thien Ngo

Abstract The main objective in the present research is to reduce temperature differ-
ence of the steel pipes in furnace and arranging pertinently a bundle of the inside steel
pipes. Moreover, transient numerical simulation was conducted by using ANSYS
FLUENT commercial software to study temperature distribution of steel pipes inside
the original and new designed annealing furnaces. The new model was designed by
adopting the elbow inlet for generating the swirl flow. The temperature difference
of steel pipes for the original furnace was 55.4 °C. However, the temperature differ-
ences of a bundle of steel pipes for the new model by using downward inlet angle
of 30° was 13 °C. Moreover, temperature difference by modified model had more
uniform than those of by Prieler et al. of 50 °C. Effect of Reynolds number, the hori-
zontal and the vertical steel spacing on the temperature difference among the steel
pipes was also examined. The transient simulation were performed to investigate
steel pipe temperature during annealing process. In short, the results revealed that a
new designed inlet configuration with this kind of geometry of an annealing furnace
could be applied to minimize the temperature difference among steel pipes in steel
annealing process.

Keywords Elbow inlet · Downward jet flow · Annealing furnace · Temperature

difference among steel pipes

L. H. Dinh · T. V. Nguyen
Engineering—Technique Faculty, Thu Dau Mot University, 6 Tran Van On, Phu Hoa, Binh Duong
Province, Vietnam
e-mail: [email protected]
T. V. Nguyen
e-mail: [email protected]
T. T. Ngo (B)
Faculty of Mechanical Engineering, Ho Chi Minh University of Technology (HCMUT), Ho Chi
Minh City, Vietnam
e-mail: [email protected]
Vietnam National University, Ho Chi Minh City, Vietnam

© Springer Nature Switzerland AG 2021 161

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_13
162 L. H. Dinh et al.

13.1 Introduction

Steel annealing is a kind of heat treatment that changes the steel properties to
enhance its ductility and also makes it more workable. The annealing process is
heating the steels arranged within furnace and holding at that temperature for a time
and then cooling at a predetermined rate. Uniform flow distribution and enhance-
ment heat transfer was investigated in recently years [1–8]. Han et al. [9] studied
the temperature distribution non-uniformity of inner grooved copper tubes during
pit furnace annealing. They proposed the methods to improve the annealing tubes
quality. Cheema et al. [10] studied the thermal fluid solid interaction of annealing
process for a steel wire in a tube furnace. The results showed that the increased gas
velocity causes an increase heat transfer due to enhanced convection. To enhance the
convection in high temperature coil annealing furnaces, the rotating cylinder tech-
nique were applied by Fatla et al. [11]. They found that the technique could reduce the
temperature difference within the furnace space and enhance the gas mixing quality.
Hachem et al. [12] investigated heat transfer and turbulent flow inside complex
industrial furnaces. Although the flow behavior and the temperature distribution of
a bundle of steel pipes inside a furnace are important, most of the previous studies
have rarely investigated the flow behavior and temperature distribution of steel pipe
simultaneously within annealing furnace during annealing process. And the arrange-
ment effect of steel pipes inside a furnace on the temperature difference among the
steel pipes have not been also published yet.
In the present study, the flow field and the temperature distribution of a bundle
of steel pipes for the original and new designed furnace model are investigated. In
addition, to achieve the small temperature difference among the steel pipes, the effect
of Reynolds number, the horizontal and vertical spacing were also examined. The
transient simulation was also performed to investigate the temperature behavior of a
bundle of steel pipes during the annealing process.

13.2 Annealing Furnace

13.2.1 The Original Annealing Furnace

The basic model with 2 inlets and 2 ourlets is shown in Fig. 13.1. The furnace has
the height, width and length of 3.5 m, 4 m and 10 m, respectively. The hot air inlet
diameter is 0.35 m and the outlet diameter is 0.5 m. Steel pipes arranged inside the
furnace have the inner and outer diameter and the length of 0.3 m, 0.32 m, and 7 m,
respectively. There are 5 rows and 5 columns of steel pipes arranged in the furnace.
The horizontal and vertical distance of the pipes are 0.5 m as shown in Fig. 13.1b.
13 Computational Fluid Dynamics Analysis of Reduction … 163

Fig. 13.1 The basic model with dimensions: a 3D view; b front-view

13.2.2 The New Designed Annealing Furnace

A new designed annealing model is performed with the elbow inlet for generating the
swirl flow and a downward injection angle as shown in Fig. 13.2. The main objective is
that the inlet flow field generates the well-mixed flow within the annealing furnace to
enhance uniform temperature distribution inside the furnace and reduce temperature
difference of steel pipes bundle than the basic model.

13.3 Mathematical Model and Boundary Conditions

The continuity and momentum equations for flow and heat transfer are expressed as
follows [13]:
Continuity equation:

Fig. 13.2 The modified model: a 3D view; b top view

164 L. H. Dinh et al.

∂ρ
+ ∇ · (ρ −
→
u)=0 (13.1)
∂t

Momentum equation:
 
∂(ρ −
→
u)  −
→ −
→   −
→ −
→  2 −
→ ∼
+ ∇ · ρ u u = −∇ p + ∇ · (μ + μt ) ∇ u + ∇ u T
− ∇· u I
∂t 3
+ ρ fi (13.2)

Energy equation:

∂ ∂ −→ 
(ρ E) + u (ρ E + p) = ∇ · (k∇T ) + Sh (13.3)
∂t ∂ xi
  
∂k ∂ ∂ μt ∂k
+ (ρku i ) = μ+ + G k + G b − ρε (13.4)
∂t ∂ xi ∂x j σk ∂ x j
  
∂ε ∂ ∂ μt ∂ε ε ε2
+ (ρεu i ) = μ+ + C1ε (G k + C3ε G b ) − C2ε ρ
∂t ∂ xi ∂x j σε ∂ x j k k
(13.5)

The transient and steady state numerical simulations were performed by ANSYS
FLUENT 18.2 software. Boundary condition and the method for the original and new
designed model are displayed in Table 13.1. The SIMPLE algorithm was selected
for coupling of pressure and velocity in this simulation. Moreover, the Second-Order
Upwind Scheme was used for the spatial discretization. The convergence criterion
was set up 10–4 for momentum, continuity, turbulence equations, and 10–6 for energy
equation. The independence mesh checking was performed on the grid amount of
4.2, 3.5, and 1.9 million. The deviation of temperature difference among the steel
pipes was 2% between two kind of grid amounts of 4.2 and 3.5 million. Therefore,
to reduce the computing time, the grid amount of 3.5 million was selected in this
study.

Table 13.1 Boundary

Parameters Value
conditions and method
Velocity inlet 0.1–1 m/s
Temperature inlet 1000 K (727 °C)
Pressure outlet 0 Pa
Turbulent model Standard k-ε
Annealing time 20,005 s
13 Computational Fluid Dynamics Analysis of Reduction … 165

Fig. 13.3 Flow field and steel pipes bundle temperature distribution within the basic model

13.4 Results and Discussions

13.4.1 The Flow Field and Steel Pipes Temperature

of the Basic Annealing Furnace

The flow field within the basic furnace model is shown in Fig. 13.3. The hot flow
transfers heat to the steel pipes which are arranged within the annealing furnace.
Convection and radiation heat loss also occur in the furnace walls. Hot flow moves
around the outer and inner of the steel pipes and then exits through the outlets at the
top of the basic furnace. Due to the perpendicular inlet flow, the hot flow impinges
to some pipes which are located at outer positions. Hence, the temperature in this
area of steel pipes is higher than the other zone.

13.4.2 The Flow Field and Steel Pipes Temperature

of the Modified Annealing Furnace

Velocity streamline at the side and bottom views of the modified model is shown in
Fig. 13.4. The model generates the swirl flow within the annealing furnace as shown
clearly in the bottom view of the furnace. This leads to a uniform flow distribution
inside the furnace. Moreover, hot flow could move through every pipe arranged
within the furnace easily to transfer heat to pipes. Therefore, temperature difference
among steel pipes for modified model is expected to reduce much more than that
for the basic model. By adopting elbow angle inlet with the downward jet flow, the
flow temperature distribution could enhance uniform as well as reduce temperature
difference inside the furnace than the results performed by Prieler et al. [14] of 50 °C.
166 L. H. Dinh et al.

Fig. 13.4 Flow field and steel pipes bundle temperature distribution within the modified model;
a side view, b bottom view, and c steel pipes temperature

13.4.3 Effect of Reynolds Number at the Inlet, the Horizontal

and Vertical Steel Spacing on the Steel Temperature
Difference Among Steel Pipes

The temperature difference steel pipes within annealing furnace as a function of

Reynolds number is displayed in Fig. 13.5a. The more increased Reynolds number,
the more increased temperature difference. This is because high speed fluid flows
to impinge on the steel pipes during the annealing process causes high temperature
difference of among the steel pipes. As Reynolds number increases from 2400 to
24,000, the temperature difference increases promptly from 3 to 13 °C.
The temperature difference among the steel pipes as a function of horizontal and
vertical spacing is shown in Fig. 13.5b, c. In this case, the distance between the
inlet and the first pipe as well as the pipe diameter are kept constant. The number
of steel pipes is varied if the horizontal spacing is changed. Higher temperature
difference occurred at smaller spacing and lower temperature difference occurred at
larger spacing. This is due to the fact that larger spacing or smaller number of pipes
leads to enough mixing of the flow inside the furnace, on the other hand, smaller
spacing or larger number of pipes hinders enough mixing and leads to smaller flow
13 Computational Fluid Dynamics Analysis of Reduction … 167

Fig. 13.5 Temperature difference of steel pipes as a function of a Reynolds number, b horizontal
and c vertical spacing to steel pipe diameter ratio

velocity. In this case, the minimum temperature difference among of steel pipes
is around 9 °C at the horizontal spacing, Sh /d of 3.13. Similarly, the trends for the
temperature difference among of steel pipes as a function of vertical spacing is shown
in Fig. 13.5c. The lowest temperature difference of 11 °C occurred at Sv /d ratio value
of 3.13.

13.5 Transient Simulation of Steel Pipes Bundle

Temperature During Annealing Process

Steel pipe temperature as a function of annealing process time is displayed in

Figs. 13.6 and 13.7. Steel temperature goes up rapidly at the first annealing period
because the high temperature difference between the hot flow temperature of 723 °C
168 L. H. Dinh et al.

Fig. 13.6 Steel pipe

temperature as a function of
annealing process time

and the steel pipe temperature of 27 °C causes high heat transfer from the hot inlet
flow to steel pipes. Then, the steel temperature will increase slowly until steady state
is reached. The temperature of steel pipes increases rapidly from 27 to around 275 °C,
almost linearly from the beginning to 8000 s. Then, for the next period from 8000 to
14,000 s, the temperature gradually rises from 27 to 350 °C. Finally, the temperature
increases slowly and then approaches the steady state value of 350 °C, for the period
from 14,000 to 20,005 s. The average temperature rising rate of a bundle of steel
pipes during around six hours is 58 °C/h.

13.6 Total Deformation of Steel Pipes During Annealing

Process

To study the total deformation of a bundle of steel pipes, one side of steel pipes
is supposed as the fixed side and the other side is the free side as displayed in
Fig. 13.8. The mechanical properties of steel pipes are given in Table 13.2. Finally,
the total deformation of steel pipes during annealing process with respect to time was
presented in Fig. 13.9. At the initial time of 600 s, the steel deformation is almost
invariable with the deformation of 3 mm. As the annealing time increases to 10,800 s,
steel deformation is to around 20 mm and finally the maximum deformation for each
steel pipe is around 43 mm at 20,005 s. The ratio of the final pipe length to the
original length in the annealing process is 0.43%.
13 Computational Fluid Dynamics Analysis of Reduction … 169

Fig. 13.7 Temperature distribution contour along steel pipes of three kind of time step: a 605 s;
b 10,800 s; c 20,005 s
170 L. H. Dinh et al.

Fig. 13.8 Constraint

condition of the bundle of
steel pipes

Table 13.2 Mechanical properties of steel pipes

Temperature (°C) Density (g/mm−3 ) Thermal Young’s modulus Poisson’s ratio
expansion (GPa)
coefficient (°C−1 )
0 0.790 1.70e−5 198.5 0.294
100 0.788 1.74e−5 193 0.295
200 0.783 1.80e−5 185 0.301
300 0.779 1.86e−5 176 0.310
400 0.775 1.91e−5 167 0.318
600 0.766 1.96e−5 169 0.326
800 0.756 2.02e−5 151 0.333

13.7 Conclusions

Numerical simulation by using ANSYS FLUENT 18.2 software was performed to

study flow behavior and temperature difference of a bundle of steel pipes inside
the original and new designed annealing furnace model. The new designed model
showed the well mixed uniform flow than the original model due to the swirl flow
and the downward jet flow. The temperature difference among of steel pipes in the
new model at value of 13 °C was also smaller than the conventional basic model at
value of 55.4 °C.
Moreover, by adopting elbow angle inlet with the downward jet flow, the flow
temperature distribution could enhance uniform as well as reduce temperature
difference inside the furnace than the results performed by Prieler et al. [14] of
50 °C.
13 Computational Fluid Dynamics Analysis of Reduction … 171

Fig. 13.9 The total deformation of steel pipes for the three kind of time step: a 606 s; b 10,710 s;
c 20,005 s

The larger increased Reynolds number, the larger increased temperature difference
among the bundle of steel pipes. The temperature difference of steel pipes decreased
as the horizontal and vertical spacing increased. Steel pipe temperature increase
sharply from the initial to 8000 s then increased gradually until steady state is reached.
Present simulation also provides a kind of design map for an arrangement of steel
pipes (horizontal and vertical spacing) in an annealing furnace. For example, the
172 L. H. Dinh et al.

Fig. 13.9 (continued)

pertinent position of the steel pipe arrangement is Sh /d = 3.13 and Sv /d = 1.56 at

Reynolds number of 24,000 to obtain the lowest temperature difference value of 9 °C.

References

1. Ngo, T.T., Go, J., Zhou, T., Nguyen, H.V., Lee, G.S.: Enhancement of exit flow uniformity by
modifying the shape of a gas torch to obtain a uniform temperature distribution on a steel plate
during preheating. Appl. Sci. 8, 1–24 (2018). https://doi.org/10.3390/app8112197
2. Ngo, T.T., Zhou, T., Go, J., Nguyen, H.V., Lee, G.S.: Enhancement of exit flow uniformity by
modifying the shape of a gas torch to obtain a uniform temperature distribution on a steel plate
during preheating. Energies 12, 1–21 (2019). https://doi.org/10.3390/en12050869
3. Ngo, T.T., Nguyen, H.V., Lee, G.S.: Enhancement of mixing and temperature distribution of
steel pipes inside an annealing furnace by adopting a modified inlet structure. Trans. Korean
Soc. Mech. Eng. B 43, 725–732 (2019). https://doi.org/10.3795/KSME-B.2019.43.10.725
4. Phu, N.M., Tu, T.N., Danh, T.L.: Experimental and numerical investigation of transport
phenomena and kinetics for convective shrimp drying. Case Stud. Therm. Eng. 14, 100465
(2019). https://doi.org/10.1016/j.csite.2019.100465
5. Phu, N.M., Tuyen, V., Tu, T.N.: Augmented heat transfer and friction investigations in solar
air heater artificially roughened with metal shavings. J. Mech. Sci. Technol. 33, 3521–3529
(2019). https://doi.org/10.1007/s12206-019-0646-x
6. Tu, T.N., Phu, N.M.: Computational fluid dynamics analysis of the heat transfer and pressure
drop of solar air heater with conic-curve profile ribs. J. Therm. Anal. Calorim. 113, 1–12 (2019)
7. Phu, N.M., Bao, T.T., Hung, H.N., Tu, T.N., Hap, N.V.: Analytical predictions of exergoeco-
nomic performance of a solar air heater with surface roughness of metal waste. J. Therm. Anal.
Calorim. 122 (2020). https://doi.org/10.1007/s10973-020-09787-5
13 Computational Fluid Dynamics Analysis of Reduction … 173

8. Phu, N.M., Hap, N.V.: Influence of inlet water temperature on heat transfer and pressure drop
of dehumidifying air coil using analytical and experimental methods. Case Stud. Therm. Eng.
18, 100581 (2020). https://doi.org/10.1016/j.csite.2019.100581
9. Han, Y., Yu, E., Han, Z.: Study on temperature distribution non-uniformity of inner grooved
copper tubes during pit furnace annealing. Int. J. Heat Mass Transf. 104, 749–758 (2017).
https://doi.org/10.1016/j.ijheatmasstransfer.2016.09.006
10. Cheema, T.A., Ali, H., Park, C.W.: Thermal-FSI based analysis of annealing process for a steel
wire in a tube furnace. Appl. Therm. Eng. 98, 340–351 (2016). https://doi.org/10.1016/j.app
lthermaleng.2015.12.070
11. Fatla, O.M.H., Medina, A.V., Robinson, F., Cihuta, M., Beynon, N.: Development of convection
in high temperature coil annealing furnaces using rotating cylinder technique. Appl. Therm.
Eng. 129, 1392–1402 (2018). https://doi.org/10.1016/j.applthermaleng.2017.10.136
12. Hachem, E., Jannoun, G., Veysset, J., Henri, M., Pierrot, R., Poitrault, I., Massoni, E., Coupez,
T.: Modeling of heat transfer and turbulent flows inside industrial furnaces. Simul. Model.
Pract. Theory 30, 35–53 (2013). https://doi.org/10.1016/j.simpat.2012.07.013
13. ANSYS FLUENT Theory Guide, pp. 133–720. ANSYS Inc., Canonsburg, PA, USA (2013)
14. Prieler, R., Mayrhofer, M., Gaber, C., Gerhardter, H., Schluckner, C., Landfahrer, M., Gruber,
M.E., Schwabegger, G., Hochenauer, C.: CFD-based optimization of a transient heating process
in a natural gas fired furnace using neural networks and genetic algorithms. Appl. Therm. Eng.
138, 217–234 (2018). https://doi.org/10.1016/j.applthermaleng.2018.03.042
Chapter 14
A DCNN Based Real-Time
Authentication System Using Facial
Emotions

A. Praveen Edward James, Mun Hou Kit, and T. Anthony Snow Ritta

Abstract The purpose of this work is to design a real-time authentication system

using Deep Convolutional Neural Networks (DCNN) based on facial emotions that
could be used for security checks in public places. The design of the system is
conducted in five stages. Initially, facial expression images for the seven categories
are captured, preprocessed and loaded into the system, so that there are 10 images for
each of the category namely angry, disgust, sad, fear, neutral, happy and surprise. The
network architecture is then defined by arranging various deep learning layers sequen-
tially. A fully connected layer is then defined for training followed by a SoftMax layer.
The training process is executed followed by classification of a facial features into
their appropriate label or category. When the training data of same person was used
a F1 score of 70.71 was obtained. When the training set from two persons, real
time classification resulted in a F1 score of 85.6. The results reveal that the accu-
racy can be increased if the training dataset was increased. The limitation was that
raw data was used and can be improved by using data augmentation techniques to
remove redundant data. The harmonic mean of real-time facial expression classifica-
tion using the proposed word was considerably high and is suitable for applications
like classification and verification.

Keywords DCNN · Emotion-based authentication · Convolution · ReLU · Data

augmentation · Batch normalization

A. P. E. James (B)
Department of ECE, Indra Ganesan College of Engineering, Manikandam, Trichy 620012, India
e-mail: [email protected]
M. H. Kit
School of Engineering, Taylor’s University, Taylor’s University Lakeside Campus, No. 1, Jalan
Taylor’s, 47500 Subang Jaya, Selangor, Malaysia
T. A. S. Ritta
Chandy College of Engineering, Mullakadu, Tuticorin 628005, India

© Springer Nature Switzerland AG 2021 175

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_14
176 A. P. E. James et al.

14.1 Introduction

Human thoughts cannot be understood in silence. When a person speaks, his intention
can be understood through his voice or expression. Recognition of a facial expression
plays a vital role in many automated system applications like robotics, education,
artificial intelligence and security. It is said that Facial expressions account for 55% of
the communicated message. Recently, a lot of work has been done for understanding
emotions. Once such method is to process the facial expressions from a dataset of
images. However, a clear understanding about images is necessary for working in
this area.
Images are indispensable in the field of information sciences. Image recognition,
enhancement, segmentation, filtering and many such image-based tasks have evolved
in the recent years. Applications like identification, verification and recognition have
opened doors for emergence for new image processing techniques.
An image is represented by a two-dimensional function, f(x, y), where x and y
are the coordinates in graphical system, and the amplitude of the function f at any
pair of coordinates (x, y) is called the intensity or grey level of the image at that
point. When x, y and the intensity values of f are all finite, discrete quantities, we call
the image a digital image. The field of digital image processing refers to processing
using special image processing systems like computers with Graphical Processing
Units (GPU), FPGA and DSP boards.
Emotional expressions can be classified into seven basic categories namely angry,
disgust, sad, fear, neutral, happy and surprise. In this work, facial expression clas-
sification is performed. Although there are several techniques available for image
processing, facial expression classification is a complex task and requires high-end
processing techniques (i.e.) Machine Learning (ML).
The commonly used techniques include Hidden Markov Model (HMM), Gabor
filtering, Bayesian Networks (BN), Support Vector Machines (SVM) and Deep
Neural Networks (DNN). A facial recognition algorithm that copes with few data,
extracts only useful data from image using a new rotation and cropping strategy
[4, p. 1]. The system uses a simple structure and the algorithm works on 7 class
facial expression recognition. Sujatha et al. [8] proposed another technique using a
convolutional adaptive Neuro fuzzy inference system (CANFIS).
In this work, feature extraction is done to obtain a combination of geometric
features like mean, mode, minimum intensity, maximum intensity, perimeter, stan-
dard deviation, orientation and centroid and face based features. Hence, the FER
algorithm is more robust in nature. A multi-stream CNN based FER techniques a
combination of appearance and geometric features for facial expression recogni-
tion [7, p. 609]. Here three data bases namely Ck+, OULU-CASIA and JAFFE are
combined and utilized by the system and cross-database evaluation is performed.
The accuracy of the system is 78.57%.
The addition of GPUs increases the performance. The authors [3, p. 185] proposed
a FER technique using Deep CNN to classify 40,000 images. A built-in GPU based
is used to improve performance, that are measured through different measurement
14 A DCNN Based Real-Time Authentication System Using Facial … 177

and visualization techniques. A FPGA platform is used to implement a FER system

using CNN [6, p. 651].
Here faces are detected from a video stream using VIFA-2000 camera. In this
method a CNN algorithm available in tensor flow is converted to C arrays and used in
Vivaldo HLS. Here, the performance is enhanced by using representational languor
features and sliding windows. A method used for Human–Robot interaction uses
CNN to extract discriminative expression representation with weight-adaptation and
Hybrid Genetic Algorithm for speeding up convergence [10]. The recognition accu-
racy of this method is higher than the implementations without HGA and is readily
utilized for expression detection by robots.
A recent work in FER was FC layer to avoid inherent error in pre-processing, arti-
ficial selection of features and reconstruction of local Binary Pattern (LBS) features
which enhances the performance [1]. Features extracted by common FER algorithms
include local Binary patterns, eigen faces, face landmark features and Texture features
[9, p. 102]. A recent work combines a CNN and Lenet architecture for FER and tested
on a combination of 3 data sets. (JAFFE, KDEF and Custom data) [5, p. 1].
The method uses small data sets Haar Cascade Library to reduce unimportant
pixels and single depth basement of pixels in the pictures resulting in loss of success
rate, reduced training time and reduced number of pixels.
Most of the previous Algorithms use off- line data sets. In order to use real
time data a method for facial expression classification of grayscale images with a
CNN and real time user interface is used [2, p. 23]. Section 14.2 covers the design
methodology, Sect. 14.3 covers the implementation details, Sect. 14.4 provides the
results and discusses their relevance in current scenario and Sect. 14.5 concludes the
topic.

14.2 Design Methodology

The proposed facial expression system forms the basis of the proposed methodology
The CNN of the proposed system is based on the above concept. The block diagram
of the proposed system is given in Fig. 14.1.
Training deep neural networks with tens of layers is challenging as they can be
sensitive to the initial random weights and configuration of the learning algorithm.
The proposed systems perform facial expression classification in a sequence of stages.

14.2.1 Image Acquisition

Initially, the training data set acquired using a digital camera. Care is taken that
the dimensions of the captured image of the same dimensions (148 × 178). It is
established by fixing the camera at the monitor and the persons involved in the
178 A. P. E. James et al.

Fig. 14.1 Facial expression classification system

process maintain the same distance from the camera. The height of the person is
chosen that the face is situated at a similar height between the different persons.
Five persons are chosen for the experiment and for each person seven standard
expressions (angry, happy, fear, neutral, sad, surprise, and exited) are captured. Each
expression is captured seven times making it 70 for each person and for 5 persons a
total dataset of 350 is captured.

14.2.2 Pre-processing

The entire image may cover regions beyond the face and make feature extraction
difficult. The face image is enough for classification. The other regions are cropped
with a face detection algorithm. The algorithm calculates face circumference based
on characteristics like the centroid of the face and the location of the eyes, ears, nose
and mouth. These are the various points that are estimated around the above facial
features based on which the contour of face is determined.
The region around the facial contour is cropped at a dimension of 50 × 50 × 3
where 3 denotes the RBG intensity of image. The colour of the cropped image is
retained. Ideally the photos are taken at a set environment where the brightness of
the environment is fixed. However, during the experiment small variations due to
the power stability may cause the brightness to vary. The dataset of cropped images
14 A DCNN Based Real-Time Authentication System Using Facial … 179

overwrites the previous dataset but with reduced dimensions and it is stored for test
purposes.

14.2.3 Feature Extraction

Once the data is pre-processed and stored it is subjected to a sequence of layer of

filters. Firstly, there are 2 sequences of convolution, normalization, Rectified linear
Unit RELU and Max pooling layers followed by convolution layer, batch normal-
ization layer, and relu layer. These layers are responsible for extracting the facial
features. Each successive sequence extracts finer features.
A deep neural network combines multiple non-linear processing layers, using
simple elements operating in parallel and inspired by biological nervous system. It
consists of an input layer, several hidden layers and an output layer. The layers are
interconnected via nodes or neurons with each layer using the output of previous layer
as its input. They are also called feature extraction layers. They perform one of three
types of operation on data namely, convolution, pooling and non-linear rectification.

14.2.4 Convolution Layer

The images are allowed set of convolution layers each of which activates certain
features from the images. It performs convolution operation and achieved using
repeated multiplications and additions. Batch normalization is a technique for
training very deep neural networks that standardizes the inputs to a layer for each
mini-batch. This has the effect of stabilizing the learning process and dramatically
reducing the number of training epochs required to train deep networks.
Rectified linear units are used to assign negative values with zero and maintain
positive values. These three operations are repeated successively to detect different
features. The Rectified Linear Unit is the most commonly used activation function
in deep learning models. The function returns 0 if it receives any negative input, but
for any positive value x it returns that value back. So, it can be written as f(x) = max
(0, x).
Max-pooling layer reduces the number of parameters that the network needs to
learn about. The ReLU function works great in most applications. But, the distribution
of the inputs to layers deep in the network may change after each mini-batch when
the weights are updated. This can cause the learning algorithm to forever chase a
moving target. This change in the distribution of inputs to layers in the network is
referred to the technical name “internal covariate shift.”
180 A. P. E. James et al.

14.3 Algorithmic Description

The classification algorithm is given below:

1. Load image data.
2. Define the network architecture.
3. Specify training options.
4. Train the network.
5. Classify the new data into appropriate labels and calculate the classification
accuracy.
The processing steps are illustrated in Fig. 14.2
1. An input layer is where you specify the image size, which, in this case, is 28-by-
28-by-1. These numbers correspond to the height, width, and the channel size.
The digit data consists of grayscale images, so the channel size (colour channel) is
1. For a colour image, the channel size is 3, corresponding to the RGB values. It is
not required to shuffle the data because the training network, by default, shuffles
the data at the beginning of training. Training network can also automatically
shuffle the data at the beginning of every epoch during training.
2. In the convolutional layer, the first argument is filter size, which is the height and
width of the filters the training function uses while scanning along the images.
In this work, the number 3 indicates that the filter size is 3-by-3. You can specify
different sizes for the height and width of the filter. The second argument is the
number of filters, which is the number of neurons that connect to the same region
of the input. This parameter determines the number of feature maps.
3. The ‘Padding’ name-value pair is used to add padding to the input feature map.
For a convolutional layer with a default stride of 1, ‘same’ padding ensures that
the spatial output size is the same as the input size. The stride can also be defined
and learning rates for this layer using name-value pair arguments of convolu-
tion2dLayer. Batch normalization layers normalize the activations and gradients
propagating through a network, making network training an easier optimization
problem. The batch normalization layers are used between convolutional layers
and nonlinearities, such as ReLU layers, to speed up network training and reduce
the sensitivity to network initialization.
4. The batch normalization layer is followed by a nonlinear activation function. The
most common activation function is the rectified linear unit (ReLU). Max Pooling
Layer. Convolutional layers (with activation functions) are sometimes followed
by max The max pooling layer returns the maximum values of rectangular regions
of inputs, specified by the first argument, the pool size. In this work, the size of
the rectangular region is [3, 3]. The ‘Stride’ name-value pair argument specifies
the step size that the training function takes as it scans along the input.
5. The convolutional and down-sampling layers are followed by one or more fully
connected layers. As its name suggests, a fully connected layer is a layer in which
the neurons connect to all the neurons in the preceding layer. This layer combines
all the features learned by the previous layers across the image to identify the
14 A DCNN Based Real-Time Authentication System Using Facial … 181

Fig. 14.2 FEC classification procedure

182 A. P. E. James et al.

larger patterns. The last fully connected layer combines the features to classify
the images. Therefore, the output size parameter in the last fully connected layer
is equal to the number of classes in the target data. In this example, the output
size is 10, corresponding to the 10 classes.

14.4 Results and Discussion

The parameters used to estimate the performance of the authentication system are
accuracy, precision, recall, and F1 score. Predictions made are either considered
Positive or Negative (under some category) and expected judgments are called True
or False (again, under a certain category). Combining them we get four metrics True
Positive, True Negative, False Positive and False. True positive is an outcome where
the model correctly predicts the positive class, True negative is an outcome where the
model correctly predicts the negative class and False positive is an outcome where
the model incorrectly predicts the positive class and False negative is an outcome
where the model incorrectly predicts the negative class Negative.
For the current system, the task involved is language modeling and results are
given in Table 14.1. The common metrics used to access performance precision,
recall, and F1 measure are described in the below section.
Precision: The proportion of correctly classified instances (True Positives) among
all the classified instances under a certain category and is given in Eq. (14.1).

TP
Precision = (14.1)
TP + FP

where TP = True Positive and FP = False Positive.

Recall: The proportion of correctly classified instances (True Positives) under a
certain category and is given in Eq. (14.2).

Table 14.1 Parameter evaluation of experiment 1

Angry Disgust Fear Happy Neutral Sad Surprise Precision Recall F1
Angry 5 3 X X X X X 5 10 6.67
Disgust X 4 X X X X 1 4 10 5.71
Fear X 1 2 X X 2 5 2 10 3.33
Happy X X X 7 X X X 7 10 8.23
Neutral X X X X 8 X X 8 10 8.89
Sad X X X X X 6 2 6 10 7.5
Surprise X X X X X X 8 8 10 8.89
14 A DCNN Based Real-Time Authentication System Using Facial … 183

TP
Recall = (14.2)
TP + FN

where TP = True Positive and FN = False Negative.

F1-measure: Combines precision and recall and is computed as the harmonic mean
between the two metrics and is given in Eq. (14.3).

2 ∗ Pi ∗ Ri
F1 = (14.3)
Pi + Ri

where F1 = Harmonic Mean, Pi = Precision under Category I and Ri = Recall under

Category I. The word category refers to the task performed. The accuracy, processing
time, precision, recall and F1 scores of the individual images are calculated and
finally, the average value indicates the final values of the FEC system.

14.4.1 Experiment 1

Real-time dataset is used in this experiment. 140 images with 10 images for each
expression are captured to form the train dataset. The real-time image is captured
using the same setup and stored in a folder. This image is compared with each of
the train dataset images to find the suitable class to which it belongs. The results are
tabulated and the parameters precision, recall and F1 score of 70.71 is determined
(Table 14.1).

14.4.2 Experiment 2

The initial dataset consists of images chosen from two different persons. 70 images
with 10 images for each expression of person 1 are captured to form the train dataset.
70 images with 10 images for each expression of person 2 are captured to form the
test dataset. The results are tabulated and the parameters namely precision, recall
and F1 score are obtained.
From the experiments the precision of the system directly correlates with the
accuracy of the system and higher values of precision implies system performance
is good. The F1 score is a measure of the harmonic mean of the system and a higher
value of 85.6 is obtained (Table 14.2).
In the first experiment, the datasets used for training and real-time testing are
from one person. Hence the performance is constrained. This shows in a real-time
system, if the train and test dataset are collected from the multiple performance can
be increased. This is validated by experiment two, using the train and test dataset
184 A. P. E. James et al.

Table 14.2 Parameter evaluation of experiment 2

Angry Disgust Fear Happy Neutral Sad Surprise Precision Recall F1
Angry 7 3 X X X X X 7 10 8.23
Disgust X 9 X X X X 1 9 10 9.47
Fear X 1 2 X X 2 5 2 10 3.33
Happy X X X 10 X X X 10 10 10
Neutral X X X X 10 X X 10 10 10
Sad X X X X X 8 2 8 10 8.89
Surprise X X X X X X 10 10 10 10

of the same performance. The results indicate that the performance is considerably
high with a F1 score of 85.6.
The number of total epoch is chosen 30 and total iteration is chosen 30 in order to
observe the training and validation accuracy behaviours where iteration per epoch is
one. The accuracy and loss relations can be seen in Fig. 14.3. The top subplot shows
the training and test accuracy for 30 epochs. In the top subplot, the smoothed line
denotes the training accuracy and the dotted line denoted the validation accuracy. The
bottom subplot shows the training and training and test loss for 30 epochs. In this,

Fig. 14.3 Training process

14 A DCNN Based Real-Time Authentication System Using Facial … 185

Fig. 14.4 Results of the

training for 30 epochs

the smoothed line denotes the training loss and the dotted line denoted the validation
loss.
It is clear from the training results that validation accuracy and loss become
constant around 8th epoch and after that point, training accuracy starts to increase.
The training is repeated with the same option and 30 epochs.
The test results and detailed information related to training can be seen in Fig. 14.4.
Training accuracy is around 100% with this configuration of the network.
The results indicate the F1 score of dataset 1 is 70.71 and that of dataset 2 is 85.6.
Although there is improvement in the harmonic mean, when the dataset is increased,
the harmonic mean can be further enhanced by using more data augmentation
techniques.

14.5 Conclusion

In this work, the performance of real-time authentication system was evaluated. The
system extracts geometric features for classification of images. This system uses the
image as such and does not perform any RGB to Grey conversion. The system has
a wide of range of applications like security, user interface for especially disabled
persons and so on.
186 A. P. E. James et al.

When combined with speech, the level of communication is much better and can
be used in gaming consoles, Flight simulator giving a real experience. A Japanese
company named Softbank has launched a humanoid robot named pepper, that can
converse in 17 language languages and can understand 70–80% human emotion.
Such inventions have opened doors for high-end communication capabilities and
create platforms for sophisticated systems in the future.

References

1. Abdulsalam, W.H., Alhamdani, R.S., Abdullah, M.N.: Facial emotion recognition from videos
using deep convolutional neural networks. Int. J. Mach. Learn. Comput. 9(1), 6 (2019)
2. Canbalaban, E., Efe, M.Ö.: Facial expression classification using convolutional neural network
and real time application. In: 2019 4th International Conference on Computer Science and
Engineering (UBMK), pp. 23–27, Sept 2019. IEEE
3. Chavan, U., Kulkarni, D.: Optimizing deep convolutional neural network for facial expres-
sion recognitions. In: Data Management, Analytics and Innovation, pp. 185–196. Springer,
Singapore (2019)
4. Li, K., Jin, Y., Akram, M.W., Han, R., Chen, J.: Facial expression recognition with convolutional
neural networks via a new face cropping and rotation strategy. Vis. Comput. 1–14 (2019)
5. Ozdemir, M.A., Elagoz, B., Alaybeyoglu, A., Sadighzadeh, R., Akan, A.: Real time emotion
recognition from facial expressions using CNN architecture. In: 2019 Medical Technologies
Congress (TIPTEKNO), pp. 1–4, Oct 2019. IEEE
6. Phan-Xuan, H., Le-Tien, T., Nguyen-Tan, S.: FPGA platform applied for facial expression
recognition system using convolutional neural networks. Procedia Comput. Sci. 151, 651–658
(2019)
7. Salmam, F.Z., Madani, A., Kissi, M.: Fusing multi-stream deep neural networks for facial
expression recognition. SIViP 13(3), 609–616 (2019)
8. Sujatha, K., Vanitha, D., Karthikeyan, V., Krishna, S., Safia, S., Bhavani, N.P.G., Srividhya, V.,
Kumar, P.D.: Facial expression recognition using convolutional adaptive neuro-fuzzy inference
system (CANFIS) (2019)
9. Vyas, A.S., Prajapati, H.B., Dabhi, V.K.: Survey on face expression recognition using CNN.
In: 2019 5th International Conference on Advanced Computing & Communication Systems
(ICACCS), pp. 102–106, Mar 2019. IEEE
10. Wu, M., Su, W., Chen, L., Liu, Z., Cao, W., Hirota, K.: Weight-adapted convolution neural
network for facial expression recognition in human-robot interaction. In: IEEE Transactions
on Systems, Man, and Cybernetics: Systems. (TIPTEKNO), pp. 1–4. IEEE (2019)
Chapter 15
Interdependence of Construction
Projects Problems Using PSO

Abtehaj Hussein, Rouwaida Hussein Ali, Hafeth I. Naji, and Naji Muter

Abstract Project-management area is going through quick progress, a sum of unde-

sirable problems remains impacting construction management projects. This paper
aim to find the interdependency between cost and time problems in construction
projects and determine which problem has major impact on the other problem to find
its solution. The results show PSO is very fast in finding the interdependency between
the problem and the method in searching is very smooth as it find the solution in the
third iteration and that the effect on the problem one on two is slightly different from
the two on one which mean that both require solution and it’s the same apply on the
problem two and three which indicate that the construction phase is very critical and
require great attention.

Keywords PSO · Interdependency · Problems · Construction

15.1 Introduction

Project-management area is going through quick progress, a sum of undesirable

problems remains impacting construction management projects. These problems
comprise use of unsuitable tools and systems for statement, organization, and
organization [1].
The word construction project is usually the use of describing an event that is
implemented as to the needs of the developer of the property and the users of the future
are met. The work of construction project concentrated on the planning, arrangement

A. Hussein · H. I. Naji · N. Muter

University of Diayla, Baquaba, Dialya, Iraq
e-mail: [email protected]
H. I. Naji
e-mail: [email protected]
R. H. Ali (B)
University of Baghdad, Baghdad, Iraq
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 187

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_15
188 A. Hussein et al.

and implementing the work of construction, resolving the construction accounts and
offering a completed construction that is complete for operation [2].
The two main concern for the management of construction are duration and
charge. Numerous issues inking to interruption and overruns of cost and differ along
with project types, sites, dimensions, and scopes. Huge construction projects with
their complexity structures and capital need have caused interest to many scholars.
The usual occurrences are the delay and cost growth in projects. Though, these are
particularly severe in developing countries [3].
Huge previous studies on cost overrun assumes that the cost increase in construc-
tion is due to uncountable reasons connecting to the construction projects features.
Thus, it is vital to examine these reasons and reading the associations that exist among
them and how far to that these reasons impact construction cost increase. This would
lead to development in the estimating practice, decrease in the scale of cost increase
and resultantly authorizing the end cost forecast [4]. Love et al. (2018) showed to the
detail that the capability to find the changes probability to the last budget would allow
the clients of construction to execute plans to manage and control cost throughout
construction.
Naji and Abtehaj have used the PSO in finding the best solution in construction
projects and made some recommendations in finding the interdependency between
these problems.
This paper aim to find the interdependency between cost and time problems in
construction projects and determine which problem has major impact on the other
problem to find its solution.

15.2 Research Methodology

The methodology of the paper includes two parts, the first one is collection of data
from different projects and using questionnaire and the second part include building
the model using PSO.
The information was gathered from previously conducted projects and projects
that under construction to obtain the information that necessary to build the expert
system, the projects include various sectors include The Ministry of Higher Education
and Scientific Research, The Ministry of Construction and Housing and Ministry of
Education.
The preparing of the design form was in two stages, the first one is open ques-
tionnaire, this phase involves leading many interviews with the experts, the expert, is
someone who has special knowledge which is attached to their career or their knowl-
edge related to research, experience, or profession and in a particular aspect of the
study. Experts are needed for giving their advice about their related subject. Experts
have extended or an intensive experience in working and teaching in a particular
field.
The second stage involve closed questionnaire, after conducting the interviews
with many specialists. The research problems of was split into numerous groups
15 Interdependence of Construction Projects Problems Using PSO 189

which embrace, pre-construction, construction problems and their impacts on the

projects, the strategies that used for each problem. Thus, the questionnaire is
formalized. As shown in Table 15.1.

Table 15.1 PSO algorithm for the interdependency

Algorithm PSO
Step 1: Define the optimization parameter
N = 10; % Number of particles
dim = 5; % dimension of input variables
max_it = maxiteration; % number of iterations
ElitistCheck = 1;
inertia = 1.0; % inertia (not used here)
correction_factor = 2.0;
Velocity = zeros(dim,1);
Step 2: Define the source of data that will be used
Step 3: Mathematical model of the problem
MAX Z = ?? (eij * Wj * xij)
% Subjected to
Ci, Sj, Sij, eij, T, B, Dij, Dji, fij (xij), fji (xji), Wj are constant
Ci * MAXj(xi.j) less B
Sj − ?? Si,j * xi.j ? T
Dj12 = f(j1j2) − 0.5 * f(j2j1)
Dj21 = f(j2j1) − 0.5 * f(j1j2)
DJ = n * (1/taoˆ-) D12 + (1 − n) * (1/taoˆ+) * D21
Dj12 < 0.5 select select ? taoˆ−1
Dj21 > 0.5 select taoˆ1
Xij = (0:1)
lmda = (0:1)
tha = (0:1)
n = (0:1)
tao = (0:1)
Wj = lmda(tha * Dj + 1 – tha * Dj) + (1 − lmda)
Step 4: Random initialization for swarm particles
this function initializes the position particles in the swarm, randomly
function [X] = initialization1(dim,N,up,down)
create x1
Step 5: Evaluation of Fitness
Step 6: Calculation of swarm velocity and update all particles of the swarm
for i = 1:N
Velocity(:,i) = correction_factor * rand * (X_Pbest − X(:,i)) + (X_Gbest − X(:,i));
X(:,i) = X(:,i) + Velocity(:,i);
end
for k = 1:N
if BestChart(iteration) == fitness(k)
X_input(:,iteration) = X(:,k);
in_dat(iteration,:) = fdata(k,:);
end
end
190 A. Hussein et al.

The second phase include building the model. This step includes using optimiza-
tion algorithms which is particle swarm algorithm in order to find the best solution
for each problem the methodology is shown in Fig. 15.1. Figure 15.1 show the whole
process and it can be seen some similarity in the general form but in fact the details for
each algorithm and the mathematical model is quite different and it will be explained
in the steps for each algorithm and each type of problem.
The different her is the mathematical modeling which consider the main part in
each algorithm, Table 15.2 will show the variables definition with an example.

Start

Construc on projects
problem, me, cost ,
effec veness ,
interdependency

One Find the

Find the
problem interdependency
Effec veness

Number of par cles

Number of par cles
dimension of input variables number
dimension of input variables
of itera ons
number of itera ons

Period of execu on Time(days) Period of execu on Time(days)

Data
spending Cost, Delay me, base spending Cost, Delay me, Trea ng
Trea ng me, Budget me, Budget

Mathema cal modeling

Mathema cal modeling

Evalua on of Fitness Evalua on of Fitness

Calcula on of swarm velocity and

Calcula on of swarm velocity and
update all par cles of the swarm
update all par cles of the swarm

End

Fig. 15.1 Flow chart of the methodology

15 Interdependence of Construction Projects Problems Using PSO 191

Table 15.2 PSO algorithm variables for two problem

Algorithm variables
i: Problem treatment
j: problems
t max: T Period of execution Time (days)
Ci: spending Cost
Sj: Delay time
Sij: Treating time
Bmax: Budgets
eij=: rating factor
W: weighing factor
tha = consist of all the problems that takes precedence over and the effects of the problem
interdependences are unfavorable (negative) and also in favorable (positive)
lmda: denotes the importance degree of the unfavorable problem interdependence relative to the
favorable one which satisfies γ ∈ [0, 1]
DI and nf: importance degree of the opposed problem interdependence relative to the
satisfactory problem interdependence which satisfies η ∈ [0, 1]
Dj = strength of problem interdependence
for one project
for example
take the C from the j and C2 from the j and for the Sm and Fj like that too
Dj12 = 0.12 − 0.5 * 0.15
= 0.045 select ?-
DJ = 0.5 * (1/2) 0.045 + 0
= 0.045
Wj = 0.9 (0.9 * 0.045 + 1 − (0.9 * 0.045)) + (1 − 0.9)
= 0.8905
Cm = ?1.1E+09 * MAX (0.5)
Cm = 547,500,000
Cm1 = 114,606,550 * MAX(0.3)
= 328,500,000
Sm = ?(21.96 − (0.03 * 0.5)+
= 21.95
Sm = 21.96 − (0.025 * 0.3)
= 21.9675
Mean select strategy 1 for F1 and strategy 2 for F2
If Cm less B and Sm less T
MAX Z
Else
Cm = 1.1E+09 * MAX (0.5)
Cm = 547,500,000
Sm = ?(21.96 − (0.03 * 0.5)+
= 21.95
If Cm less B and Sm less than T
MAX Z
192 A. Hussein et al.

15.3 Results and Discussion

In this section the interdependency between two problem is calculated in order

to know which problem has the greater effect on the other and for the two phase
(Table 15.3).
As it can see the effect of the problem one on two is less than the two on one
which mean that problem one require quick solution (Figs. 15.2, 15.3, 15.4, 15.5,
15.6, 15.7, 15.8, 15.9, 15.10, 15.11, 15.12, 15.13, 15.14, 15.15, 15.16, 15.17, 15.18,
15.19, 15.20 and 15.21).
PSO is very fast in finding the interdependency between the problem and the
method in searching is very smooth as it find the solution in the third iteration
(Table 15.4 and Figs. 15.22, 15.23, 15.24, 15.25, 15.26 and 15.27).
The results show that the effect on the problem one on two is slightly different
from the two on one which mean that both require solution and it’s the same apply on
the problem two and three which indicate that the construction phase is very critical
and require great attention.

15.4 Conclusions

Based on the work of the research, the following conclusions can be drawn:
1. The problems in the pre-construction phase are more than the construction phase
but less impact.
2. The construction phase considers very critical and has an impact of about 25%
on cost and 20% on time.
3. The interdependency between problem quite equal in the construction problems
rather than pre-construction problems.
4. PSO algorithm shows an effective tool in finding the best solution in term of
speed and method of searching.
5. All the velocity of the PSO is zero which indicate the PSO is fast and also in the
method of searching as seen in problem 7 and 9.
Table 15.3 PSO for interdependency in the pre-construction phase (Project One)
Problem z xij tha lmda nf tao Dj Di Wj v
1.2 0.891 0.990001 0.010001 0.53485 0.010001 0.502513 5.00E−07 9.95E−07 1 1.70E−05
1.3 0.891001 0.990001 0.443174 0.552126 0.251424 0.567938 0.08 0.14086 1 0.003839
1.4 0.891001 0.990001 0.2835 0.71647 0.479893 0.502513 5.00E−07 9.95E−07 1 0.000487
1.5 0.891 0.990001 0.404891 0.670826 0.156737 0.523392 5.00E−07 9.55E−07 1 0
2.1 0.9702 0.990001 0.583358 0.669248 0.990001 1.246471 5.00E−07 4.01E−07 1 0.001872
2.3 0.93 0.946928 0.674535 0.983106 0.765017 1.519137 0.0165 0.010861 1 0
2.4 0.9702 0.990001 0.623623 0.49123 0.010001 0.502513 0.05 0.0995 1 0.006957
2.5 0.9702 0.990001 0.964595 0.990001 0.990001 1 0.0165 0.000165 1 6.04E−07
3.1 0.967911 0.987665 0.510789 0.866043 0.078818 0.508821 5.00E−07 9.83E−07 1 7.87E−05
3.2 0.9702 0.990001 0.504084 0.010001 0.424556 4.864892 5.00E−07 1.03E−07 1 0.000205
3.4 0.9702 0.990001 0.583358 0.669248 0.990001 1.246471 5.00E−07 4.01E−07 1 0.00016
3.5 0.967671 0.987421 0.028391 0.530824 0.085799 0.618452 0.083 0.134206 1 1.39E−06
4.1 0.9702 0.990001 0.990001 0.990001 0.463808 0.568066 0.0165 0.029046 1 0.00E+00
4.2 0.967671 0.987421 0.028391 0.530824 0.085799 0.618452 0.0165 0.02668 1 2.89E−05
15 Interdependence of Construction Projects Problems Using PSO

4.3 0.9702 0.990001 0.623623 0.49123 0.010001 0.502513 0.0165 0.032835 1 8.10E−06
4.5 0.9702 0.990001 0.990001 0.691863 0.010001 0.664394 5.00E−07 7.53E−07 1 0.00E+00
5.1 0.891 0.990001 0.010001 0.097888 0.010001 0.502513 5.00E−07 9.95E−07 1 0.00E+00
5.2 0.891 0.990001 0.186701 0.929889 0.010001 0.502513 0.05 0.0995 1 6.08E−06
5.3 0.891 0.990001 0.990001 0.990001 0.071149 100 0.0165 0.000165 1 0.00E+00
5.4 0.891 0.990001 0.010001 0.183033 0.761051 100 0.0165 0.000165 1 0.00E+00
193
194 A. Hussein et al.

Fig. 15.2 PSO for interdependency in the pre-construction phase (Project One, Problem 1.2)

Fig. 15.3 PSO for interdependency in the pre-construction phase (Project One, Problem 1.3)

Fig. 15.4 PSO for interdependency in the pre-construction phase (Project One, Problem 1.4)
15 Interdependence of Construction Projects Problems Using PSO 195

Fig. 15.5 PSO for interdependency in the pre-construction phase (Project One, Problem 1.5)

Fig. 15.6 PSO for interdependency in the pre-construction phase (Project One, Problem 2.1)

Fig. 15.7 PSO for interdependency in the pre-construction phase (Project One, Problem 2.3)
196 A. Hussein et al.

Fig. 15.8 PSO for interdependency in the pre-construction phase (Project One, Problem 2.4)

Fig. 15.9 PSO for interdependency in the pre-construction phase (Project One, Problem 2.5)

Fig. 15.10 PSO for interdependency in the construction phase (Project One, Problem 3.1)
15 Interdependence of Construction Projects Problems Using PSO 197

Fig. 15.11 PSO for interdependency in the pre-construction phase (Project One, Problem 3.2)

Fig. 15.12 PSO for interdependency in the pre-construction phase (Project One, Problem 3.4)
198 A. Hussein et al.

Fig. 15.13 PSO for interdependency in the pre-construction phase (Project One, Problem 3.5)

Fig. 15.14 PSO for interdependency in the pre-construction phase (Project One, Problem 4.1)

Fig. 15.15 PSO for interdependency in the pre-construction phase (Project One, Problem 4.2)
15 Interdependence of Construction Projects Problems Using PSO 199

Fig. 15.16 PSO for interdependency in the pre-construction phase (Project One, Problem 4.3)

Fig. 15.17 PSO for interdependency in the pre-construction phase (Project One, Problem 4.5)
200 A. Hussein et al.

Fig. 15.18 PSO for interdependency in the pre-construction phase (Project One, Problem 5.1)

Fig. 15.19 PSO for interdependency in the pre-construction phase (Project One, Problem 5.2)
15 Interdependence of Construction Projects Problems Using PSO 201

Fig. 15.20 PSO for interdependency in the pre-construction phase (Project One, Problem 5.3)

Fig. 15.21 PSO for interdependency in the pre-construction phase (Project One, Problem 5.4)
 202

Table 15.4 PSO for interdependency in the construction phase (Project One)
Problem z Xij tha lmda nf tao Dj Di wj v
1.2 0.861301 0.990001 0.990001 0.010001 0.010001 100 5.00E−07 5.00E−09 1 0
1.3 0.861301 0.990001 0.010001 0.010001 0.990001 0.626363 0.08 0.127722 1 0
2.1 0.871201 0.990001 0.893624 0.364161 0.815136 6.926421 5.00E−07 7.22E−08 1 0
2.3 0.871201 0.990001 0.990001 0.220145 0.41776 1.859264 0.0165 0.00907 1 0
3.1 0.871201 0.990001 0.010001 0.960172 0.701381 0.549202 0.0165 0.030044 1 0
3.2 0.871201 0.990001 0.990001 0.010001 0.814334 100 5.00E−07 5.00E−09 1 0
A. Hussein et al.
15 Interdependence of Construction Projects Problems Using PSO 203

Fig. 15.22 PSO for interdependency in the construction phase (Project One, Problem 1.2)

Fig. 15.23 PSO for interdependency in the construction phase (Project One, Problem 1.3)

Fig. 15.24 PSO for interdependency in the construction phase (Project One, Problem 2.1)
204 A. Hussein et al.

Fig. 15.25 PSO for interdependency in the construction phase (Project One, Problem 2.3)

Fig. 15.26 PSO for interdependency in the construction phase (Project One, Problem 3.1)

Fig. 15.27 PSO for interdependency in the construction phase (Project One, Problem 3.2)
15 Interdependence of Construction Projects Problems Using PSO 205

References

1. Sidawi, B.: Management problems of remote construction projects and potential IT solutions;
the case of kingdom of Saudi Arabia. J. Inf. Technol. Construct. (ITcon) 17(7), 103–120 (2012)
2. Kasprowicz, T.: Tech. Trans. 2, 177 (2010)
3. Le-Hoai, L., Lee, Y.D., Lee, J.Y.: Delay and cost overruns in Vietnam large construction projects:
a comparison with other selected countries. KSCE J. Civ. Eng. 12(6), 367–377 (2008)
4. Coffie, G.H., Aigbavboa, C.O., Thwala, W.D.: Modelling construction completion cost in Ghana
public sector building projects. Asian J. Civ. Eng. 20(7), 1063–1070 (2019)
 Part III
IoT Applications in Biomedical
Engineering
Chapter 16
Automated Irrigation System with Pest
Detection Using IoT with OTSU
Algorithm

A. V. Kalpana, T. Chandrasekar, S. Rukmani Devi,

and T. C. Jermin Jeaunita

Abstract Internet of Things (IoT) is a pristine technology to track and govern equip-
ment anywhere in the world. A novel strategy for automated water irrigation together
with a platform for pest detection, which can be used to control the water level and
water the crops in agricultural lands, accordingly, is presented in this paper. The
water pump is triggered depending on the water level in the soil. In addition, we have
implemented a new algorithm in this method to identify the pests in the plants. It
will take reasonable measures to eradicate it, depending on the nature of infection.
The proposed algorithm employed is built on the extraction of appropriate features
from the plant leaves and those features are utilized for classification. A comparison
of the proposed algorithm with current algorithms like k-NN and decision tree was
set up to yield admirable results.

Keywords IoT · Smart irrigation · OTSU · Pest · Automatic

16.1 Introduction

The Internet of Things (IoT) requires digital internet connectivity of electronic,

software-embedded computers, sensors and actuators that allow transmission of data
in a seamless fashion. It provides a chance for the direct contact of human world
with world of computer systems and its efficiency and accuracy were tremendous
reducing human involvement. Agriculture is that the backbone and one in every of
important human activity of our nation. In agricultural sector, water is that the most

A. V. Kalpana · T. C. J. Jeaunita
Department of Computer Science & Engineering, R.M.K. Engineering College, Chennai, India
T. Chandrasekar (B)
Department of Business Administration, Kalasalingam Academy of Research and Education,
Krishnankoil, India
e-mail: [email protected]
S. R. Devi
Department of Computer Science & Engineering, R.M.D. Engineering College, Chennai, India

© Springer Nature Switzerland AG 2021 209

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_16
210 A. V. Kalpana et al.

used resource. Irrigation helps to avoid wasting prodigious deal of water. Manual
irrigation is the traditional method utilized in agricultural land and should require
skillful laborers on bigger farm. There are no new approaches to use in conventional
methods for automatic identification of plant diseases and their classification. Unless
recognized in due time, it results in a reduction and loss of enormous quantities and
quality of agricultural production. It needs continuous monitoring of the farm which
will need more laborers and more big farms experts. In far-flung areas, farmers have
to go long distances to get assistances from specialists.
Automated irrigation system with automatic detection of plant infection is a signif-
icant and cultural area of research, as it can help in developing countries’ cere-
monies such as India for earning income without further manual intervention. An
integrated intelligent irrigation network with identification of plant disease by visu-
ally seeing signs is used to eliminate drawbacks of conventional method using IoT.
The proposed system consists essentially of an automated framework of Irrigation
device and surveillance of plant disease by disease signs that are externally detected
in stems and leaves. The irrigation system works at the inputs provided by wireless
temperature and humidity sensor which will be set up in the sector by way of the
Internet of Things for coordination and delivery. Visual disease identification module
is used for taking images, and these images are captured and processed via video for
suggesting a treatment for the corresponding disease. The IoT enables the identifi-
cation/tracking of objects through prevailing network infrastructure remotely. IoT is
augmented with sensors and actuators and the technology turn into an example of
IOT of general category of cyber-physical system, which furthermore comprehends
technologies such as sensible grids, smart homes, smart farming and smart cities.

16.2 Background and Related Work

Irrigation network mechanisms have been used to achieve water savings based on
specific methods such as thermal imaging, direct measurements of the soil water,
crop water Stress index (CWSI), etc. A popular irrigation control technique called
Thermal imaging is focused on the dissemination of plant shaded temperature [1].
S Maheswararajah et.al used unconnected nodes termed as orphaned nodes in the
process of network initialization due to their inability to receive a network address
from a router-capable parent node [2]. A decision-support system using fuzzy logic to
support irrigation management in agriculture, based on the combination of wireless
sensor and actuation network technology is proposed by Viani et al. [3]. M. Ayaz
et al. highlights the promise of wireless sensors and IoT in agriculture, as well as
the challenges that the convergence of this technology with conventional farming
practices will pose [4].
S. N. Daskalaki et al. proposed to implement a novel, energy efficient and low-
cost leaf sensing device with a new plant sensor node. The above, when connected
to an irrigation network, will result in water waste prevention (energy efficiency)
[5]. Goap et al. offers an open-source smart device based on technology to predict a
16 Automated Irrigation System with Pest Detection Using IoT … 211

field’s irrigation requirements using ground parameter sensing such as soil wetness,
temperature of the soil and environmental conditions alongside weather forecast
data from the Internet [6]. Kanungo et al. proposed k-means clustering and simple
methodology for grouping data into a certain set of clusters. The aim is to discover
centroids, one for each group. Primarily, it splits n amount of objects into s non-empty
sub-groups or cluster and then seeks the median point) of the cluster centroids in any
cluster/subgroup. At that time, the distances between each point with the centroids
is measured and assign each entity to a particular cluster, wherever the width of the
centroid is small. The process is iterated by re-assigning points and defining new
cluster centroids [7]. O’Shaughnessy and Evett [8] suggested an automatic irrigation
schedule built on direct measurements of the soil water using water skillfully over
a labor-intensive irrigation system [8]. Gutiérrez et al. proposed to utilize a GPRS
module for an automated irrigation system for saving irrigation water in WSN. In
this device a network of sensors of soil moisture is built in a crop arena for real-
time observing with controller and irrigation is controlled [9]. Gill et al. proposed
a grounding approach prediction of moisture using temperature of the air, relative
humidity of the air and temperature of the soil enabling vector machines [10]. Jaguey
et al. developed smart-phone based irrigation sensor. The optical camera is used to
detect soil moisture and utilized to practice RGB to leaden for the wet to dry estimate
ratio in the field. Wetness and dryness ratio is transmitted to controller of the water
motor via gateway. A Mobile Program Application is built to monitor the operation of
sensor such as wake-up and set sensor to sleep mode [11]. Roopaei et al. modeled an
Intelligent, thermal imaging based irrigation control system. The proposed method
uses Drone mounted camera for thermal imaging. An algorithm is established using
procedures for the image processing to identify water needs, leaf water capacity and
irregular irrigation to be utilized to control irrigation [12].

16.3 Proposed Work

The proposed method as shown in Fig. 16.1 has different kinds of sensors like soil
moisture sensor, temperature sensor, flow sensor, pH sensor and ESP8266 are utilized
to intellect the quantity of water in the soil. Using the spatial and temporal distri-
bution disparity of available water for irrigation marks a significant ultimatum on
water conservation methods. Henceforth, an Automated Irrigation System delivers
a justifiable elucidation to boost water use productivity in the agricultural fields
[13]. This scheme permits farmers to apply the precise amount of water required
at the precise time. According to a predefined condition, this scheme can irrigate
the agricultural fields robotically. It distributes water based on the requirement of a
particular crop [14]. The results have shown that the system would be a paramount
option for moderate size agriculture field and this could minimize the regular irriga-
tion consumption by 50%, reduces the power consumption and also reduces human
intervention significantly. The motor is triggered with a relay switch when the quan-
tity of water is less than the predefined threshold and these values are sent to the
212 A. V. Kalpana et al.

Fig. 16.1 Smart irrigation framework

cloud by means of ESP82600 module and can be monitored by the agriculturalists.

Apart from that, wireless sensor nodes are deployed in the field and these nodes are
used to capture the images of the leaves of a plant as shown in Fig. 16.2.
Using Bluetooth, the images that are captured are transmitted to a central node.
The central node obtain the images of a leaf and recognizes whether the plant has a

Fig. 16.2 Arduino with ESP8266

16 Automated Irrigation System with Pest Detection Using IoT … 213

disease or not and those values are sent to Arduino by means of a Zigbee module.
The Zigbee receiver of the Arduino is utilized to trigger the DC motor driver that is
linked to the spray machine. To avoid infectious pathogens that cause ailments like
Mildew, Ascochyta blight, Anthranose, etc., Copper grounded sprays can be utilized
in the system.
The entire system be made up of two phases namely the training phase and the
testing phase as shown in Fig. 16.4. The images are collected and categorized for
every category during the training phase to create a dictionary. Utilizing the features
that are extracted from the patches of the images of the leaf that are segmented and
dictionaries are created and acted upon during the testing phase. Finally, the distance
computation between the features of the test and dictionaries that are constructed
during the training phase are made and is used to estimate whether the plant is in
good condition or not using the classification. If the plant is infected, then the type
of the disease is also predicted by means of OTSU algorithm.
The acquisition of the image caught by the sensor node is the major step in
the training phase and are enhanced using a sharpening filter. Using the mask, the
sharpening operation is achieved and defined as
⎡ ⎤
−1 −1 −1
ψ = ⎣ −1 −8 −1 ⎦
−1 −1 −1

Using OTSU algorithm, the images are segmented according to the disease and
finally sharpened with the segmented output as shown in Figs. 16.4 and 16.5.
The RGB patches, which are segmented, transformed to grey-scale images which
are used in the feature extraction process. The OTSU algorithm is given below:
Step 1 Draw the histogram of the digital image and calculate the various intensity
levels.
Step 2 Initialize ωi (0) and μi (0).
Step 3 Calculate all probable threshold values from t = 1 … n
Step 4 Update ωi and μi for all possible values.
Step 5 Calculate σb2 for all threshold values t.
Step 6 Desired threshold relates to maximum of σb2 .
A dictionary is constructed with the features that are obtained from each category.
For each group of infectious leaf images and uninfected leaf images, these dictio-
naries were created. The camera gets the image of the leaf that is to be tested during
the testing phase. The distance factor is computed between each dictionary and the
feature needs to be tested.
214 A. V. Kalpana et al.

16.4 Experimental Setup

The experimental setup of the hardware for automated irrigation is shown in Fig. 16.3.
The performance assessment is carried out using the images of the leaf used as
the experimental data derived from the analysis. In our experiment, 11 species of
pests were picked. In this, the images were distributed into twelve partitions and the
assessment was carried out using one section as the study and the rest of the images
are used for testing. The whole cycle was repetitive for ten times, and it determines
the mean average. The accuracy of the classification was calculated and then the
subsequent categories were preferred from the dataset.
Two frequently used metrics called Accuracy and Sensitivity were used. Accuracy
is termed as the proportion of prediction correct. It’s Computed as.
1. Accuracy

The most intuitive one is Accuracy, which is termed as the ratio of the correctly
labelled subjects to the total set of subjects.
Accuracy = (True Positive + True Negative)/(True Positive + False Positive + False
Negative + True Negative).
2. Precision

Precision is the correctly labeled ratio of our system positive to all + ve labels.

Fig. 16.3 Experimental setup

16 Automated Irrigation System with Pest Detection Using IoT … 215

Fig. 16.4 Plot of recall vs epoch

Precision = True Positive/(True Positive + False Positive).

3. Recall

Recall is defined as the ration of True Positive to the sum of True Positive and False
Negative.
Recall = True Positive/(True Positive + False Negative).
4. Specificity
Specificity is defined as the ratio of True Negative to the sum of False Positive
and True Negative.
Specificity = True Negative/(False Positive + True Negative).
It has been concluded that the proposed algorithm outperforms Naïve Bayes and
SVM for detecting the pests which are evident from Figs. 16.4 and 16.5.

16.5 Conclusion and Future Work

In this paper we suggested a new smart irrigation scheme and it is also used to
detect pests that are there in the field using OTSU algorithm. The water needed
for the irrigation is adaptively selected in this scheme based on the soil’s water
content. We have also proposed anew algorithm to classify diseases of plant leaf. This
algorithm has been executed on a database which is publicly accessible. We found
that the proposed algorithm accomplished best in terms of classification accuracy
as compared to all sophisticated techniques. The prototype of the system is cost
effective, since it is built on mainstream open technology. Automotive mode allows
216 A. V. Kalpana et al.

Fig. 16.5 Plot of precision vs recall

as a smart device and can be customized for application specific examples. We intend
to do a water saving in the future study based on suggested multi-node algorithm to
minimize the cost of the system.

References

1. Wang, X., Yang, W., Wheaton, A., Cooley, N., Moran, B.: Efficient registration of optical and IR
images for automatic plant water stress assessment. Comput. Electron. Agric. 74(2), 230–237
(2018)
2. Maheswararajah, S., Halgamuge, S.K., Dassanayake, K.B., Chapman, D., Management of
orphaned-nodes in wireless sensor networks for smart irrigation systems. IEEE Trans. Signal
Process. 59(10), 4909–4922 (2011)
3. Viani, F., BertollI, M., Salucci, M., Polo, A,: Low-cost wireless monitoring and decision support
for water saving in agriculture. IEEE Sens. J. 17(13), 4299–4309 (2017)
4. Ayaz, M., Ammad-Uddin, M., Zubair S., Ali, M., El-Hadi M., Aggoune, E.H.: Internet-of-
Things (IoT)-based smart agriculture: toward making the fields talk. IEEE Access 7, 129551–
129583 (2019)
5. Daskalakis, S.N., George, G., Assimonis, S.D., Tentzeris, M.M., Georgiadis, A.: A uW
backscatter-morse-leaf sensor for low-power agricultural wireless sensor networks. IEEE Sens.
J. 18(19), 7889–7898 (2018)
6. Goap, A., Sharma, D., Shuklab, A.K., Rama Krishna, C.: An IoT based smart irrigation manage-
ment system using Machine learning and open source technologies, Comput. Electron. Agric.
155, 41–49 (2018)
7. Kanungo, T., Mount, D.M., Netanyahu, N.S., Piatko, C.D., Silverman, R., Wu, A.Y.: An efficient
k-means clustering algorithm: analysis and implementation. IEEE Trans. Pattern Anal. Mach.
Intell. 24(7), 881–892 (2002)
16 Automated Irrigation System with Pest Detection Using IoT … 217

8. O’Shaughnessy, S.A., Evett, S.R.: Canopy temperature based system effectively schedules and
controls center pivot irrigation of cotton. Agric. Water Manage. 97(9), 1310–1316 (2010)
9. Gutiérrez, J., Villa-medina, J.F., Nieto-Garibay, A., Porta-gándara, M.Á., Gutierrez, J., Villa-
medina, J.F., Nieto-Garibay, A., Porta-Gandara, M.A.: Automated irrigation system using a
wireless sensor network and GPRS module. IEEE Trans. Instrum. Meas. 63(1), 166–176 (2014)
10. Gill, M.K., Asefa, T., Kemblowski, M.W., McKee, M.: Soil moisture prediction using support
vector machines. J. Am. Water Resour. Assoc. 42(4), 1033–1046 (2006)
11. Hemanth Kumar, G., Ramesh, G.P.: Reducing power feasting and extend network life time of
IoT devices through localization. Int. J. Adv. Sci. Technol. 28(12), 297–305 (2019)
12. Ramesh, G.P., Kumar, N.M.: Design of RZF antenna for ECG monitoring using IoT. Multimed.
Tools Appl. 79, 4011–4026 (2020)
13. Jaguey, J.G., Villa-Medina, J.F., Lopez-Guzman, A., Porta-Gandara, M.A.: Smartphone
irrigation sensor. IEEE Sens. J. 15(9), 5122–5127 (2015)
14. Roopaei, M., Rad, P., Rad, K.K.R.: Cloud of things in smart agriculture: intelligent irrigation
monitoring by thermal imaging . IEEE Cloud Comput. 4(1), 10–15 (2017)
Chapter 17
IoT Based Telemedicine System

L. K. Hema, Rajat Kumar Dwibedi, R. Karthikeyan, and V. Vanitha

Abstract Latest findings suggest that early and skilled pre-operative hospital admin-
istration helps live in emergencies. Ambulance workers do not typically have the tech-
nical skills and expertise needed to cope with emergency emergencies first. Economic
and logistical considerations do not provide for the inclusion of specialized physi-
cians in emergency vehicles. We also created a small medical system within the
context of the Health Telematics software that enables specialist doctors to use tele-
scope tracking, long-range assistance and telecommunication from mobile health
providers. The system enables critical biosignals and photographs of patients to be
submitted to the hospital from site. Data’s are transmitted over a GSM network or
Wi-Fi connection. Owing to the need to share and record data during telemedicine
sessions, we have aligned the consulting network with a digital database that can
store and handle the “ambulance” program captured data.

Keywords Health · Security · Wireless communication · Cloud · Emergency ·

Ambulance

17.1 Introduction

The IoT which simple used as a operation of any device anywhere from the world
capable of transmitting data through a network without the need for interconnected
computer creation, machine-driven and automated mainframe, artifacts, unique iden-
tifiers (UIDs) and humans. Communication from machine or from person to device.
And moreover it is composed of Cloud-enabled intelligent technologies which use
embed technologies such as CPU, trackers and communications equipment pick,
transfer and process data along with existing milieu. The sensors sensor data exchange
that they obtain by linking to an Internet of things database that transfers or analyzes
the data to the cloud. Such systems often communicate with apps, and operate

L. K. Hema (B) · R. K. Dwibedi · R. Karthikeyan · V. Vanitha

ECE Department, AarupadaiVeedu Institute of Technology, Vinayaka Mission’s Research
Foundation (Deemed To Be University), Salem, Tamilnadu, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 219

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_17
220 L. K. Hema et al.

on respect of each other they collect [1–3]. Without human interaction, computers
execute most functions, but humans can communicate with computers-for example,
horizon them up, providing them directions. The networking parameters for these
systems worked on the platform also rely on different IoT applications [4–6]. The
Internet of Things provides numerous benefits for companies. Some advantages are
unique to manufacturing, and others are common in other fields. Track their overall
operating processes and enhance consumer service (CX), save time and resources,
increase efficiency of the staff, incorporate and adjust business models. Helps more
financial choices and gets more money. IoT helps companies to reconsider how they
manage their markets and provide them with the means to develop their market
plan [7]. Using sensors and other IoT tools, IoT is typically prevalent in automo-
tive, logistics and service firms; but, it further seen usage context for businesses
in the cultivation, telecommunications and control sectors, pushing several firms to
digital transformation. IoT-farmers may profit from simplifying their jobs. Sensors
may gather rainfall info, precipitation, degree conditions and soil moisture, quite as
great as other variables that aid to robotize agricultural strategy. Observing func-
tionality activities around the groundwork is one of the factors that may help IoT.
Sensors may be used, for example, to detect incidents or improvements in existing
houses, bridges and other infrastructures [8]. It offers advantages such as expense
reduction, time reductions, improvements in working process efficiency and docu-
ment less development environment. A smart home unit company use IoT to control
and manage a building’s mechanical and electrical systems. The attributed cities can
broadly empower people improve efficiency and resources expenditure. It affects any
sector, including hospitals, banking, retail and manufacturing companies [9]. Many
IoT apps are enabled to do their key functions despite external user intervention.
IoT systems must be capable of functioning dependent on their telemetry data or
the corresponding system input. For illustration, an intelligent smoke detector will
switch on its person’s order, while the person is meters away [10, 11].
In India, the healthcare sector faces multiple problems such as the shortage of suffi-
cient facilities for diagnosing, managing and tracking patient development in little
towns and villages in the countryside. However, this situation is slowly evolving
with (IoT), a network which enables actual-time tracking and knowledge gath-
ering, collection, and knowledge interpretation accompanied by decision-making.
This technological advances will contribute to addressing the technical challenges
in the Indian health sector and ensuring that our large population receives less assis-
tance of medical care. Medical professionals able to integrate mobile platforms to
optimize consumer communication and experiences through pro-programming, e-
mail, and patients-oriented user portals, to name a few [12]. The Indian healthcare
providers are expected to spend $1.2 billion on IT services and products in 2015
and are 7 percent higher than in the previous year, said Gartner. Installation of core
infrastructure, equipments, data centers, IT systems, computers and wireless carriers
network is required for hospitals and other medical services. The IoT positioned
general patient monitoring network model is depicts in Fig. 17.1.
17 IoT Based Telemedicine System 221

Fig. 17.1 General architecture of IoT stationed patient supervise system

17.2 Biomedical Automation in Internet-of-Things

The IoT application is easily widespread by the business world. This is demonstrated
by the substantial improvement in service effectiveness and productivity due to the
use of IoT technology in medical facilities and by incorporating a especially high
advantage for the elderly, survivors of injured persons, critically ill patients and
those in need of ongoing analysis. By certain estimates, funding on healthcare IoT
technologies would reach $1 trillion by 2025 and would likely pave the way for
highly tailored, inexpensive as well as on-time health services for everyone. Such
networks are linked via antenna to frequency Wi-Fi, Zig-Bee, Radio and Ethernet.
In total Bluetooth is not favored since there is a long haul isolation here [13].

17.3 Existing System

In severe situations, IoT based Bio-medical systems are so necessary to save the life of
patients. As with Intensive Care Units to screen all important elements such as ECG,
Pulse Oxygen, Heart Rate, Body Temperature etc., in an overall Mobile Ambulance
system supported by a healthcare provider it is not able to achieve greatness just as all
of these structures are not feasible for the use in a mobile vehicle [14]. Extraordinary
events could lead to a delay in entering the Medical Centre [15]. A new concept to
build a system that is able of communicating the biomedical signals to the Health
Centre when the physician’s documentation is tracked and on wheels via IoT is
carried out to resolve these deficiencies and provide protection for life by through
incentives for savings [16, 17].
222 L. K. Hema et al.

17.4 Proposed System

The system proposed consists of three planes of service as depicted in the Fig. 17.2.
They are:
1. Processing, Sensing, Communication plane.
2. Cloud computing and Concentrations of data.
3. Pre and Post processing by Physician, Analytics.

17.4.1 Sensing, Processing, Communication Plane

The first thing the individual who travels in the ambulance is that the vehicle is fitted
in digital terms through the data acquisition system with various sensors such as ecg,
temperatures, pulse oxygen sensors and performance evaluation. The data are then
interpreted and transmitted via the microcontroller device via a transceiver.

17.4.2 Data Concentration and Cloud Computing

Data is transmitted to the public cloud via mobile phone from the local ground station.

Fig. 17.2 Block diagram of IoT enabled ambulatory system

17 IoT Based Telemedicine System 223

17.4.3 Analytics, Pre and Post Processing by Physician

The data analyzes are performed at this stage and the relevant data is forwarded to
the local phone or laptop through the web. The post analysis or emergency care is
performed in the bulk itself after the data is reviewed by the doctor and based on the
criteria gathered about the person.

17.5 Data Acquisition and Transmission in Ambulance

Using IoT

All medical hardware and patients on a web system are called to the one server named
the Host. In the scenario of a moveable ambulance the patient’s vital physiological
parameters may not be transferred with a central computer to distant doctor locations.
The IoT [8, 9] Data gathering is responsible for several physiologically-backed sensor
devices, such as the ECG, body temperature, respiratory rate, pulse oxygen level and
posture [10, 11], which are designed to measures variables. Sensing information,
this essential data is conveyed by a wireless device (Zigbee, Wi-Fi or Bluetooth),
from either the client to the multiple data aggregator that is usually a mobile. The
consolidated information is then transferred via smartphone or WiFi from a distant
area, e.g. from the client’s home or from an accident scene to the nearest health centre.
Sometimes the transitional signal strength of the data aggregator does not meet data
transfer and storage for various uses and the cloud network can then be used. The
public cloud is nothing but a computer with more data analysis capabilities for the
desktop/laptop and functions as a web server [13]. The next step is the transmission
of data from the cloud network into the network for every 5 min or even less through
IoT based cloud infrastructure. In the data center the clinical data such as entry
time, physiological condition, stance, etc. is stored. Doctors can obtain the collected
information about the particular Web servers. The cloud storage are updated regularly
by analysis tools and post-processing [15]. The database stores long-term knowledge
and helps physicists to treat and take medication. Such data can also be saved in the
cloud which will aid in the creation of e-Health records, discharge summary etc. We
also use mathematical methods to map sensory data as graphs and plots in our model.

17.6 Results and Discussions

We established an internet-of-things-enabled ECG and pulse [18, 19] oximeter that

detects air, the oxygen saturation level and the heart rate (Fig. 17.3). The ECG
sensors and pulse oximetric sensors are mounted on the patient’s body and the critical
information is detected by these devices. This information is then transmitted to and
processed by the node microcontroller. Then the data is sent to the mobile telephone
224 L. K. Hema et al.

Fig. 17.3 Prototype of IoT enabled ambulance system

(intermediate data aggregator). The data is stored locally via wireless Internet via
mobile telephone. In your remote desktop, you can view the values and plots using
the doctor’s cloud storage.

17.7 Conclusion

The IoT allows health care in India to be provided in a friendly way. Through ‘digital
the ICT infrastructure also serves the healthcare industry implicitly. The system
originally designed improves the feasibility of the care of the patient, enhances the
effectiveness of the meds and reduces the provider considerably. The obstacle that
we face is that network connectivity is always available.
17 IoT Based Telemedicine System 225

References

1. Pardeshi, G., Kakrani, V.: Challenges and options for the delivery of primary health care in
disadvantaged urban areas. Indian J. Community Med. 31(3), 132 (2006)
2. Evans, D.: The internet of things: How the next evolution of the internet is changing everything.
CISCO. Int. J. Internet 3(2), 123–132 (2011)
3. Hassanalieragh, M., Page, A., Soyata, T., Sharma, G., Aktas, M., Mateos, G., Andreescu, S.:
Health monitoring and management using internet-of-things (IoT) sensing with cloud-based
processing: Opportunities and challenges. In: 2015 IEEE International Conference on Services
Computing, pp. 285–292 (2015)
4. Babu, S., Chandini, M., Lavanya, P., Ganapathy, K., Vaidehi, V.: Cloud-enabled remote
health monitoring system. In: 2013 International Conference on Recent Trends in Information
Technology (ICRTIT), pp. 702–707 (2013)
5. Rolim, C.O., Koch, F.L., Westphall, C.B., Werner, J., Fracalossi, A., Salvador, G. S.: A cloud
computing solution for patient’s data collection in health care institutions. In: 2010 Second
International Conference on eHealth, Telemedicine, and Social Medicine, pp. 95–99 (2010)
6. Zhao, W., Wang, C., Nakahira, Y.: Medical application on internet of things (2011)
7. Hu, F., Xie, D., Shen, S.: On the application of the internet of things in the field of medical and
health care. In: 2013 IEEE International Conference on Green Computing and Communications
and IEEE Internet of Things and IEEE Cyber, Physical and Social Computing, pp. 2053–2058
(2013)
8. Nalinipriya, G., Kumar, R.A.: Extensive medical data storage with prominent symmetric algo-
rithms on cloud-a protected framework. In: International Conference on Smart Structures and
Systems-ICSSS’13, pp. 171–177 (2013)
9. Pang, Z.: Technologies and architectures of the internet-of-things (IoT) for health and well-
being (Doctoral dissertation, KTH Royal Institute of Technology) (2013)
10. Fezari, M., Rasras, R., El Emary, I.M.: Ambulatory health monitoring system using wireless
sensors node. Procedia Comp. Sci. 65, 86–94 (2015)
11. Mehta, M.: ESP 8266: A breakthrough in wireless sensor networks and internet of things. Int.
J. Electron. Commun. Eng. Technol. 6(8), 7–11 (2015)
12. Hema, L.K., Murugan, D., Chitra, M.: Wireless sensor networks’ emergence and Growth-A
survey. Int J Comput Eng Res (IJCER), pp. 2250–3005 (2012)
13. Pandey, A., Prakash, G.: Deduplication with attribute based encryption in E-health care systems.
Int. J. MC Square Sci. Res. 11(4), 16–24 (2019)
14. Shahada, S.A.A., Hreiji, S.M., Shamsudheen, S.: IOT BASED GARBAGE CLEARANCE
ALERT SYSTEM WITH GPS LOCATION USING ARDUINO. Int. J. MC Square Sci. Res.
11(1), 1–8 (2019)
15. Balaji, K., Kumar, R.A.: Adaptive super-twisting sliding mode controller-based PMSM fed
four switch three phase inverter. In: Intelligent Computing in Engineering, pp. 327–337 (2020)
16. Sivasankar, C., Kumanan, T.: Friend list-based reliable routing in autonomous mobile networks.
In: Intelligent Computing in Engineering, pp. 409–415 (2020)
17. Ramesh, G.P.: Design of digital fir filters for low power applications. In: Intelligent Computing
in Engineering, pp. 433–440 (2020)
18. Hema, L.K., Mohana Priya, R., Indumathi, R.: Real time ecg monitoring using IoT gateway.
Int. J. Psychosoc. Rehabil. 10499–10504 (2020)
19. Hema, L.K., Mohana Priya, R., Indumathi, R.: Smart pulse oximeter systems for Spo2
measurement based on IoT. Int J Psychosocial Rehabil. 10499–10504 (2020)
Chapter 18
Tracking Greenhouses Farming Based
on Internet of Technology

Sebastián Gutiérrez, Rafael Rocha, David Rendón, Juan Carlos Bernabé,

Luis Aguilera, and Vijender Kumar Solanki

Abstract The Internet of Things plays role in all arenas. In this manuscript, we have
to consider the growing flowers in a garden, vegetable, fruit, and other farming. We
are considering the greenhouse which aims to introduce the productions of yields.
Of course, the growth of plants, and farms are vital and need of everyone, keeping
in view of this manuscript is aimed to discuss and study in line of IoT and agricul-
ture. In this work, we propose a greenhouse automation system based on Arduino
for the monitoring of temperature, humidity, and moisture of the soil. Arduino can
obtain data on the environmental conditions of the greenhouse from various sensors
and transfer the data to the ESP8266 module. Consequently, it’s possible to change
the state of greenhouse control devices like fans, lamp heater, and water pump in
obedience to the necessary conditions of the crops. These parameters are modified
by the type of plant to maximize their growth, the Aloe Vera plant was used in this
project. For the architecture of the Internet of Things was used Blynk coming from
the embedded board and the communication link with the Blynk Server was through
the Wi-Fi protocol. Results indicate that the system allows the control and monitoring
in real-time of the greenhouse correctly. As a future improvement, it is intended with

S. Gutiérrez (B) · R. Rocha · D. Rendón · J. C. Bernabé · L. Aguilera

Facultad de Ingeniería, Universidad Panamericana, Aguascalientes, México
e-mail: [email protected]
R. Rocha
e-mail: [email protected]
D. Rendón
e-mail: [email protected]
J. C. Bernabé
e-mail: [email protected]
L. Aguilera
e-mail: [email protected]
V. K. Solanki
Department of Computer Science and Engineering, CMR Institute of Technology, Hyderabad, TS,
India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 227

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_18
228 S. Gutiérrez et al.

the data obtained, to search for the best optimal conditions for plant growth through
artificial intelligence.

Keywords Arduino · ESP8266 · Greenhouse · Home automation · Internet of

things

18.1 Introduction

In recent years, large-scale modern greenhouses have been used for precision agri-
culture. In addition, greenhouses must have different species of swamps for different
seasons, for this, it’s necessary to monitor a number of parameters such as tempera-
ture, humidity, and light through sensors that collect information about the weather
condition of the greenhouse and send it to a base station for monitoring [1]. In order
to replicate a greenhouse for residential houses and small farms, it is also required
some sensors and actuators, all these real time parameters are measured and admin-
istrated by a base station too. The base station is able to modify weather conditions
according to the preset parameters of the desired plant to be grown. ESP8266-based
wireless monitoring and control system for greenhouses is designed of a base station
and front-end devices including sensors and electrical actuators organized as a star
network shown in Fig. 18.1 [2–6].
This paper presents the system that gets and controls condition of greenhouse
environment for Aloe Vera crops by using sensors and actuators. The control system
monitors the humidity, temperature, light intensity and the soil moisture of the green-
house. The prototype collects the information of the sensors to control the weather

Fig. 18.1 Star network applied in greenhouse

18 Tracking Greenhouses Farming Based on Internet of Technology 229

condition of crops depending on internal environment factors of greenhouse based

on the preset parameters registered in the software.
This document is organized as follows: Sect. 18.2 describes the properties
of the Aloe Vera plant. Section 18.3 describes the design system of the proto-
type. Section 18.4 describes the hardware implementation of the proposed system.
Section 18.5 describes the display data. Section 18.6 summarizes the results and
Sect. 18.7 gives the conclusions of the study.

18.2 Aloe Vera Plant

Aloe Vera (aloe barbadensis miller, belongs to liliaceae family) is a medicinal plant
that has various health properties and is used in various beauty treatments, mainly at
the level of hair and skin. It is a perennial plant, growing to the high of 45 ~ 76 cm.
The leaves are long and thick, the two sides of the leaves have thorns structure with
a spiny tip [7]. Figure 18.2 shows a typical Aloe Vera plant. In order to grow this
plant, the following conditions are needed.

Fig. 18.2 Aloe Vera plant

230 S. Gutiérrez et al.

18.2.1 Soil

Aloe Vera grows in warm conditions of humidity and rain. It’s cultivated in all types
of soils, but well-drained soil with high organic matter is the most suitable for its
growth. These types of plants grow well in bright sunlight. Shady conditions could
cause plant health discomfort; also it’s very sensitive to stagnant water. Hence, the
Aloe Vera grows best when supplied with an excess of 50 cm of rain per year, in
alkaline soils rich in nitrogen. Ideally, soil nitrogen should be kept at 0.40–0.50.
While most aloe species tend to grow in sandy soil [7, 8].

18.2.2 Irrigation

Aloe Vera is slightly tolerant to drought. Aloe can be successfully grown under both
watering and rainy conditions. The irrigation supply immediately after planting and
during the summer season will ensure a good yield [7].

18.2.3 Indoor Cultivation

The Aloe Vera plant usually grows openly in the field, but it can also be grown
indoors, in pots that have all-day sunlight. For the plant to grow properly, the soil
of the pot should be well-drained, the plant does not tolerate waterlogging. The soil
can be sandy or clayey, thick kidney, broken brick, or crushed limestone can be used.
The plant should be potted in the spring and watered carefully until stiff. During the
summer, water can be added to the pot as long as the soil is dry, but in the months of
September to March very little water is needed, enough to prevent the leaves from
wilting [9].

18.3 System Design

The structure of the prototype is shown in Fig. 18.3, it encloses an ESP8266 micro-
controller as the main processing unit and it gets inputs data from the Arduino. The
embedded board, Arduino, read the temperature and humidity sensor, LDR sensor,
soil moisture sensor and water sensor. From the data obtained since the sensors the
program controls the actuator components (light bulb, fan and water bomb) to achieve
the requirements of the system. It also uses a smartphone to display the data obtained
from the sensors. The light bulb, cooling fan, and water bomb are connected to the
Arduino, relay control water pump and fans. Presets set parameters accord an Aloe
Vera plant requirement [10, 11].
18 Tracking Greenhouses Farming Based on Internet of Technology 231

Fig. 18.3 Block diagram of the adaptive greenhouse system

18.4 Hardware of the System

The ESP8266 is a low-cost Wi-Fi Module. It’s integrated with a Tensilica L106 32-
bit microcontroller (MCU) and an ultra-low-power 16-bit RSIC. The power-saving
architecture of the embedded board works in 3 modes: active mode, sleep mode and
deep sleep mode. ESP8266 consumes approximately 60 mA in deep standby mode
with the RTC clock still running. Integration of sensors or actuators with ESP8266
board is easy via GPIO pins, which creates a way to connect the ESP8266 board
to the outside world. In this project, the ESP8266 module was used as a wireless
communication module with the cloud through the Internet of Things. The advantage
of using the ESP8266 board is that in addition to its connectivity to the cloud, it has
an integrated microcontroller integrated, which allows sending and receiving data
from sensors and actuators [12].
The architecture of the greenhouse prototype is divided into three parts and is
shown in the Fig. 18.4. The first part is the acquisition of sensor data; sensors detect
an ambient parameter of temperature, humidity and luminous intensity. After Arduino
get data from sensors, it transmits the sensors readings to ESP8266. The second part
is the reading of the sensor parameters, where the Arduino Mega sends the data to
ESP8266. The third part is the Blynk APP that shows the data to the user.

18.4.1 Temperature and Humidity Sensor

There are different types of temperature sensors available on the market, which
can vary in accuracy depending on the application. The DHT11 sensor is a low
232 S. Gutiérrez et al.

Fig. 18.4 Architecture of the greenhouse prototype

cost, medium precision relative humidity and temperature sensor. The sensor has a
capacitive humidity sensor and a thermistor that allows you to measure the circu-
lating air, it is possible to obtain a digital signal through the data pin. This sensor is
widely used in applications related to automatic temperature control, air conditioning,
environmental monitoring, and other applications.
18 Tracking Greenhouses Farming Based on Internet of Technology 233

18.4.2 Light Sensor

The LDR (Light Dependent Resistor), also known as a photoresistor, is a device

whose resistance varies according to the amount of light it receives. They are very
useful for automatic lighting control projects, sunlight trackers, twilight switches,
etc. [13].

18.4.3 Water Sensor

The water sensor is used for liquid detection, which can be widely used in detecting
precipitation, water level, and even some kind of liquid leakage. It can be widely
used to detect the presence or absence of water, level, and volume. With the help of
this sensor and a microcontroller, it is possible to monitor and control various actions
based on the presence of a liquid [14].

18.4.4 Actuator and Power Supply

As Fig. 18.3 shows, the actuators in the proposed prototype are an electric motor that
drives a water pump, a cooling fan, and a light bulb. In general, for most actuator
elements such as electric motors, fans and lights can be controlled by relays if the
power consumption is low, as the prototype shows. However, for more advanced and
powerful systems, a motor drive is needed, which allows controlling the start of the
motor and the speed of the pump.
Usually, motor drives require DC/AC converters, detailed information about these
types of circuits can be found in [15–17]. Regarding funs, the problem is the same,
a power electronic interface such as a motor drive can control the speed of the fun,
so a more advanced control can be obtained. The light bulb intensity can also be
controlled. In this case, the correct circuit is a DC/DC converter [15]. Finally, if all
of these actuators are driven by power electronics, a bigger power supply is needed.
Therefore, the batteries must be replaced with a grid-connected power supply which
requires a rectifier and a DC/DC converter [18].

18.5 Data Display

Blynk is a digital dashboard where it’s possible build a graphical interface for any
type of project by simply dragging and dropping widgets. It is possible to control the
hardware in real-time, display data from connected sensors and it is also possible to
store data [19]. Figure 18.5 shows the Blynk cloud architecture and the user interface.
234 S. Gutiérrez et al.

Fig. 18.5 Blynk cloud architecture

To understand how Blynk works, the main parts that make up the system are:
• The Blynk APP application allows you to create control interfaces for any project.
• The Blynk server is responsible for all communications between the cell phone
and the hardware. You can use the online server or install the local Blynk server.
The Blynk server is an open source application.
• Blynk libraries are available for different development platforms, allowing
communication between the server and hardware resources.
Concerning the above, the greenhouse interface could be carried out correctly
simply and easily using the Blynk APP. Figure 18.6 shows the interface of the values
of temperature, humidity, and presence of water.

18.6 Results

To analyze and optimize the performance of the system, an experiment has been
carried out in Aguascalientes City. The geographical coordinate of the Greenhouse
was 21°52 56 N 102°16 57 O. The values of humidity, temperature, and light
intensity, presence of water and soil moisture were monitored in the system. The
information was sent to the Arduino Mega located outside the greenhouse. For the
experiment, information was taken every hour from 9:00 to 18:00 h. Figure 18.7
shows the experimental setup of the greenhouse. In the image, it’s possible to observe
the different sensors used in the prototyping in addition to the greenhouse structure.
Figure 18.8 shows the parts of computational capabilities, including the network
connection made by the ESP8266 card, in addition you can see an LCD screen, where
it is possible to show the information or some data failure in the greenhouse.
18 Tracking Greenhouses Farming Based on Internet of Technology 235

Fig. 18.6 Blynk user

interface

Table 18.1 presents the measurements obtained from the monitoring of the devel-
oped greenhouse. Temperature (°C), humidity (%), water presence (Yes/No), soil
moisture (High/Low) and LDR (High/Low) are displayed.
236 S. Gutiérrez et al.

Temperature/
Humidity sensor

Moisture sensor
Light sensor

Water sensor

Fig. 18.7 Setup of the greenhouse

Fig. 18.8 Computational capabilities of the greenhouse

18 Tracking Greenhouses Farming Based on Internet of Technology 237

Table 18.1 Temperature, humidity, water presence, soil moisture and LDR measurement results
Time Temperature (°C) Humidity (%) Water presence Soil Moisture LDR
9.00 27 52 Yes High High
10.00 28 46 Yes High High
11.00 30 35 Yes High High
12.00 32 18 Yes Low High
13.00 31 19 Yes Low High
14.00 32 18 Yes Low High
15.00 33 16 Yes Low High
16.00 33 16 Yes Low High
17.00 32 18 Yes Low Low
18.00 30 35 Yes High Low

18.7 Conclusions

Greenhouse prevents the plant from the weather effects and insects, which calls the
attention of the agricultural community. Can avoid pests, It can help improve farmers
if it is proposed for large greenhouses.
The automation and high efficiency of greenhouse environment monitoring and
control are crucial. This project shows the real meaning of the IoT and how to properly
manage the growth of the plants. The importance of real-time monitoring is to be
able to know the health of the plant.
Our greenhouse purpose is to help people that do not have access to a field, yard,
or a place to grow plants and want a plant or fresh vegetables right at their home.
In homes, apartments, buildings, there is no place to harvest. The greenhouse proto-
type offers a plant growing with maximum efficiency, with low-cost technologies to
encourage people that want to have their small farm and start making a change at
their house.

References

1. Mohanty, N.R., Patil, C.Y.: Wireless sensor networks design for greenhouse automation. Int.
J. Eng. Innov. Technol. 3 (2013)
2. Wan, Z., Song, Y., Cao, Z.: Environment dynamic monitoring and remote control of green-
house with ESP8266 NodeMCU. In: Proceedings of 2019 IEEE 3rd Information Technology,
Networking, Electronic and Automation Control Conference, ITNEC 2019 (2019)
3. Taha, F.M.A., Osman, A.A., Awadalkareem, S.D., Omer, M.S.A., Saadaldeen, R.S.M.: A
Design of a Remote Greenhouse Monitoring and Controlling System Based on Internet of
Things. In: 2018 International Conference on Computer, Control, Electrical, and Electronics
Engineering, ICCCEEE 2018 (2018)
238 S. Gutiérrez et al.

4. Gutiérrez, E., Gutiérrez, S., Becerril, J.A., Rodríguez, F.: Low Cost Prototype for Monitoring
and Remote Control in Greenhouse for Homes. In: 2018 IEEE International Autumn Meeting
on Power, Electronics and Computing (ROPEC 2018) (2018)
5. Lara, J.C.D., Gutierrez, S., Rodríguez, F.: Low Cost Greenhouse Monitoring System Based on
Internet of Things. In: 2019 IEEE International Conference on Engineering Veracruz (ICEV),
pp. 1–6 (2019)
6. Soto-Zarazúa, G.M., Romero-Archuleta, B.A., Mercado-Luna, A., Toledano-Ayala, M., Rico-
García, E., Peniche-Vera, R.R., Herrera-Ruiz, G.: Trends in automated systems development
for greenhouse horticulture. Int. J. Agric. Res. (2011). https://doi.org/10.3923/ijar.2011.1.9
7. Biswas, B.C.: Cultivation of medicinal plant. Indian Fertil. Mark. News. 41, 1–4 (2010)
8. Manvitha, K., Bidya, B.: Aloe vera: a wonder plant its history, cultivation and medicinal uses.
J. Pharmacogn. Phytochem. (2014)
9. Rajeswari, R., Umadevi, M., Rahale, C.S., Selvavenkadesh, S., Kumar, K.S., Bhowmik, D.:
Aloe vera: The miracle plant its medicinal and traditional uses in India. J. Pharmacogn.
Phytochem. (2012)
10. Faris, D.M., Mahmood, M.B.: Data acquisition of greenhouse using arduino. J. Babylon Univ.
Appl., Sci. (2014)
11. Vimal, P.V., Shivaprakasha, K.S.: IOT based greenhouse environment monitoring and control-
ling system using Arduino platform. In: 2017 International Conference on Intelligent
Computing, Instrumentation and Control Technologies, ICICICT 2017 (2018)
12. Kodali, R.K., Mahesh, K.S.: Low cost ambient monitoring using ESP8266. In: Proceedings
of the 2016 2nd International Conference on Contemporary Computing and Informatics, IC3I
2016 (2016)
13. Efficiency, E., Circuit, E., Dependent, L., Circuits, R.: Light dependent resistor (ldr). Energy
(2010)
14. Grove—Water Sensor. https://wiki.seeedstudio.com/Grove-Water_Sensor/
15. Mohan N., Undeland T.M., Robbins W.P., Power electronics converters, applications and
design, 3rd edn, Wiley (2002)
16. Florez-Tapia, A.M., Ibanez, F.M., Vadillo, J., Elosegui, I., Echeverria, J.M.: Small signal
modeling and transient analysis of a Trans quasi-Z-source inverter. Electr. Power Syst. Res.
(2017). https://doi.org/10.1016/j.epsr.2016.10.066
17. Ibanez, F.M.: Bidirectional Series Resonant DC/AC Converter for Energy Storage Systems.
IEEE Trans. Power Electron. 34, 3429–3444 (2019). https://doi.org/10.1109/TPEL.2018.285
4924
18. Ibanez, F., Echeverria, J.M., Vadillo, J., Fontan, L.: High-current rectifier topology applied to a
4-kW bidirectional DC–DC converter. IEEE Trans. Ind. Appl. (2014). https://doi.org/10.1109/
TIA.2013.2271471
19. B. Inc., “Blynk,” Blynk Inc. https://www.blynk.cc.
Chapter 19
Impact of Internet of Things
and Artificial Intelligence on Human
Resource Development

Sonal Pathak and Vijender Kumar Solanki

Abstract In today’s smart organizations, Human Resource Management processes

and practices are hugely affected by information technology. To retrieve organiza-
tional knowledge from human knowledge, Artificial Intelligence plays a vital role.
Adoption of innovative and robust HR practices to beat the market competition,
can be fulfilled by advanced technology like Automation, Augmented Intelligence,
Robotics, which can help in the reshaping their workforce. The impact of Artificial
Intelligence takes hold on mainly four areas. The first is customer satisfaction, which
comes through their product experience. Until we make our Human Resources ready
to learn machine language to increase the quality of customer service until then,
consumer expectations and satisfaction cannot be fulfilled. The second area is the
recruitment process, where Artificial Technology can be used in different ways to
improve the overall outcomes. The next aspect can be around training and develop-
ment. As we can see, changes in platforms based on customer’s preferences due to
multi options available. Therefore, to manage global competition, our HR should be
trained enough to deal with innovative methods based on automation supported tech-
nology. The last one is based on the analysis of data based on people. Organizations
have varied data generating on a daily basis, so it is time to make use of information
retrieved from data analysis. Therefore, HRM processes should now be updated and
technology-based. Every organization must look at these four areas when they think
about the influence of AI on their HR practices. In this chapter, the analysis of the
impact of Artificial Intelligence on the HRM process has been discussed. Application
of AI tools for screening of candidates, engagements of employees, and their career
enhancement have been explained thoroughly. Some challenges in HR practices
based on data science are extensive HR phenomenon, and there are many restraints
due to small data sets. In this chapter, we have analyzed the gaps between the realism

S. Pathak (B)
Manav Rachna International Institute of Research and Studies, Faridabad, India
e-mail: [email protected]
V. K. Solanki
CMR Institute of Technology, Hyderabad, TS, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 239

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_19
240 S. Pathak and V. K. Solanki

of Artificial Intelligence in managing Human Resources and expectations. Sugges-

tions have also been discussed for the progress to be made. The methods to have
sustainable HR and talent acquisition, training through the utilization of technology
have also been discussed in the present chapter. Practical examples to understand the
collaboration of working with AI are considered as well.

Keywords Human resources · Human resource management practices · Artificial

intelligence · Automation · e-HRM · Enterprise resource planning ·
Human-machine interaction · Digitization

19.1 Introduction

The nature of jobs are changing in the current global environment. Brain and finger
professions are overpowering the jobs requiring muscle power. This chapter describes
the impact of Artificial Intelligence on Human Resource practices and how it is
reinventing HR processes. India has recently come across these changes though
these radical changes were mentioned by Charles Handy a long back in 1984. Now
networking can be enhanced via the internet in this highly competitive world; tech-
nology is creating a global village through fast up-gradation. Artificial Intelligence
is helping organizations in performing any function. The HR managers understand
the importance of using technology in strategic decisions, and higher mangers are
under pressure in involving HR mangers in strategic decisions. Senior management
has recognized the significance of technology-based Human Resource systems. A
quick response is very much required to manage rapid changes in the business envi-
ronment. The present chapter explores and discusses Research and innovation to
develop new technologies to compete in the global and digitized market. People who
are not competent in data analysis are using Artificial Intelligence to prompt review
of a given data. Platform powered by Artificial Intelligence supports all HR func-
tions. Artificial Intelligence does the integration of the different systems from hiring,
training, job growth, performance, and compensation management.

19.1.1 Employees as Resource

Persons who do work for an organization or a business and the division which handles
employees together called Human Recourse. In 1960s, terms like—motivation, orga-
nizational behavior, and selection assessment have stared, and took shape and atten-
tion was given to the importance of labor relations—“Human Resources”. We can
define Human Resources as: “The individuals that operate an organization and work
as staff” as compared with the material and financial resources of an organization.
19 Impact of Internet of Things and Artificial Intelligence … 241

People are an essential part of any organization. The vital objectives of any organiza-
tion are hiring employees, their skill development, keeping their level of performance
high through continuous motivation, and ensuring their commitment for the firm [1].
To make a triumph at the end, the HR department act as an essential division
for optimal utilization of Human Resources/workforce of an organization. Employ-
ment relations among personnel are always expected to showcase cohesiveness and
harmony. It is also the responsibility of HR department to develop good working envi-
ronment for the staff/personnel and consequently makes a difference for an impro-
vised business world. Personnel managers are responsible for each aspect of the
employee life cycle in a corporation.

19.1.2 Management of Human Resource

Component of management, which consists of the employee’s job division with

their connections, is called Human Resource Management (HRM). HRM comprises
methods that support the organization’s deal effectively along with their individuals
with numerous stages of the service cycle: prerecruitment, choice, and after the selec-
tion procedure [2]. The fundamental purpose of the management of Human Resources
is to confirm the readiness of capable and skilled staff. The Human Resource depart-
ment is purely a tag along with division with employee-centered plans. The responsi-
bilities include planning of employment and way of selecting competent employees,
posting an advertisement of job, appraising the employees’ performance, filtering
aspirants’ applications of job, organizing meetings, and aiding in the procedure and
make sure experience checking.
Managing payroll of employees and various administrative allowances are the
additional job of Human Resource executives which deals with certifying breaks of
employees for which they are entitled, appraising payroll, and contributing in aids
jobs responsibilities, like approving invoices, reconciling benefits statements, and
claim solutions for payment. Most recent work is frequent fund preservation, and
another assignment is to keep HR records and existing HR documents up to date,
preserving employee’s rights and work standards regarding benefit-related/payroll
settlements. In an institute, management of Human Resource is a support word which
is used to explain the development and management of employees. Also, employees
or talent management (though these words are a bit outdated), Human Resource
Management requires overseeing all matters related to handling an organization’s
human investment. Human Resource administration needs emerging and managing
plans which intended to the growth the efficiency of an organisation. It comprises the
complete continuum of managing, generating, and developing the manager-worker
association [3].
For most corporations, agencies, and enterprises, the Human Resources division
is responsible for mainly the following activities:
• Supervising job recruitment, selection, and promotion
242 S. Pathak and V. K. Solanki

• Overseeing and developing employee wellness and benefits programs

• The building, organizing, and enforcing employees’ policies
• Fostering employee career development and job training
• Offering orientation programs for new employees
• Supporting guidance regarding corrective actions
• Providing as a critical contact for work-site grievances or accidents.

19.2 Traditional Methods of Human Resource

Management Practices

Methods, procedures, and systems, which affect the employee’s behavior and perfor-
mance of employees are called HRM (Human Resource Management.) According
to Dessler, Procedures of Human Resource involves analyzing the requirement
of Human Resources, their screening and employee’s recruitment, rewarding,
employees’ training and development, labor relations, health and safety, legal matters
[1]. Current years have been showing increasing significance of organizations which
are incorporating knowledge-based methodologies [4]. In the private sector of India,
candidate’s viewpoints are also involved in the multi-policy, which consists of new
approaches of screening, recruitment, training, and development. Fast environmental
modifications will keep up along with such procedures. Organizational performance
has a significant relationship with the innovative Human Resource practices and
its cost-effective implementation. Countries such as India are showing growth in the
study of modern management, which covers topics like factors dealing with HR prac-
tices, new trends of recruitment and selection, employee training, and their career
development, etc. Hence, the framework of HR practices and its analysis enhances
the importance of research-based on this subject and reflects economic and technical
circumstances and variables [5].
Although no integrated definition endures for Human Resource Management,
usually scholars tend to decide that it assumes that people should be considered as
recourse and not assets as they are the strength of any company. Walsh discusses
that Human Resources are precious assets because they are exceptional, inimitable,
irreplaceable, and rare. Hence, the company should build appropriate methods of
appointment, deploy, improve, and retain them. In the context of Human Resource
Management, these methods are referred to as the HR bundle practices [6].
In a related vein, Nunn [7] argued that even although certain specific HRM prac-
tices are considered as better to others, a single improved HRM practice not including
other encouraging procedures in the scheme is inadequate to lead viable performance
effects. He added that specific “best practices” ought to be part of a bigger, univer-
sally enhanced HRM system to sustain feasible success. Empirical support tends to
indicate that excellent HRM practices are related to work-connected outcomes, such
as organizational commitment and job satisfaction and organizational citizenship
behavior [8].
19 Impact of Internet of Things and Artificial Intelligence … 243

Job design,
recruitment and
Dynamic selection processes
Environment
Human resource
development plan
Human Strategic
Vision organizational
Resource
outcomes
Strategy Diversity
Strategic
management plan Performance
Business Plan
Productivity
Effectiveness
Key Stakeholders Cost effectiveness
Remuneration and
Human reward systems Profitability
Resource plans
and policies
Achievement of
Performance
business outcomes
Management
Dynamic strategies
Environment

Career
development and
succession plan

Employee relations
strategy

Fig. 19.1 Role of human resource plans and policies in achievement of business outcomes

As shown in Fig. 19.1, the necessary HR practices include appointing, training

and development, performance appraisal, compensation, and flexible work arrange-
ments, which are interrelated functions. Although these methods are distinctive, they
are interconnected. To be successful in the success of the organizational objectives
and goals, they have to be accomplished as a system. As per the Kochar, assessment
of performance encompasses the extent, estimation, and modification of fields of
workers’ accomplishment to certify that their achievements and production provide
adequately to the attainment of managerial targets. In other words, management of
personnel is that element of the administrative practice that focuses in managing
persons in the firm [9]. This entails sequence of choices which affect employers
and workforces. HRM is mainly proposed to control the expediency of personnel
and managers, and it involves numerous regions which has implications for the
global. By managing workers of an organization, Human Resource Management
practioners influence meaningfully, administrative implementation of policies for
cost-proficiency, service or product excellence, distribution, and client gratification.
HRM methods are called successful if consumers and workers are inclined to be
further convinced and by obtaining a satisfactory reputation for himself in the market-
place. In other terms, genuine task for personnel supervisor is to invite, encourage and
maintain great persons [10]. As shown in Fig. 19.2, the structure of administration of
personnel comprises five key movement regions that manage in an overlying field:
preparing; recruitment; workers’ training and growth; reimbursement and, workers
relations. Assessment of employee’s performance arises as a complicated procedure
244 S. Pathak and V. K. Solanki

Fig. 19.2 Human resource management functions

that has impact and significance on every primary occurrence. Kopulos examines the
HRM apprehensions, which includes planning as the primary action. Internal and
external situations (includes government regulations and legislation, sociocultural,
economic, market and unions, demographic, people, global sub-systems, technolog-
ical aspects of the working environment). HRM includes data gathering, and retention
of employees’ information from both internal and external sources is one of the tradi-
tional responsibilities of HR executives. Planning and Preparation for the fulfillment
of policies of the compilation of organizational approaches and practices is a manda-
tory role of HRM. This includes synthesis of work design as well. Exact Human
Resources planning improves and minimizes the ambiguities of environmental risks
and ensure the success of each approach [11].
The Operational actions act as outcomes of planned methodology because of
comprised techniques about confirming suitability to numbers, value chain, supply
chain management, individuals etc. This is for bringing and maintaining social skills
and correct counting, which can assist in the accomplishment of organizational goals.
Human Resource planning consists of preparing for recruitment, screening, and
employment in the organization.
While planning for employment, a human supply manager targets on generating
employment plans through quantifiable analysis, which can predict the requirement
of Human Resources. He also analyzes the numbers, categories, skills of future
employees. This Preparation is a sensitive practice as it directs staffing, appropriate
19 Impact of Internet of Things and Artificial Intelligence … 245

approach, behavior, and this leads to many intrinsic reasons. There can be many long-
lasting repercussions of inappropriate staffing for an organization. It can increase
functional cost and profitability of the organization and regular retraining will be an
additional cost. Whereas, selecting right employees will be a long-term investments
for an organization. Apart from this, experienced personnel can be engaged to achieve
competitive advantages over its competitors. As part of employment agreement with
the association, an employee gets satisfied with various types of economic returns,
real services, and other reimbursement. An employee’s payment for the work done
is a very pervasive and sensitive activity. The Executives have always confronted
the problem of generating results-based amount. These results can either be the
capacity, type of work, skills, involvement of the member and his knowledge etc.
Personnel becomes the strength of any group, industry by generating quality goods
and services or the worker’s negotiation power and customer relation management.
Therefore, optimal wage management of workers is very much required to sustain
the financial stability. The workers’ prospects and demands, their wage proposal
expenses should be balanced out by the Human Resource manager. The workforce
efficiency can be enhanced by tooling and abilities—optimization events, correct
instructions, upgrading projects etc. with a point of view to improving employees’
execution, and consequently the productivity of the organization. Skill training, shape
of training and management development are frequently conducted under the training
and development of employees. The organizations which have market value are
eventually developed by their efforts about development of workforce. This action
has consequences beyond improving workers’ proficiencies parameters, but also to
expand their motivation level. This necessary factor is essential to the fulfillment of
a job, increased efficiency, and long-term dedication for the organization [12].
In any organization, the role of any employee is characterised by set of activi-
ties, designed programs, loyalty, sense of belongingness etc. The responsibility of a
human resource manger is to look after employee’s well being, their safety, fairness
in compensation, discipline, career development, updates of procedures and other
affairs related to employment. He helps in raising employees’ perception, enhance
reputation and group’s value along with the attainment of workers’ career objec-
tives. Through this significant approach, worker’s loyalty for the organization and
achievement of set objectives can be assured. Excellent worker’s affairs procedures
often ensure a reduction in differences among the personnel. Management Execution
requires quantity, appraisal, a modification for a certain range of workers’ performing
for confirming that managerial goals are accomplished by employee’s actions and
production. In the latest years, innovation and various studies have a remarkable
impact on the area of Human Resources, and it is expected to push the extent to
some exceptionally new methods in the future as well. Despite the pervasive adop-
tion of e-HR procedures, there are several issues about whether these new methods
allow companies to attain their HR goals. In addition, the current HR structures have
a number of limitations involving the fact that they (a) utilize one-way interaction
systems, (b) are unfriendly, (c) passive, and (d) may prevent individuals with minimal
computer skills from getting access to careers. Bhardwaj has also pronounces that
these days, revolution in industries and automation of several HR functions has caused
246 S. Pathak and V. K. Solanki

a kind of “outsourcing” for Human Resources [13]. The “automated revolution” will
change the traditional industry into industry 4.0. Most of the e-Human Resource
research has been on e-tutoring and e-hiring with an exceedingly small research on
other HR regions. Secondly, the general take-away from any e-Human Resource
research is that several of the traditional human research also leading into compo-
nents of e-Human Resources. Though the new and innovative know-how are being
used in HR terminologies but the basic goal is still empowering our Human Resource:
e.g., the purpose of employment is to make an effective placement, the goal of variety
is to employ the diverse and most talented employees, and the purpose of training
is to improve the skills and knowledge of the workforce. In spitefulness of these
objectives, most of the research on Human Resources has engaged in approaches
for implementing automated processes and expanding their recognition. Compara-
tively, little investigation has analyzed the extent to which e-Human Resource enables
corporations to achieve their fundamental goals of Human Resource. The movement
toward e-Human Resources is expected to expand in the future, and the expectation
is that the exploration of e-Human Resources will continuously grow.

19.3 Artificial Intelligence

‘Artificial Intelligence’ can be explained as “an ultimate ‘intelligent’ system that

is an adaptable logical agent which recognizes its environment and takes measures
that maximize its chance of accomplishment at some goal.” The work burden of the
employees could be decreased with the help of Artificial Intelligence by using the
latest techniques for data analysis. Artificial Technology promotes the machine-to-
machine dependencies rather than human-to-human or human-to-machine depen-
dency. There are enormous prospects for HR to think about employing AI in their
organizations. There can be some ethical questions as well regarding legal issues and
reactions of employees on management decisions due to the transition from tradi-
tional methods to technology-driven practices. Skill-based industries are gradually
replacing the labor-centered activities into information-based businesses and process
automation. The work pressure of the employees can be reduced, and any process
of HR and other segments can be smoothened through Artificial Intelligence. AI
technology helps in connecting machines with people and tools to devices. Artificial
Intelligence system can be utilized in many HR functions such as—screening of
candidates, engagement of employees, career guidance and development, etc. There
are the number of Artificial Intelligence methods and procedures which are meant
for organizational support to handle organizations with realities of the global work-
place, to understand the impact of international competition and globalization. There
is a definite requirement to shift traditional Human Resource Management practices,
which are no more adequate for the growth of today’s organizations. Eligibility of
employees in any of the industries has been changed from labor into multiskilled and
customer-focused. Artificial intelligent techniques are more focused on the promo-
tion of global workplaces and work culture. We cannot deny the significance of
19 Impact of Internet of Things and Artificial Intelligence … 247

assessed and recognized HRM systems; therefore, to enhance corporate performance,

the combination of the novel paradigm and traditional approach is required [14].
Meyer discussed lots of concepts of Artificial Intelligence (AI), which are limited
strictly to ‘intelligent agents’ that represent ‘cognitive behavior’ are need to be
understood. In this chapter, we include a comprehensive approach to systems
supports expert decision-making systems, modeling and simulation, natural language
processing (NLP), robotics, therefore, we are including augmented, autonomous, and
assisted intelligence in many ways that human are replaced or supported by Arti-
ficial Intelligence [15]. Many technologies have not been observed nowadays as
changed process and have become routinized and are no more ‘cutting edge’. The
cobots (robots collaborating with humans) and the chat boxes (where a computer
program or an AI converses with a customer via speech or text) are challenging
the old process of customer relationship management. The future of HR driven by
Artificial Intelligence can be predicted after 8–10 years based on the trend in the
past of Human Resource practices before 8–10 years. Its significance may differ
in some areas due to process mechanism. It may be just enhancement of current
device and features, incorporation of new functionality in the existing system, or
utterly unique process reengineering. AI building can work for aspirants screening,
engagement of workers, career development, etc. AI, by applying thorough abilities
source and reasonable procedures, generate robust work explanations armed with
appropriate marker subsequent ideal resume match. The organizations can find the
right candidates for the post with the help of AI based tools. These tools acts and
thinks logically to avoid the biasness for the applicant. It can give a competitive
edge over competitors. The contemporary organizations face numerous challenges
that bring new prospects alongside. The experiments faced by the new groups can
be tackled with the help of Artificial Intelligence. Due to tremendous usage of social
media, a huge data has been generated which is called “Big Data” which can be used
carefully to understand latest trends. Big Data analytics for Human Resource is the
process of examining big data in pursuit of trends, models, correlations, and percep-
tions about human behavior. Today’s global market requires expanded attention about
business effectiveness. Artificial Intelligence raises the efficiency of enterprises in
the international market.
The various phases from data gathering and utilizing this huge data into decision-
making by training an AI model or machine is called “AI development cycle” as
shown in Fig. 19.3. Information about this cycle can help professionals of Human
Resources in planning and strategy making by becoming more active. Data curation
is the very first stage of any AI development life cycle. In the growth of any industry,
Artificial Intelligence plays a game changer role. An AI tool can analyze a huge data
with high precision and less time. In many sectors, applications of Artificial Intel-
ligence, professional systems, neural network, simulated reality, natural language
programming, including the Cognitive computing method, has been applied. We
discover in detail the four broad challenges to AI incorporated to HR processes of any
industry. This includes—a complication of HR phenomena, insignificant data, legal
and ethical restrictions, and worker responses to technology-based administration.
248 S. Pathak and V. K. Solanki

Fig. 19.3 Conventional artificial intelligence life cycle

Data Generation is the Difficult phase in the fundamentals of HR processes A

significant cause of difficulty might be the circumstance which is not laidback to
determine data for deciding a “good worker,” or “Inefficient Worker”. The Role neces-
sities are very extensive and method of checking of job consequences is inadequate.
Prejudices connected with measuring personal identity when performance of a whole
team has been analysed based on target achievement. Additionally, multiple roles are
interdependent, along with additional responsibilities. A team’s assessment is gener-
ally inseparable from the individual performance analysis [16]. It is extremely diffi-
cult to describe a good performer as performance assessment is substantially tricky
in the standard procedure of Human Resource Management. In terms of actions, we
can say that, not all functions of HR are restrained; particularly all operations can
not be digitally traced, and all suggestions cannot be applicable and transformed into
a beneficial arrangement at a reasonable cost. For example, in recruitment process,
companies may not track the sources through which candidates come to them—
from recommendations versus visiting website versus job panels, and so forth. Most
companies collect an incomplete amount of data on candidates, and they do not recall
it. This type of selection mechanism limit the kinds of analyses that can be achieved
and the deductions that can be strained. There is no list of variables that companies
choose to finalized and to rearrange their complete HR processes.
Artificial Intelligence is still a prolonged shot of any kind of sphere of Human
Resource movement, but the pace of evolution towards specific technology systems
in the automobile industry, social media and health care, marketing, advertising is
considerable.
19 Impact of Internet of Things and Artificial Intelligence … 249

Adler and Borys explain why the data science society is entirely unconvinced
about causally intelligent AI systems. Randomization is one of the theory that
improves judgments which are based on algorithms. First, converting the responses
to an algorithm is like research and can assist in determining causation. Second,
randomly deciding an HR conclusion with the possibility predicted by a process
where we cannot calculate outcomes with considerable accuracy. This also helps in
recognizing the fundamentally stochastic kind of HR effects and unavoidable mistake
of methods. Formalizing procedures is also required to build appropriate processes. It
ensures the corporations about assumptions constructed into any means, the costs of
developing it, and the possible challenges from workers who are negatively affected
by such tools. As per the procedure, validation should be facilitating other than
intimidating [17].
Line executives should rejuvenate their competence and skill sets. AI should
involve them in an “augmented intelligence,” an updated application of analysis of
staff insights. This helps in the process of management decision-making. The infor-
mation is based on evidence according to the Bayesian approach to methodically
apprising administrative principles for new data [18]. Conflict regarding the ratio-
nality of proficiency, relevance, can occur which disturbs the existing structure. By
leveraging the use of the amount of AI tools, the company can connect the candidate
more successfully before and after they enter the corporation. Fast-paced digitiza-
tion improves in integrating various systems and can deliver a unified platform that
can help a full range of HR functions, starting from enrollment, selection, training,
development, payment, and performance management.

19.3.1 Importance of Artificial Intelligence in Transitioning

Human Resource Management Practices

Merits of using Artificial Intelligence in Human Resource Technology has trans-

formed the current businesses. Computers have now shortened the task of adminis-
tering and investigating large data sets. These tools have confirmed to be an invaluable
aid to the HR executives from the personnel to employee’s self-service. Across the
last 15 years, at least 800,000 jobs in the UK single-handedly have been lost due
to automation, not only in manufacturing but additionally in the front office and
back office, which were consist of routine tasks. HR division has accepted contin-
uous improvements to deliver more quantifiable and tactical value to the corporation.
Lemaignan [19] clearly illustrates that for Artificial Intelligence, smooth interaction
between robots and humans will be a great challenge. Except for emotional intelli-
gence, all other physical work can be done by AI-based technology in a similar way.
Though everyone is talking about AI and innovative technologies, but it also has
many pros and cons. Artificial Intelligence radically transforms Human Resources
Management functions.
250 S. Pathak and V. K. Solanki

Fig. 19.4 Role of AI in human resource management. Source Littler (2018). https://www.racont
eur.net/hr/Artificial-Intelligence-hr-human

Reilly [20] examines that in the future, organizations will face a drastic change
in the way of Human Resource practices with the usage of Artificial Intelligence. To
sustain and persist in the competitive environment, the impact and applications of AI
on Human Resource is a big challenge. The role of Artificial Intelligence has been
continuously increasing in Human Resource Practices. Data Analytics is an integral
domain of AI. Figure 19.4 depicts the intrusion of Artificial Intelligence and Data
Analytics in different HR practices [46].
Human Resource executives will be utilizing AI functionality, such as machine
knowledge and natural language processing in HR functions. The Artificial Intelli-
gence presents a much great prospect to promote the Human Resource world. It will
support the corporations to accomplish their goal in less time. With the growing need
for technology, the company will be needing competent employees who can produce
the machine to achieve the assignment as per the necessity. Artificial Intelligence
will assist the person in managing their employment and personal life effectively.
Staff members will be capable of finishing their job prior to the limit [21].
The need for employees may be decreased in the organisation. AI facilitates the HR
executives to power difficult tasks such as information collection, analyse available
information and to separate employees. It determines the HR delinquent by using the
information to the largest potential. The organisations will appreciate the benefits of
utilizing Artificial Intelligence in the long term. Though investment for AI schemes
will be costly but it will power and will become an asset for the companies. Artificial
Intelligence is about constructing technologies that can believe and comprehend like
a human being.
The benefits of AI-based Human Resources practices comprises of:
19 Impact of Internet of Things and Artificial Intelligence … 251

1. Enabling computer-generated aids via data gathering or dissemination.

2. Expanding worker abilities for full effectiveness in tasks necessitating cross-
preparation skills.
3. Coordinating numbers on various participant requirements.
4. Controlling the flow of data all the way through computerization, supervision
and restructured conditions.
5. Consuming lesser time.
6. Corporations would be capable of concentrating on tactical judgments.
7. Work Precision will be improved.
8. The constraints can be easily conquered.
9. Duties and arrangements of operations can be easily managed.
10. Intelligent card-based systems can be used to discover the schemes.
11. For extreme risk zone, face detection tools can be used by the corporations.
12. To save time and energy of employees, apps which detects personnel numeri-
cally can be utilized by organizations.
13. A manager can sometime be emotionally biased in taking a decision, but an AI
tool does not get biases due to sentiments, emotions and is more competent in
delivering rational decisions.

19.3.2 How Artificial Intelligence is Reinventing Human

Resources

In the process of Human Resource management practices, the crusade of effective-

ness and apprehensions of success cannot be presumed. It can only be developed, and
spontaneity can only be assured after the implementation of AI tools. Application of
Artificial Intelligence can be helpful in the following domains, which can reinvent
the typical HR practices. As depicts in Fig. 19.5, Out of 100% received resumes,
only 0.4% turned out to be a selected candidate after so many phases of assess-
ments. Artificial Intelligence based recruitment tools will be capable of providing
useful projection and proposals that will enhance company competitiveness and
productivity.
There are some of the methods that are enhanced and widened the use of AI
is programmed hiring procedure, recommended training techniques, data-driven
performance reviews, and predictable attrition and turnover which are discussed
as follows:

Fig. 19.5 A typical recruitment process

252 S. Pathak and V. K. Solanki

Hiring and Training of Employees

• Hiring Processes

A recruiter is provided with huge data in comparison to company’s recruitment

approach which has very few candidates. AI based tools which are used in hiring
can reach to large number of aspirants through different resources. Recruiters and
Candidates practice experiences tailored to their particular needs.
• Screening of Candidate

AI can assist the recruitment executives in assessing the applicant rapidly and effec-
tively. Organizations encourage applicants by helping them in the chatbox or AI tool,
responding to common queries about the job reports and specifications. It might also
deliver information in times of feedback about the applications.
• Engagement of Candidate

With the support of AI tools, the company can undoubtedly engage the candidates
using Marketo-like or Hubspot systems. Indeed, AI is able to be incorporated into
these kinds of applicant automation; however, these ideas, responses, and activities
with AI be actual-time and exclusive to the individual need and not just guided by a
position, tags, locations or categories.
• Re-Engagement

Artificial Intelligence tools aids the very beset group of applicants to determine their
attention level for a role, category or location, while by means of AI aids, candidates
gets opportunity to know about new places, skills and work experiences that might
have learned.
• Post-Offer

Once an applicant accepts a job proposal, a dark area exists—usually two weeks after
when a nominee gives notification to when they begin working with the organisation.
AI could provide a differences in this process by engaging and subsequent up with
the nominee.
• New Hire Onboarding

The training program shows an incredibly significant role for the newly appointed
employees. It supports recognizing the corporation processes, culture and policies. AI
can smartly and effectively respond to other shared questions, data, and informations
that may assist the new members of the firm.
• Training of Team
19 Impact of Internet of Things and Artificial Intelligence … 253

Systematic and modern tools can be utilised to train employees. These intelligent
agents/machine can even help individual to do their self skill assessment and under-
stand their weaknesses and coping methodologies. Team executives will be provided
a list of training prospects for the team members.
• Employee Relation and Career Development

It offers assistance to scale a professional development program or business training

for each employee. Employees must have some questions related to assistance
coverage, vacation time, and how they are paid, which need in-depth discussion
with an HR executive or manager. Once information feed in the AI assistance tool,
AI can respond to all inquiries in chat form. Artificial Intelligence knowledge can
be applied in email, chat type, or a simulated meeting room, giving option, even
booking a business meeting between employees and Human Resource manager.
• Scheduling

Whether it is scheduling a work meeting or booking interviews, spending time to

organize availability and schedules is an unproductive task that not only wastes
everybody’s time but retains our email inboxes full and overflowing. Visualize a
world where a machine records meeting, schedules and even arrange food without
human intervention. Human Resource executives and organizational leaders have
faith that amalgamating Artificial Intelligence (AI) into HR purposes like adminis-
tration and onboarding of payments will definitely enhance the overall experience
and knowledge of employees.
Other Examples of Applications of Artificial Intelligence (AI) in the Workplace
• Personalized Employee Experiences

IBM officials in their study have discussed that how AI could successfully be
constructed into an employee’s onboarding program. New employees who typically
want to connect people and develop relations and requires a lot information about
organisation. They may request to their counter neighbor colleague. But what if
she/he performs in a different division. What if the department head welcomed the
newly hired employees on his mobile device that was customized to his first task? IBM
has created a AI tool that will respond to a new employee’s about their job-related
questions to aid them up. For example, an AI could deliver training recommendations
or offer the locations, names, and contact info for individuals he/she should look to
get along with on his first day or so [14].
– Vacation Requests Control Mechanism—Such tools help in getting information
about total number of vacation application for a certain period of time and help
employees in scheduling and getting approval for their slots accordingly.
– Mood Control Mechanism—If an executive takes a call from the client. After the
conversation, the executive receives feedback about his discussion with the client
so that he can decide to take a break before the next meeting before exhaustion.
254 S. Pathak and V. K. Solanki

• Cognitive-Supporting Decision-Making

IBM is using a cognitive AI tool such as IBM Watson, which assist employees in
making essential day-to-day judgements in their workplace and are necessitates for
daily routine activities. Such tool help them by offering suggested solution to a
problem and help employees in such cognitive decisions. Usually, such task are to
be managed by HR team members.
• Analytics of Smarter People

Organisations have to maintain a database about their employees to gain perceptions

to foresee their future performances. HR team members have to gather and maintain
a lot of data to go for individuals’ analytics. AI plays a significant role in deciding
about which kind of data can be useful to pursue, analyze and protect for a Human
Resource database. With the help of such tools, organization can take decision about
ability, can find inventive ways to bring talent together.
• Removing Biases

As per the survey by the Human Resources researchers, Loftus et al. found that even
when corporations strive to be comprehensive, they may unknowingly tend toward
applicants who are similar to them, which is called “unconscious bias.” Another
bias, language bias, has been revealed by a psychological means called the Implicit
Association Test (IAT) that shows that individuals’ subconscious word companies
indicate discrimination. These biases discover their way into job explanations, as
well as resume choices. AI algorithms can be deliberate to help managers to iden-
tify and eliminate these prejudice patterns in the language they utilize to improve
their employing networks and welcome different applicants, and AI could also offer
executives with candidates who might have been examined out owing to human
inclination to choose candidates with comparable competencies and traits [22].
• Identifying Employee

AI programs are designed to identify workers that may be planning to exit. It works by
following workers computer activity—internet browsing, emails, keystrokes, etc. and
store this data for one month. AI method examines the data to establish a baseline
of typical activity models in the organization. It flags outliers and informs higher
authority about identified changes. Bases on modern AI based HR tools employees’
relations can also be judges and correct measures can be taken before time.
In the US economy, roughly 60% of all expenditure is on Human Resources. In
service businesses, the data is much higher. Processes constitute the trend of concern,
such as how an association hires workers. One of the justifications for the importance
of applying statistics science tools to labor is because employee performs so many
jobs and a considerable amount of money is spent on them. Most of the operations
and functions of HR are repetitive with similar mechanism and procedures. Each
of these procedures includes organizational routine tasks, way of performance, and
19 Impact of Internet of Things and Artificial Intelligence … 255

each comprises specific job roles, written guidelines, and instructions to execute and
activities for all parties. These repetitive processes produce quantified data, in the
shape of texts, recordings etc. As organisations move to the simulated space, many
of these task will be based on “numerical dissipate” which is established data on
numerical activities (e.g., skill assessment and online job applications) that may be
utilized to build enrolling algorithms [23].
Human Resource information systems, candidates’ tracking methods, digital drain
etc. are all critical inputs for the data production. Usually, this input must be obtained
from numerous databases, transferred to a standard format, and enlisted together
before an assessment can take place. By “machine learning” (ML) we describe to a
comprehensive set of methods that can adjust and learn from information to create
algorithms that play better for any assignment or to predict new assignments and
assess requirement of changes. The most interactive application of machine learning
skills has been “conducted” in which a data researcher creates a machine learning
algorithm, defines the most suitable metric to evaluate its accuracy, and focuses on
the procedures.
• Artificial Intelligence in Learning and Advancement

Employees can get AI facilitated personalized points to improve their skill sets
through execution surveys. This platform helps employees in figuring out the skills,
new aptitudes they should learn at their work front. It is now feasible with the help of
cutting-edge technology to have an individual’s personalized experiential learning.
Artificial Intelligence tools can also help in structuring the skill chart of staff and can
provide data about which employees need improvement advancements before they
become out of date in the industry.
Dixon discussed that guide inducements and other tools are used as “nudging”
programs for workers and suppliers. IBM uses algorithms to recommend employees
on the bases of knowledge of other employees at the same level. Among the most
applicable prediction algorithms, such as random foster and logistic regression help
in conclusion from statistical calculations such as correlations, regression, time series
etc. These AI tools can also be utilized in scoring characteristics of employees related
to job performance and make use of that to skills enhancements schemes in the
future. “Algorithmic Management” the method of procedure based on algorithm
can be used to focus on employees need just similar to Netflix proposes content
based on customer preferences or Amazon suggests products based on browsing
their purchasing behaviour [24].

19.4 Comparison Between Traditional HRM and AI-Based

HRM

In the contemporary global context, Human Resource Management is ever-changing

continuously and adequately conquer and coping with existing and new challenges of
256 S. Pathak and V. K. Solanki

the changed environment. In Corporations, very few HR practices match the concept
of developing people in a new way. Some key factors are as follows:
• Accountability of Line managers have been increased
• Requirement of Cognitive Skills
• Purpose of HRM has been extended as a business partner
• Language modification
• Elastic building
• Adaptable and multiskilled teams
• Client-focused [25].
As shown in Fig. 19.6, we can see the new roles and rules for Human Resources.
The demand for a shift to favor working methods has forced the expansion of HRM
abilities to line executives throughout the companies. The changes promote scope
of management of Human Resources to a distinct, diverse methods. As a new role,
line managers got the responsibility to assess employee performance, their new role,
new mind set, acceptance for the changes, training about new changes etc. as they
are in the direct contact of employees. These new HR role are enlisted and discussed
below:
1. HR executives direct the line managers about their added responsibilities. From
this authority distribution to line managers, the HR manager can utilize their time
for new planning and innovating other HR aspects.
2. Human Resource management work has been expanded as a business asso-
ciate with this modified responsibility of updated Human Resource. With the
distributed authority to line managers, the HR procedure has also been rede-
fined. The decision-making of Human Resource executives is necessarily paired
with allowing the free flow process for different methods Viz. Formation of new
groups.

Fig. 19.6 New rules and roles for human resource

19 Impact of Internet of Things and Artificial Intelligence … 257

3. The shift in the organizational vocabulary in new direction and orientation is a

pervasive tool. To develop personal writing and bring change in the terminology
of critical businesses, the difference in the infrastructure and structure of the orga-
nization is required. In modern terminology—the word supervisor, supervisors
have been replaced by names such as coach, Leader.
4. New organizations are changing with the acknowledgment that consumers,
whether internal or external, are the backbone of a business, and therefore client’s
satisfaction and their needs are on top priority. Based on this new philosophy,
the emphasis has been redirected from supervisors to the consumers. In other
terms, the modern philosophy emphasized on worker’s job satisfaction, and more
attention is on consumer satisfaction and care.
5. The modern organization was looking to develop a workforce that is multi-
talented, which is flexible and adaptable, who can perform different responsi-
bilities. This philosophy affects the planning policy, recruitment and selection
policy, and hiring the person who could be easily re-skilled in a different area
of competences and skills. The AI based tools which works on algorithms have
made differences in some critical ways from conventional approaches used in the
Human Resource department. The continuous study on generating new norms of
employment, is reinventing the HR processes in today’s industrial scenario [26].
Organizations now are facing complex workforce challenges at higher rates.
Hiking expectations of the employer combined with the visualization of the work-
force equipped with the novel skill set, technology driven and unique of it’s on to fit
in the job perfectly. The HR plays a crucial role in addressing these challenges. The
upcoming technologies or the updating in the earlier one are becoming the savior
of the HR to take the challenges of fitting in the employers expectations. With the
advancement of the cloud computing, data decision matrix and Internet of the Things
the life of a HR became so easy and another saviors’ in the pipe line is the Cogni-
tive computing emerging to help the business outcomes by expanding the expertise
of the Human and improving its decision making. The digital age is bringing the
opportunity, challenges and trends to impact the HR functions of the organizations
across the globe. Machine learning, in contrast, produces one algorithm that creates
the use of several variables [27]. Indeed, one of the benefits of Machine Learning
is its inquiry of non-conventional factors because the objective is to build a better
forecast rather than improving the theory of the ground in which the investigator is
based by giving testimony on theories. “Judgment-making,” the final phase, is very
helpful in transactions with the way in which we use perceptions from the machine
learning model in daily operations.
The following are the few software which is being used in HR with the help of
automation algorithms of Artificial Intelligence.
Talla—This is an auto platform that is based on AI, which helps in supporting
customers. This answers the queries automatically and generate report whenever
required and help in the managerial decision-making process.
258 S. Pathak and V. K. Solanki

My Interview—This platform is cloud-based, which brings employees and candi-

dates together for the interview process. It helps in the pre-screening of candi-
dates as well. This platform has features of managing applications of candidates
and customized questions and reviewing and to provide comments. This helps in
enhancing effectiveness of recruitment process. This ensures the personality of the
candidates in front of employer and help in selecting best fit candidate.
Hemingway App—This App is used in increasing readability and saving time. It is
based on writing by NLP and improves writing structure by using primitive Artificial
Intelligence.
Skype Translator—This text translator in more than 500 languages enhanced
prompt messaging and diminished the barrier to different styles as it can work in
more than eight languages.
Saberr—This platform helps individuals in receiving reports on Big 5 Personality
and Schwartz, which will further help to increase the self-awareness about how to
work and act at work. It helps in creating a relationship between technology and
people.
Google Smart Reply—This is a tool that helps employees in responses to the mails
and updation of mails and their smart answers. Machine learning suggests a quick
reaction after analyzing the emails.
Paradox—It helps in searching the candidates with the help of VCV. This is a video
interview which calls candidates with questions with the help of voice recognition
and more focusses on candidate management.
Clarke AI—This helps in managing conference calls through Artificial Intelligence
Bot and does nothing during the entire meeting and email conference notes after
completion of the conference directly to the email box.
Therefore we can say that in the war of talent, Artificial Intelligence is making its
space effectively starting from recruitment and selection to complete the employee
life cycle along with personalized experience, career designing, and succession plan-
ning. Managers today usually have the possibility of disregarding evidence about
expectations, using it as they see fit, and creating their own information about actions
like employment in the form of conversations [28].

19.5 Implications of Artificial Intelligence in Human

Resources Practices

There is a massive eruption of machinery-technology and new vendors in the market-

place for HR specialists to think about and execute to help with their use issues.
However, the big question is about the readiness of Human Resources. Based on a
review study, researchers found that nearly 80% of HR experts felt that they might
19 Impact of Internet of Things and Artificial Intelligence … 259

possibly see a business world where chatbots would be responding to employee

queries but our human resource and management may not be ready for this transition
due to many factors such as excessive finance and fear of joblessness, etc. That is
probably alarming when you see how we are developing more relaxed with chatbots
in our employees ands customer lives.
However, study also reports that only 46% felt happy with the concept of Artificial
Intelligence That is slightly disappointing but perhaps, reflects how willing HR is
to accept these machinery-based services today. The study also suggests that regret-
tably, there is a massive number of HR specialists who believe that they are still not
experienced in the vicinity of AI. This represents us with a core area for enhancing
skilled HR to realize and accept power of this kind of technology and automation.
As Machine Learning and AI are playing with the technological landscape of the
HR, it is becoming very essential for the HR administrator to make their employees
ready to accept changes and to bring equilibrium in between the human space and
technological advancements. The implementing technology in to a existing system
is mire difficult than adopting a new technology. Figure 19.7 depicts some of the
pros and cons of Artificial Intelligence in the present contexts. If an organization
has observed to get started and understand how automation, Artificial Intelligence,
augmentation, and the new prospect of work are influencing the HR function, then
AI can be implemented successfully in Human Resource practices. The aims should
bring to deliver with a strong foundation of the growth of new methodologies incor-
porating machine learning and AI means to HR. This new approach will become
much more assured in understanding the impact of AI on Human Resources and get
ready for coming future of organisations.

Fig. 19.7 Pros and cons of artificial intelligence. Source Data Fair
260 S. Pathak and V. K. Solanki

19.6 Limitations of Artificial Intelligence Based HR Tools

We have discussed in this chapter that better productivity can be foreseen through
Artificial Intelligence in comparison to humans. At present, AI, machine learning,
NLP are emerging technologies, and around 1 lakh new roles will be conquered the
traditional HR practices by the year 2036. A few organizations have foreseen the
future and undergoing into a transition phase and faced few challenges which are
discussed below:

19.6.1 Challenges

1. Various matrices are used to categorize ‘a good employee,’ but this is a very
complex phenomenon. One of them is the performance appraisal score, which has
validity issues, and it shows biasness. Therefore, many organizations have to stop
using it [29]. Drawback of performance appraisal score is that does not differen-
tiate team performance and individual performance because of interdependencies
[30].
2. Small data sets is another constraint. The performance of an employee cannot
be declared as weak due to less data availability as most of the organization has
its employee count on thousand instead of millions, so the “BIG DATA” tools
analysis is not applicable in all organizations.
3. At the time of hiring or firing of employees, procedural and distributive justice,
and an approach needs to be followed. Measurement of socio-psychological
concerns should be taken care of as well [14]. Based on gender biases shown by
an algorithm of a recruiting tool, Amazon had to put down the AI recruiting tool
[31].
4. Meskó et al. suggested that organizational outcomes might be affected by bluffing
with an algorithm of AI-based tools as an adverse reaction of employees. Under
a monitored environment, masking of one’s actual capabilities is a universal
human psychological phenomenon. Confirmation of real characteristics must be
captured through spontaneity [32].

19.6.2 Solutions

1. The first challenge is the prediction of an overly sensitive issue of declaring

a “Good Employee” based on metrics used by the AI application tool and the
environment. As an example of metric used by Tavis and Cappelli (2017) is
performance appraisal scores, which cannot be a determinant of the decision of
“good employee” but maybe a reward or cause. Apart from this, the construct can
be created, and metrics can be identified, to provide a more precise notion about
19 Impact of Internet of Things and Artificial Intelligence … 261

‘a good employee’. For instance: five metrics can be used combinedly, which
can be efficiently used for the decision. These five indices can be as follow:
• Employee Productivity Index
• Training Effectiveness Index
• Absence Rate
• Employee Happiness Index
• Employee Ethics Index [33].
2. An outsized impact can be achieved through précised data powered by AI [34].
An example of the United States Postal Service (USPS) is worth to be discussed
here. Sorting of mails has been automated by using optical character recognition
technology. Almost 98% of the mails can be read with precision without human
intervention. Thus, the following steps can be taken to make the expected result
for an organization
(a) Meaningful results can be generated by first setting objectives and goals with
the personnel who tie back to the aim of the business and handling cross-
functional assignments. The nature of AI is prescriptive. Therefore, signif-
icant results will come if the data set of organization and cross-functional
teams work accordingly.
(b) Another important step is taming the data confusion, which will help in
filtering the data by eliminating the noise. The stages of employee’s lifecycle
attached systems and impact on business should be focused on while devel-
oping a framework for actionable mining values and taming the disorder of
data [34]. The utmost exactitude is required for the selection of the right tech-
nology. For a growing ecosystem, machine learning and AI offers a gamut
of tools which is quintessential. In this transformation, Top management
plays an important role [35–37].
3. In 2017, Amazon used an AI-based hiring tool, but the results were “gender
Bias” [38]. The Amazon model was trained over ten years to verify applicants,
and by chance, mostly men’s applications were observed, and this leads to gender
dominance [39]. However, these shortcomings were overcome with the help of
newer technologies as discussed in Table 19.1.

4. As per the study, LinkedIn considers that there can be a slight change in capturing
patterns to make video interviews a prominent tool. The pre-recorded questions
can be replaced by real-time, and the candidate will answer each through the
camera instead of moving to the next question manually in the whole process. A
bias-free interview can be conducted by not giving much time to forge expres-
sions. Interview transcript, tone, body language, and word choice are the few
parameters on which results of a video interview are derived. This technology
uses Facial Expressions and Natural Language Processing [43].

There are many other HR technologies as well, which can be utilized to manage
human recourses. The functions of these technologies need to figure out in HR spaces.
262 S. Pathak and V. K. Solanki

Some of them are Google as a personal assistant, SOAR, Free HAL, Cyc—currently
being used by Cleveland Clinic, Open Mind Common Sense. AI implementation
in HRM has lacked a clear framework yet. A quite enchanting current practice of
AI techniques in HRM has been observed. Innovations are regularly coming by
overcoming the challenges and problems of Ai in HRM. A recent report by Deloitte

Table 19.1 Description of technologies used in AI based recruitment process

Sr. No Technology Description
1. Knockri Analyses of the video in which pre-recorded
questions have been answered by a candidate are
done by this platform. Candidate’s gender, skin
color, ethnicity does not take by Knockri. The
four parameters are the basis of scores provided
by algorithm, namely, collaboration,
communication skills, enthusiasm, and empathy.
Based on scoring, suitable candidates are
selected and sent to HR recruiter. This platform
integrates with SAP—success factors, the client
company’s mobile Application Tracking
Platform (ATS), Kenexa-Brassing, etc. As per
the data by company websites, a 62% decrease
in cost to fill has been observed. Screen time has
decreased to 68%, and diversified shortlists are
increased to 24%. Unilever has reported that
90% of hiring time has been reduced, and
diversity in hiring has been increased to 16%,
and total saving is £1 million [40]
2. HireVue This is a game-based assessment that is more
engaging and fun for candidates. This is also a
video-hiring tool. This challenge is posed to the
candidate who is game-based video questions.
Around 250,000 data points are created through
this, which is further analyzed on audio, speech
patterns, and tonality points. At HireVue,
Industrial—psychologists and data scientists
work together by using standards that are
professional, legal, and validated—based on
assessment results, hiring decisions taken by
taking care that it should not impact on protected
groups. An assessment algorithm is created by
the team to remove data that affects negatively
on assessment’s predictive accuracy [41]. In
2015, HireVue’s interview and platform was
used by Atlanta Public Schools in their
recruiting process in all 80 schools. As per the
data, the percentage of filling all principal
vacancies was 100% on the very first day, and
that year 66% reduction was observed in student
criminal offenses
(continued)
19 Impact of Internet of Things and Artificial Intelligence … 263

Table 19.1 (continued)

Sr. No Technology Description
3. Humantic (earlier Deeper Sense) This platform analyses the applicants’ social
media instead of the responses of video
performance. The Frrole developed a
Humanistic tool based on the candidate’s public
information besides intruding into their personal
data. Dr. Tom Janz (I/O psychologist, Founder:
The Talent Analytics Group) has invented 7
Deeper Sense personality traits, which has
negative and positive correlation was the
parameters of the AI tool—Humanistic [11, 42]

has quoted well that industry must understand Gen Y—millennials to know more
about the future of HR [44, 45].
In administrative terminology, maintenance, creation of comparatively flat classi-
fied assemblies, enables well-organized procedure for administrations, in relations of
organizational-time possessions, which decrease managerial levels. For this situation,
smooth edifices enable an earlier association between structural associates to erad-
icate inflexibilities in their communications. Close interaction among employees is
of immense worth—which in turn stipulates slender and flat administrative construc-
tion. Occasions are not provided by the executive team members for mutual knowl-
edge but also influential in subsiding mechanical barriers. Additionally, they deliver
associates and prospects to obtain additional information or abilities because they
transfer from one mission to another. Incorporation of reward policy in which an
employee can be shifted to another position and location will help in increasing
the high level of elasticity and help employees for growth and career development.
Emerging multi-talented personnel can be developed through this approach. More
emphasis should be given on developing people-oriented administration and culture
instead of competitive administration. A pleased and motivated staff can only be
generated through HR policy which is for well-being and welfare of their personnel.
Customer gratification can be attained through a satisfied labor force. The main
Human Resource administration task is to generate such a work environment in which
constant development and improvement of employees can be ensured along with the
development of a knowledge-based organization. As a method of enabling higher
incentives, workers aids should be documented and properly rewarded whenever any
worker showcases a higher level of performance. Training modules should not be non-
experimental and theoretical. Further analysis should be done if existing methodolo-
gies and practices are not matching with the proposed expectations from the planned
modules. Other methods can opt if schemes are not flexible and recommendatory in
nature and fit the organizational culture.
264 S. Pathak and V. K. Solanki

19.7 Conclusion

Organizational accomplishment can be improved by incorporating AI-based Human

Resource Management practices. Though the cognitive and emotional qualities of
humans can not be comparable to AI-based products, in comparison to traditional HR
methods, AI-based HR functions are more efficient in reference to data maintenance,
diagnose issues, accurate analyses, and timely prediction about future prediction. As
we are moving towards industry 4.0, in which IOT, Artificial Intelligence, automated
procedures are going to overpower all other organizational aspects, therefore, one
should be ready to adopt new trends of the fourth revolution of industry. This new era
can affect employment by job reducing across a few sectors but its responsibility of
the HR leaders to train and prepare their employees for adopting and excel in these
new technologies. Many existing companies have integrated AI-based devices in HR
systems, be it recruitment, staffing, on-the-job training, performance, and potential
assessments and especially important factor-retention of skilled employees, etc. If
AI is employed in an appropriate way in HR systems, it enhances better opportu-
nities. AI-based HR practices should be considered as enthusiastic prospects. HR
team members require to be stabilized their processes as per the technical and intel-
lectual developments along with the precision because Human Resource technology
panorama is continuously affected by Artificial Intelligence. Managers and experts
must have a real insight into judgments that are rendered to lessen unintention-
ally, introducing prejudice about systems. Candidness will be crucial in becoming
convinced about personnel faith in new expertise. The order can exceed the individual
point of ability, intelligence. Human Resource database will enhance database that
discovers about events. The AI scheme information processor acquires a successful
path of accomplishment of the purpose of any task. This technological improve-
ment will support the executives to handle the procedure, employees, and strategies
successfully. This will be helpful for both employer and the employee. Companies
are no further reliant about the human power; employees can make their job done
around the system efficiently. Though it might decrease need of Human Resource but
nonetheless managers would need cost-effective administrators, trained personnel,
and qualified workers to engage in along with the procedure. This would support
people for lessening the document efforts; a long method would no prolonged drawn
out. The anxiety of the workers will be decreased because quickly, assessment can be
done based on personnel data. The new technical AI structure might be affected by
external and internal forces. Issues based on internal factors viz: workers’ struggle
to understand roughly new workforces, the unattainability of funds, etc. The compo-
nent which may alter the method, such as conversion in the legal area, may have
various governmental principles, through effects on the economy, nation’s economy,
new legislations, employment contract, etc. The equipment has altered the system of
the business.
Acknowledgment for this change in the business methodologies has gone to
the internet, programmers, workstations, and the processors. Globe has become
compact, and the forthcoming group would utilize improved AI-based tools. We can
19 Impact of Internet of Things and Artificial Intelligence … 265

conclude that employee productivity can be raised through AI-based HR applications.

Employee outcomes and needs can be precisely focused by analyzing, predicting,
diagnose, and becoming become a more capable resource. Challenges of the talent
gap, privacy, integration capabilities, maintenance, and limited successful applica-
tions can be there as a subsidiary of AI-based Human Resource Management. But
unnecessary issues can be easily solved by deploying services based on Artificial
Intelligence. HR Systems based on AI should be carefully managed by obtaining
reliable data sets, finding the correct approach, avoiding biases, and looking for
clarity

19.8 Future Scope

The future vacant positions can be correctly predicted by using succession planning
if it will be based on Artificial intelligence. But a lot of studies is required to develop
a tool that can accurately analyze candidate’s applications and help in taking the
decision about the right candidate. Using Artificial Intelligence such as online quizzes
and using fast games by using an intranet can help in employee’s engagement in a
better way. Apart from this, AI can also help HR professionals in conducting exit
interviews through video and audio recordings of those candidates who have already
left or going to leave.

References

1. Dessler, G.: Human Resource Management, 11th edn. Prentice-Hall, Englewood Cliffs (2007)
2. Stone, D.L.: The influence of technology on the future of human resource management. Hum.
Resour. Manag. Rev. 25, 216–231 (2015)
3. Rousseau, D.: The Oxford Handbook of Evidence-Based Management. Oxford University
Press (2014)
4. Strohmeier, S., Piazza, F.: Artificial intelligence techniques in human resource management—
a conceptual exploration. In: Intelligent Techniques in Engineering Management, vol. 87,
pp. 149–172 (2015)
5. Bratton, J., Gold, J.: Human Resource Management: Theory and Practice. MacMillan Business,
Hampshire, UK (1999)
6. Walsh, D.: Employment Law for Human Resource Practice. South-Western, Mason, Ohio
(2017)
7. Nunn, J.: Emerging Impact of AI on HR. Forbes Technology Council (2019)
8. Smyth, R.: Exploring the usefulness of a conceptual framework as a research tool: a researcher’s
reflections. Issues Educ. Res. 14, 43–51 (2004)
9. Kochar, R.: Now Artificial Intelligence Will Find the Right Job For You. Available at www.ent
repreneur.com (2016)
10. Oyesola: Managing health and safety in workplace: employee’s perspective, pp. 31–33. Daily
Sun (2011)
11. Kopulos, A.R.: What Does Artificial Intelligence AI Mean for HR? Available at www.employ
eeconnect.com (2016)
266 S. Pathak and V. K. Solanki

12. Lee, M.K., Kusbit, D., Metsky, E., Dabbish, L.: Working with machines: the impact of algo-
rithmic, data-driven management on human workers. In: Proceedings of the 33rd Annual ACM
SIGCHI Conference, pp. 1603–1612. ACM Press, New York, NY (2015)
13. Bhardwaj, R.: How AI is Revolutionizing the Human Resource Functions (2019)
14. Anon: The Business Case for AI in HR. Available at https://www.ibm.com/talent-management/
ai-in-hr-business-case/ (2019)
15. Meyer, D.: Amazon Reportedly Killed an AI Recruitment System Because It Couldn’t Stop
the Tool from Discriminating Against Women. Fortune. Available at https://fortune.com/2018/
10/10/amazon-ai-recruitment-bias-women-sexist/ (2018)
16. Pfeffer, S.: Hard facts, dangerous half-truths, and total nonsense profiting from evidence-based
management. Strateg. Leadersh. 34(2), 35–42 (2006)
17. Adler, P., Borys, B.: Two types of bureaucracy: enabling and coercive. Adm. Sci. Q. 41, 61–89
(1996)
18. Barends, E., Rousseau, D.: Evidence-Based Management: How to Use Evidence to Make Better
Organizational Decisions, pp 21–30. Kogan (2018)
19. Lemaignan, S.: Artificial cognition for social human-robot interaction (2017)
20. Reilly, P.: The Impact of AI on HR Function. Institute of Employment Studies (IES) (2018)
21. Netessine, S., Yakubovich, V.: The Darwinian workplace. Harv. Bus. Rev. 90(5), 25–28 (2012)
22. Loftus, R.J., Russel, C., Kusner, M., Silva, R.: Causal reasoning for algorithmic fairness (2017).
Available at https://doi.org/10.1111/phc3.12470
23. Acktar, R., Winsborough, D., Ort, U., Johnson, A., Premuzic, T.C.: Detecting the dark side of
personality using social media. Pers. Individ. Differ. 132, 90–97 (2018)
24. Dixon, N.M.: Helping US HR professionals into action learning. In: Action Learning in
Practice, 2nd edn. Aldershot (1991)
25. Geetha, R., Reddy, B.: Recruitment through artificial intelligence: a conceptual study. Int. J.
Mech. Eng. Technol. 9(7), 63–70 (2018)
26. Lind, E.A., Van den Bos, K.: When fairness works: toward a general theory of uncertainty
management. Res. Organ. Behav. 24, 181–223 (2002)
27. Ahmed, O.: Artificial intelligence in HR. IJRAR 5, 971–978 (2018)
28. Bloomberg, J.: Don’t Trust Artificial Intelligence? Time to Open the AI Black Box.
Forbes. Available at https://www.forbes.com/sites/jasonbloomberg/2018/09/16/dont-trust-art
ificial (2018)
29. Cappelli, P., Tambe, P., Yakubovich, V.: Artificial intelligence in human resources management:
challenges and a path forward. SSRN Electron. J. (2018)
30. Jia, Q., Guo, Y., Li, R., Li, Y., Chen, Y.: A conceptual artificial intelligence application frame-
work in human resource management. In: ICEB Proceedings 91. Available at https://aisel.ais
net.org/iceb2018/91 (2018)
31. Dastin, J.: Amazon Scraps Secret AI Recruiting Tool That Showed Bias Against Women. Avail-
able at https://www.reuters.com/article/us-amazon-com-jobs-automationinsight/amazon-scr
aps-secret-ai-recruiting-tool-that-showed-biasagainst-women (2019)
32. Pfeffer, R.S.: Evidence-based management. Harv. Bus. Rev. Available at https://hbr.org/2006/
01/evidence-based-management (2019)
33. New Age: Artificial Intelligence for Human Resource Opportunities and Functions. Ernst &
Young LLP (2018)
34. Saklani, P.: Sometimes “small data” is enough to create smart products. Harv. Bus. Rev.
Available at https://hbr.org/2017/07/sometimes-small-data-is-enough-to-createsmart-products
(2019)
35. Dietvorst, B.J., Simmons, J.P., Massey, C.: Overcoming algorithm aversion: people will use
imperfect algorithms if they can (even slightly) modify them. Manage. Sci. 64(3), 1155–1170
(2016)
36. Murgai: Role of artificial intelligence in transforming human resource management. IJTSRD
2(3), 877–881 (2019)
37. Webb, P.: Managing today’s workforce to meet tomorrow’s challenges. Front. Health Serv.
Manag. 35(4), 3–10 (2019)
19 Impact of Internet of Things and Artificial Intelligence … 267

38. Insight—Amazon Scraps Secret AI Recruiting Tool That Showed Bias. Avail-
able at https://in.reuters.com/article/amazon-com-jobs-automation/insightamazon-scraps-sec
ret-ai-recruiting-tool-that-showed-bias-againstwomen-idINKCN1MK0 (2019)
39. Integrity Management & Consulting Group: Man vs. Machine in the HR World—The Conclu-
sions. Available at https://www.integritymcg.com/blog/hiring-machines-decidinghuman-fates-
pros-cons-of-using-ai-in-recruitment (2019)
40. Knockri—AI Video Recruiting I Assess People and Skills I Not Just Shortlist Resumes.
Available at https://www.knockri.com/#results (2019)
41. HireVue Case Studies | HireVue. Available at https://www.hirevue.com/customers/case-studies
(2019)
42. Meskó, B., Hetényi, G., Győrffy, Z.: Will artificial intelligence solve the human resource crisis
in healthcare? BMC Health Serv. Res. 18(1) (2018)
43. LinkedIn Releases Global Talent Trends Report. Available at https://news.linkedin.com/2019/
January/linkedin-releases-2019-global-talent-trends-report (2019)
44. Challenges in Managing the Gen Y Workforce | Forbes India Blog. Available at https://www.
forbesindia.com/blog/business-strategy/challenges-inmanaging-the-gen-y-workforce/ (2019)
45. Global Human Capital Trends | Deloitte India | Human Capital: Available at https://www2.del
oitte.com/in/en/pages/human-capital/articles/hctrends-2019.html (2019)
46. Aaron C.: How Can Artificial Intelligence Work for HR? Available at https://www.littler.com/
publication-press/press/how-can-artificial-intelligence-work-hr(May,2018)
Chapter 20
IoT Based Intelligent Ambulance
Monitoring and Traffic Control System

J. Jijin Godwin, B. V. Santhosh Krishna, R. Rajeshwari, P. Sushmitha,

and M. Yamini

Abstract The evolving problem of traffic problems is an inevitable situation in wide

and rising cities worldwide. Population growth has led to a growth in the number
of cars leading to absolute congestion development. Life is sacred and can not be
repatriated until lost. The reaction time required by the emergency responders plays a
vital role in the face of disasters. This might be a department for medical, disaster, or
defense. The greatest challenge they face is congestion from flow. To address this, an
intelligent traffic system is needed using a preemptive equation that adapts rapidly
to environmental conditions. The aim of this article is to identify the emergency
call headed back to the location and to monitor the congestion system in order to
provide efficient facilities. This journal also sets out a method that uses a Zigbee
component and Internet of Things (IoT) to transmit the treatment request from the
ambulance to the nearby hospitals, while ambulance attaining the road junction, the
smart traffic system which in turn changes the traffic signal cycle. This system can
be implemented throughout the city thereby reducing the delay.

Keywords Preemption algorithm · Traffic congestion · IoT Zigbee module · Smart

traffic control system

J. J. Godwin · R. Rajeshwari · P. Sushmitha · M. Yamini

Department of Electronics and Communication Engineering, Velammal Institute of Technology,
Chennai, Tamil Nadu, India
e-mail: [email protected]
R. Rajeshwari
e-mail: [email protected]
B. V. S. Krishna (B)
Senior Assistant Professor, New Horizon College of Engineering, Bangalore, Karnataka, India
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 269

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_20
270 J. J. Godwin et al.

20.1 Introduction

Traffic congestion reduction is the main goal in the development of efficient traffic
control system. Many research works focus in the field of traffic safety control system.
In addition, giving priority to the vehicle and changing the traffic signals must be the
important functions for all the emergency vehicles to develop the safety measures in
the road transportation [1]. Ambulance and police vehicles are the emergency vehi-
cles, must be reach the location without a maximum delay. The principle behind the
preemption is that an emergency vehicle is detected by the sensor at each intersec-
tion, when a RFID analyses an emergency vehicle, it will transmit the control signal
to change the traffic lights turn into green. The advanced development in electronics
and communication systems provides different traffic control techniques [2–4].
Traffic control system using Preemption algorithm is designed for giving green
light for emergency vehicles and red signal for other vehicles. A centralized mecha-
nism is used for monitoring the traffic control mechanism to clear the traffic conges-
tion and give information about the shortest path for the ambulance services [4].
One of the most efficient traffic control system for freeway system is ramp metering
[5]. This paper focuses on traffic control system using preemption algorithm with
Radio Frequency Identification (RFID) technology. Another technology (Zigbee
module and IoT) is that the conditions of the patient will be pre intimated to all
the nearby hospitals, the hospital which is ready to take care of the patient will give
an acknowledgement to the nurse present along with the patient in the ambulance
[6, 7].
IoT may be implemented in internet-connected embedded applications which
interact with one another, goods and services. Most of the monitoring systems that
showcasing the status of the patient are frequently in emergency medicine systems
like operation theatres. But the doctor cannot be alerted all the time where there is
an emergency [8]. Although 24 h monitoring by the doctors are not possible and the
details will not be exchanged with other physicians and relatives who are experts in
that area. Many technologies that enable both these things are accessible and many
people in developed countries can’t view and manage them [9]. Hence, the threats
will be swept off by using IoT.
Health monitoring system in the ambulance monitoring the different health param-
eters and these data will be sent to IoT for further analysis [10, 11]. Different sensors
are used for measuring those parameters. In the proposed system, various health
parameters like Electrocardiogram (ECG), temperature and heart beat rate are moni-
tored and transmitted through wireless technology Zigbee. The details about the
different health parameters are collected and transmitted to Zigbee and IoT through
Arduino Uno [12].
Acknowledgements from the hospital is received by Zigbee module via Arduino
Uno. The IoT architecture functions are served to the clients using uninterrupted web
service [13]. The physical connections are interconnected with various sensors like
IR sensor, Heart rate sensor and much more. IoT in cloud provides the connection
20 IoT Based Intelligent Ambulance Monitoring and Traffic Control … 271

which is necessary to share the information between the devices. Cloud Computing is
another important technology [14]. Cloud enables user to access data from anywhere.

20.2 Proposed System

20.2.1 Block Diagram

The block diagram is divided into three parts: Traffic section, Ambulance section
and Hospital section.
The block diagram of traffic section is shown in Fig. 20.1 in which two RFID
readers are connected to Arduino Uno to control the traffic signals [15].

POWER
SUPPLY

RFID READER
(1)
TRAFFIC
ARDUINO
LIGHTS
UNO

RFID
READER(2)

Fig. 20.1 Traffic section

Fig. 20.2 Ambulance

section POWER
SUPPLY

HEART ZIGBEE
RATE
SENSOR
ARDUINO
UNO IoT
IR
272 J. J. Godwin et al.

Fig. 20.3 Hospital section

ARDUINO ZIGBEE
POWER
UNO
SUPPLY

The block diagram of ambulance section is shown in Fig. 20.2 where the heart rate
sensor and IR sensor are connected to Arduino Uno which can be used to transmit the
information of the patients to the nearby hospitals and receive an acknowledgement
from the hospital which is ready to give a treatment to the patient through Zigbee
and IoT [16].
The block diagram of hospital section is shown in Fig. 20.3 where the Arduino
Uno is connected to Zigbee to receive the request and transmit the response to the
ambulance.

20.2.2 Hardware Components

1. Heart Rate Sensor: The device provides optical heart beat production. By using
LED light source and an LED light sensor, the heart rate monitor tests the cardiac
output in beats per minute (BPM). The sensor tests the light output that reflects
back as the refracted through your surface of the skin [17]. The reflection of light
is varying as blood pulses under your skin past the light.
2. IR Sensor: An Infrared sensor is an electronic device. An IR sensor can measure
the temperature of the patient as well as detects the motion of an object. Here,
Passive infrared sensor (PIR) is used [18]. These types of sensors measures only
IR rather than emit.
Infrared Radiation is invisible to the human eyes, which can be identified by the
help of IR sensor. The transistor is merely an IR LED and the sensor is merely
an IR Photodiode responsive to IR light of the spectrum close to that produced
by the IR LED. As IR light reflects on the photodiode, the sensitivity and the dc
voltage change proportionally to the intensity of the obtained IR light [19].
3. RFID (Radio Frequency Identification) Reader: It is a machine which uses
frequency signals to transmit the information wireless network between it and
an RFID tag/label to recognize, characterize and monitor properties. It is used to
integrate an RFID tag. In this project, two RFID readers are used, one for detect
the normal vehicles and another for detect the emergency vehicles.
The transmitter and receiver pins of the RFID reader are connected to the trans-
mitter and receiver of Microcontroller respectively. The reader then detects the
information from the Tag and sends the perceived data through signal port to the
Circuit board. It is one method of Automatic Identification and Data Coupling
(AIDC).
20 IoT Based Intelligent Ambulance Monitoring and Traffic Control … 273

4. Zigbee Module: It is IEEE 802.15.4 based architecture, low power, low data rate
supporting wireless networking standard, which is commonly used for two-way
communication between sensors and control system. In this project, this module
is used for the communication between the ambulance section for sending a
request and hospital section for sending a response.
5. Arduino Uno: It is an open source electronics platform based on easy to use
hardware and software. It is an AT Mega 328p based Microcontroller board. It
has 14 digital input or output pins in which 6 pins can be used as PWM output
pins. USB cable can be used to connect with a computer. It is used in traffic
section to control the signals, ambulance section to transmit a request to the
hospital and hospital section to transmit an acknowledgement to the ambulance
section.

20.3 Algorithm

20.3.1 Preemption Strategy

An efficient preventive signaling and route choice algorithm is expected in this study.
This approach direct to Toggle traffic control management method system to clear the
traffic congestion before Emergency vehicle (EV) arriving the junction. In order to
reduce the travel time, emergency vehicle should resume the transportation without
any disturbances. To find the best route from start of the emergency vehicle to the
terminal is the aim of the path selection strategy. Shortest path to reach the destination
is considered as the best path and it has the least harmful effect on daily traffic.
This logic calculates the minimum detection distance of emergency vehicle using
the following equation:

T ≥ Tso + Tm + ST

where T is the notification period, Tso is the changeover period of the traffic signal
head, Tm is the discharge signal cycle, and ST is the safety time interval.
Interval of switching the signal state is said to be Tso (Switchover Time). By using
the normal length of line and rate of discharge from queue of the emergency vehicle
with the help of historical data, discharge time Tm can be calculated. Two seconds
is the constant value for Safety time interval (ST).
The necessary green indicator head time on the Emergency vehicle can be used
to measure response by the following equation:

Tgr = T −(Tso + Tm + ST )

Once the emergency vehicle is figure to detach from the junction, the green time
(Tgr ) can be halted instantly for the unilateral step.
274 J. J. Godwin et al.

In the category with path crossings, it is presumed that if the gap between two
consecutive junctions is less than the shortest distance to detecting emergency vehi-
cles. Then the emergency vehicle detection point for first junction has been affected.
The date set for notice has also taken into an considering the release duration of both
crossings. The distance of detection for the second intercept is computed equivalent
to that of a single intersection when the gap between the two successive collisions
is greater than the duration of observation. The flow chart for preemption algorithm
has been explained in Fig. 20.4. In the next section, the best path selection will be
explained.

Fig. 20.4 Flow chart for

preemption algorithm
20 IoT Based Intelligent Ambulance Monitoring and Traffic Control … 275

20.3.2 Path Selection Strategy

The aim of the best path selection is to reduce the response time. In emergency
vehicle preemption, the shortest path selection is always crucial. By using Dijkstra’s
algorithm, the shortest path from emergency vehicle’s origin to the destination will
be calculated with the help of location of emergency vehicle and the given destination

T (i, j) = Li[1 + C(i, j)]


C(i, j) = Bk[P(k, j) · V (k, j)]

[1 + V (k, j)]

where V(k, j) is the Passenger amount reached regulator k on the link i during j; Bk
is the weight for the detector k in the link i

Bk = 1;

Li = [0 − 1];

P(k, j) = {1: if detector k is occupied by a vehicle at the end of j; 0: otherwise.}

where T (i, j) is the connection transport expense I during the period interval j is
determined from the duration of the link Lj and the traffic flow level C (i, j) during
period j.

20.4 Result and Discussion

The proposed output has been explained using Figs. 20.5 and 20.6.
In ambulance section, whenever, the accident takes place, ambulance has to come
to the accident spot and pick the injured person, IR sensor is used to detect the patient
on stretcher, once detected heart rate sensor is used to monitor patient’s pulse rate.
Through serial monitor the Nurse has to input the type of injury, so that request is
continuously send to the near by hospitals.
ARDUINO SOFTWARE IDE is used for searching the hospital, once the nurse
entered the type of disease the message sent to all nearby hospitals (Fig. 20.7).
In Hospital section, it has facility for that injury will respond to ambulance
(Fig. 20.8).
276 J. J. Godwin et al.

Fig. 20.5 Traffic section

Fig. 20.6 Ambulance section

20.5 Conclusion

This article represents controlling the traffic congestion to give freeway systems for
Ambulance and monitoring the ambulance. Preemption algorithm is used to control
the traffic congestion automatically. The request sent to the hospital from the ambu-
lance and the acknowledgement send to the ambulance from the hospital is done by
IoT. The main advantage of this project is that the request from the ambulance and
response from the hospital is done efficiently by IoT and automatic traffic freeway
system for ambulance done by preemption algorithm formulates normal and emer-
gency traffic control strategies. Internet of things-based emergency surveillance is a
new technologies which works for people. In future, new methods can be developed
for complex road networks.
20 IoT Based Intelligent Ambulance Monitoring and Traffic Control … 277

Fig. 20.7 ARDUINO SOFTWARE IDE

Fig. 20.8 Hospital section

References

1. Huang, Y.S., Weng, Y.S., Zhou, M.: Design of traffic safety control systems for emergency
vehicle preemption using timed Petri nets. IEEE Trans. Intell. Transp. Syst. 16(4), 2113–2120
(2015)
2. Qi, L., Zhou, M., Luan, W.: A two-level traffic light control strategy for preventing incident-
based urban traffic congestion. IEEE Trans. Intell. Transp. Syst. 19(1), 13–24 (2016)
278 J. J. Godwin et al.

3. Pasquale, C., Sacone, S., Siri, S., Papageorgiou, M.: Optimal control for reducing congestion
and improving safety in freeway systems. IEEE Trans. Intell. Transp. Syst. 19(11), 3613–3625
(2018)
4. Rajeshwari, S., Hebbar, S., Golla, V.: Implementing intelligent traffic control system for
congestion control, ambulance clearance and stolen vehicle detection. IEEE Sens. J. 15(2)
(2015)
5. Gnanaraj, V.V., Ranjana, P., Thenmozhi, P.: Patient monitoring and control system using internet
of thing. Int. J. Innov. Technol. Explor. Eng. (IJITIEE) (2019). ISSN: 2278-3075
6. Hari Kishore, K., Surendra Nath, K.V., Hari Krishna, K.V.N., Pavan Kumar, D., Manikanta,
V., Basha, F.N.: IOT based smart health monitoring alert device. Int. J. Innov. Technol. Explor.
Eng. (IJITIEE) (2019). ISSN: 2278-3075
7. Priyadharshini, K., Manikandan, S.K.: Automatic traffic control system based on the vehicular
density (2019)
8. Patil, S., Pardeshi, S.: Health monitoring system using IoT. Int. Res. J. Eng. Technol. (IRJET)
(2018). e-ISSN: 2395-0056, p-ISSN: 2395-0072
9. Bhilawade, V., Ragha, L.K.: Intelligent traffic control system. Int. J. Sci. Res. Publ. (2018).
ISSN: 2250-3153
10. Kumar, G.H., Ramesh, G.P., Avadi, C.: Novel gateway free device to device communication
technique for IoT to enable direct communication between homogeneous devices. Int. J. Pure
Appl. Math. 118(16), 565–578 (2018)
11. Shahada, S.A.A., Hreiji, S.M., Shamsudheen, S.: IOT based garbage clearance alert system
with GPS location using Arduino. Int. J. MC Square Sci. Res. 11(1), 1–8 (2019)
12. Kumar, G.H., Ramesh, G.P.: Intelligent gateway for real time train tracking and railway crossing
including emergency path using D2D communication. In: 2017 International Conference on
Information Communication and Embedded Systems (ICICES), pp. 1–4, Feb 2017. IEEE
13. Badawi, W.A.: Underground pipeline water leakage monitoring based on IOT. Int. J. MC Square
Sci. Res. 11(3), 01–08 (2019)
14. Babu, V.H., Balaji, K.: Survey on modular multilevel inverter based on various switching
modules for harmonic elimination. In: Intelligent Computing in Engineering, pp. 451–458.
Springer, Singapore (2020)
15. Kumanan, T.: Link quality and energy-aware metric-based routing strategy in WSNS. In:
Intelligent Computing in Engineering, pp. 533–539. Springer, Singapore (2020)
16. Manikandan, G., Anand, M.: Radix-2/4 FFT multiplierless architecture using MBSLS in
OFDM applications. In: Intelligent Computing in Engineering, pp. 553–559. Springer,
Singapore (2020)
17. Sundaram, A., Ramesh, G.P.: Investigation of solar based SL-QZSI fed sensorless control of
BLDC motor. In: Intelligent Computing in Engineering, pp. 779–787. Springer, Singapore
(2020)
18. Snehalatha, N., Shiny Angel, T.S., Amudha, S.: Remote display access using remote frame
buffer and IO streaming. Int. J. MC Square Sci. Res. 8(1), 23–40 (2016)
19. Rajesh, D.: CH panel based routing scheme for mobile wireless sensor network. Int. J. MC
Square Sci. Res. 8(1), 183–198 (2016)
Chapter 21
IoT Based Emergency Alert System

K. Chinnusamy, D. Nandhini, A. S. Subhashri, M. Hemavathy, V. Baskar,

M. Kavimani, and M. Vignesh

Abstract There is rapid increase in population, due to which associated challenges

in safety and health care are also in raise. No proper assurance is given to safety,
especially for women. Moreover, elder people in isolation are facing many health
issues due to which continuous monitoring is essential. In order to pave path for these
issues, a method is proposed which will ensure road safety, women safety as well
elder people safety. On the whole, the major threats and difficulties faced by people
will be prevented or solved using this application.

Keywords IoT · Geo fencing algorithm · Sensor · Fall detection · Safety

21.1 Introduction

The personal security of women and elderly people are highly vulnerable in the
modern society at different times and venues. Sexual harassments and also other
forms of sexual violence are very often in public places such as nearby schools,
colleges, parks. This way, the freedom to girls and women is not ensured thereby,
preventing them from exploring this competitive world with hidden talents. A wear-
able smart band is designed which will ensure safety for women during emergencies
[1]. Many systems were designed for women’s safety of which, a safety system is
described in [2]. In addition to this, road safety is not ensured now a days, a certain
proportion of which results from untimely treatments and secondary accidents. These
accidents need to be prevented before occurring or immediately further action must
be taken to rescue the victim. The accidents are detected using different algorithms
such as weighted extreme learning algorithm which is explained in [3]. To detect the

K. Chinnusamy (B) · D. Nandhini · A. S. Subhashri · M. Hemavathy · V. Baskar · M. Kavimani ·

M. Vignesh
Department of Electronics and Communication Engineering, Velammal Institute of Technology,
Chennai, India
e-mail: [email protected]
D. Nandhini
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 279

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_21
280 K. Chinnusamy et al.

accidents, the collision of vehicles need to be detected which is done using speed
sensors and it is described in [4]. Also [5], discusses about controlling the speed of
the vehicles using accelerometer sensor so that the speed can be controlled. More-
over, the traffic incidents that occurred needs to be classified to find the severity
of the incident. This can be achieved using advanced machine learning and active
learning algorithms [6]. Further, the number of elder people living alone at old age
homes and the number of single-resident houses are also increasing worldwide. In
order to monitor the elder people, their heart beat rate is taken into consideration
but many real time heart beat signals contain artifacts that needs to be removed and
the buried accurate signals must be recovered which is discussed in [7]. However,
the independent life becomes quite difficult for them. The cause of injuries in the
elderly people is maximum due to instability that leads to falls. These falls need to
be detected using a corresponding device that is handy to use. The falls must be
detected efficiently which is discussed in [8]. The general activities of old age people
such as their walking speeds are measured which can be used to train the support
vector machine and random forests algorithms so that the cognitive impairments were
detected earlier [9]. In [10], without any physical contacts with the people, a face-
based heart rate evaluation technique is proposed. In [11], the person’s front-door
events such as entry, exits etc. are classified using an algorithm to prevent dementia.
Also, the emergencies occurring with different severities are captured and responded
immediately which is described in [12].
The existing systems mentioned above includes few demerits due to which the
purpose of those systems were not served as expected. These demerits include unfash-
ionable designs so that the people find it difficult to access those applications with
ease. Moreover, the existing applications include facilities like tilts that are done
unnoticed due to which the authorities are alerted without necessity. In order to over-
come those hurdles, an application is developed which can be accessed by public with
ease. This application is can be accessed even to ensure general safety such as road
safety and importantly, the elder people living as single residents can lead a healthy
life without health issues. Section 21.2 describes about the proposed method which
includes wireless alert, the modules used and the algorithm. Section 21.5 depicts the
results and discussion. Finally, Sect. 21.6 provides the conclusion.

21.2 Proposed Method

21.2.1 Wireless Alert Through Internet of Things (IoT)

Figure 21.1 depicts the alert to the respective authorities which is wireless. The
problem is detected by the sensor which sends the data to the interface system for
the amplification process. The alert signal after amplification will be sent to the
microcontroller to proceed with the further action of sending the data to the main
server. From the main server, the location, voice and picture of the victim’s location
21 IoT Based Emergency Alert System 281

Fig. 21.1 Algorithm of IoT alerting system

will be sent to the respective authorities as URL links via GPS and GSM modules. In
the proposed system, the android application includes three sensors for three different
purposes. Vibration sensor is used for accident detection, heart beat sensor is used
for health monitoring and MEMS sensor is used for fall detection.
Whenever an accident is detected, the vibration sensor will be triggered, if an old
age person is suffering from health issues, the heart beat sensor will be triggered
and also if the person faints, the MEMS sensor will detect the fall [13]. The sensor’s
output as values will be interfaced using an interface system. The interface system
includes amplification of the output signal. The amplified output will be sent to the
PIC microcontroller which will be paired to the sensor using Zigbee [14]. The alert
signal is sent to the main server (mobile server) after which the GPS and GSM
modules will be activated. Using GPS, the location will be tracked and sent as URL
links to the concerned people of the victim (Hospitals, patrols, relatives, friends,
neighbors). Similarly, using GSM, the Short Message Service (SMS) will be sent to
the respective authorities with the location of the incident. In case of harassment, the
victim’s image and recorded voice as URL links will be sent via SMS. The wireless
communication plays a major role in sending alert messages to the server and to the
respective authorities.
Figure 21.2 depicts the block diagram of the proposed system to indicate the
inputs and output of the design. The components involved in the proposed system
are four sensors namely vibration sensor, MEMS sensor, temperature sensor and
heart beat sensor. These sensors will play the role of detecting the hurdle faced by
the victim. The vibration sensor senses the collisions during accidents, the MEMS
sensor detects the fall in case of elderly people and the temperature sensor will sense
the abnormal temperatures to treat the victim. The heart beat module is used to detect
the abnormal heart beat rates of the victim.
282 K. Chinnusamy et al.

Fig. 21.2 Block diagram of

the alerting system

21.2.2 PIC 16F877A

The PIC 16F877A is a 40 pin micro controller which is used to implement the
programs with ease. It requires an operating voltage of 2–5.5 V. The micro controller
supports communication protocol. The output of this controller is the alert signal sent
in the form of URL links to the mobile server. The main merit is its fast performance
with less power consumption.

21.2.3 Sensors

The sensors used in this project are vibration sensor, heart beat sensor and MEMS
sensor. These three sensors are used for three different purposes. The vibration sensor
to detect the accident, the heart beat sensor to monitor the health condition of old
age people and the MEMS sensor to detect the fall of a person which will be applied
to women’s safety. The MEMS sensor is an analog to digital sensor which will be
connected to the ADC pin of the micro controller. Its output will include the ADC
values which consists of three coordinate angles namely X, Y and Z angles based
on which the fall of old people can be detected. The MEMS sensor output along
with the output signals of the other two sensors each will be sent to the interface
consumption.
21 IoT Based Emergency Alert System 283

21.3 Module

21.3.1 Zigbee

The Zigbee is a wireless technology which is used to control and monitor the devices
with low power consumption at low cost. Its low data rate makes this standard more
suitable to handle many devices where data rate of 250 kbps is used for transmission
between embedded systems both the ways. Zigbee operate at 2.4 GHz and also
between 902 and 928 MHz. This project uses Zigbee instead of Bluetooth as there
is no pairing required with other devices as in Bluetooth so that even other devices
can connect to it without un-pairing the existing device.

21.4 Algorithm

21.4.1 Geo Fencing

Step 1—The algorithm used for the detection of the above mentioned problems is
geo fencing algorithm.
Step 2—It is a location-based application that uses GPS, wireless fidelity and RFID
that are preprogrammed in it.
Step 3—Geo fencing fetches the longitude and latitude values of the location.
Step 4—The fetched values will be converted as URL links which will be sent via
SMS to the respective authorities.
Step 5—By selecting the received links, the concerned location of the incident will
be viewed.
Step 6—This algorithm is accurate in spotting the location and in fetching the corre-
sponding latitude-longitude values. For example, the latitude of Chennai is 13.0827°
N, 80.2707° E.
Step 7—This value will be displayed which must be selected to view the current
location.
Step 8—Importantly, the geo fencing algorithm is introduced in this project as java
codes.
Step 9—The respective coding is written for this logic to make it fetch the location
of the incident.
Step 10—Hence, the mobile application is developed with the frame work of java.
284 K. Chinnusamy et al.

21.5 Results and Discussion

The objective of the project is to ensure safety to all people in general by providing
them with an android application so that they can access the facility whenever
required. The components used are inexpensive with low power consumption which
is the ultimate aim of the project. The uniqueness of the project is that, a single
application is developed to ensure road safety, women safety and good health of
people.
Figure 21.3 indicates the main three options displayed in the application that will
enable the user to register their details such as name, guardian name and respective
address.
Figure 21.4 depicts the details to be entered by the user such as their guardian’s
name, guardian’s address and their mail address. These details of the individual will
help the application to contact the respective authorities during any emergencies such
as during accidents or harassments or during any health issues.
Figure 21.5 represents the options displayed by the application for women to
access it in case of any emergencies. By selecting the panic button or the volume
down button thrice, the front camera will be enabled so that the picture of the victim
will be captured and sent to the respective authorities to draw their attention and to
take further actions.
By downloading this application, the public can access it whenever required. This
application is not specific for any particular purpose, it can be accessed to ensure
safety while driving vehicles, while walking in an isolated place or while residing
as a single resident such as in old age homes meant for elder people. The existing
applications were designed in order to help women especially during their hard times.
By accessing those applications, other people cannot serve their purposes. Applica-
tions such as Kavalan SOS and Life360 were designed and developed especially for

Fig. 21.3 The application

with options to register one’s
details
21 IoT Based Emergency Alert System 285

Fig. 21.4 The details of an

individual to contact them
when required

women’s safety. The proposed application will serve the purpose of people in general
which is the uniqueness of this application.
Table 21.1 depicts the uniqueness of the proposed application over other applica-
tions. The difficulties that are faced in few other applications such as unnoticed tilts
are overcome in our proposed IoT based emergency alert application.
Table 21.2 shows the experimental results of the emergency alert system by
alerting the respective authorities during emergencies in case of women or accident
detection or of elder people to monitor their health.

21.6 Conclusion

An emergency alert system using IoT has been designed. The proposed system
includes three sensors to operate for different purposes such as for road safety, health
286 K. Chinnusamy et al.

Fig. 21.5 The three access

facilities for women, panic
button for elder people and
during accidents

Table 21.1 Comparison between the proposed application and other safety applications including
Kavalan SOS
Purpose Kavalan SOS and other safety Proposed application
applications
Functionality Ensures safety especially for women Ensures women safety, road safety and
elder people safety
Tilt facility Tilt is sometimes done unnoticed which Such discomforts are avoided
alerts all the authorities by mistake

Table 21.2 The success rates

Locations Response time (s) Percentage of achievement (%)
of the proposed system
Location 1 40 90
Location 2 30 94
Location 3 35 92

monitoring of old age people and for women’s safety. This module is designed to alert
the respective authorities such as patrols, hospitals and relatives during emergencies.
These emergencies include accidents, harassments and abnormal health conditions.
The system is capable of ensuring safety among public rather than a particular safety
such as women safety alone which implies that this alert system can be accessed by
everyone in general.

Acknowledgements The authors would like to thank Velammal Institute of Technology for
providing facilities in their Research and Development laboratory.
21 IoT Based Emergency Alert System 287

References

1. Devkota, B., Alsadoon, A., Prasad, P.W.C., Singh, A.K., Elchouemi, A.: Image segmentation for
early stage brain tumor detection using mathematical morphological reconstruction. Procedia
Comput. Sci. 125, 115–123 (2018)
2. Mathur, N., Meena, Y.K., Mathur, S., Mathur, D.: Detection of brain tumor in MRI image
through fuzzy-based approach. In: High-Resolution Neuroimaging—Basic Physical Principles
and Clinical Applications. InTech, Rijeka, Croatia (2018)
3. Alam, M.S., Rahman, M.M., Hossai, M.A.: Automatic human brain tumor detection in MRI
image using template-based K means and improved fuzzy C means clustering algorithm. Big
Data Cogn. Comput. 3(27) (2019)
4. Prastawa, M., Bullitt, E., Ho, S., Gerig, G.: A brain tumor segmentation framework based on
outlier detection. Med. Image Anal. 8, 275–283 (2004)
5. Karthik, R., Menaka, R.: A multi-scale approach for detection of ischemic stroke from brain
MR images using discrete curvelet transformation. Measurement 100, 223–232 (2017)
6. Hemanth Kumar, G., Ramesh, G.P., Avadi, C.: Novel gateway free device to device communi-
cation technique for IoT to enable direct communication between homogeneous devices. Int.
J. Pure Appl. Math. 118(16), 565–578 (2018)
7. Selkar, R.G., Thakare, M.: Brain tumor detection and segmentation by using thresholding and
watershed algorithm. Int. J. Adv. Inf. Commun. Technol. 1, 321–324 (2014)
8. Arzoo, M., Rathod, K.: K-Means algorithm with different distance metrics in spatial data
mining with uses of NetBeans IDE 8.2. Int. Res. J. Eng. Technol. 4, 2363–2368 (2017)
9. Wu, Y., Zhu, M., Li, D., Zhang, Y., Wang, Y.: Brain stroke localization by using microwave-
based signal classification. In: IEEE International Conference on Electromagnetics in Advanced
Applications, pp. 828–831 (2016)
10. Ay, H., Furie, K.L., Singhal, A., Smith, W.S., Sorensen, A.G., Koroshetz, W.J.: An evidence-
based causative classification system for acute ischemic stroke. Ann. Neurol. 58(5), 688–697
(2005)
11. Database, Brainweb. Available at: https://www.bic.mni.mcgill.ca/brainweb/. Accessed 10 Sept
2018
12. Pendem, S., Ramesh, G.P.: 75 GHz 5G frequency spectrum analysis. In: Balas, V., Kumar,
R., Srivastava, R. (eds.) Recent Trends and Advances in Artificial Intelligence and Internet of
Things. Intelligent Systems Reference Library, vol. 172. Springer, Cham (2020)
13. Fiebach, J.B., Schellinger, P.D., Geletneky, K., Wilde, P., Meyer, M., Hacke, W., Sartor,
K.: MRI in acute subarachnoid haemorrhage; findings with a standardised stroke protocol.
Neuroradiology 46(1), 44–48 (2004)
14. Fiebach, J.B., Schellinger, P.D., Jansen, O., Meyer, M., Wilde, P., Bender, J., Hähnel, S.: CT
and diffusion-weighted MR imaging in randomized order. Stroke 33(9), 2206–2210 (2002)
Chapter 22
An Industrial Internet of Things
Approach for Pharmaceutical Industry
Growth

Deepak Kumar Sharma, Gurmehak Kaur, and Mohita Sharma

Abstract This chapter deals with the basic introduction to the concept of Internet
of Things (IoT) which is a concept that enables a device to connect to the Internet
or other devices, hence forming a giant ecosystem. The IoT platform acts as the
brain of the system, while the devices linked together via IoT function as limbs.
These devices can be controlled from anywhere. There are certain requirements that
a device must fulfill to be called an IoT device, such as its capability to connect to
the Internet or if it’s integrable with technology involving sensors and software. A
perfect example of such devices is the Radio-Frequency Identification (RFID)—tech-
nology which uses electro-magnetic fields to automatically recognise and track tags
attached to objects. Another well-known and useful technology is the Global Posi-
tioning System (GPS)—using which real-time continuous data or information can
be gathered from vehicles, ensuring ease in shipping facilities. There are numerous
advantages that IoT could provide us with, time-saving being a crucial one. On the
other hand, it is equally important to consider some demerits of these smart devices
as many fear that their increasing popularity may take control of life eventually, with
increasing AI technology. This chapter shall cover the fundamentals of IoT and how
IoT devices are gaining popularity in this era. IoT becomes a lot more interesting
and useful when applied to the Pharmaceutical Industry, which will be our major
area of focus in the next section of the chapter. There have been a lot of discussions
on the obstacles faced by the pharma industry which include issues like—lack of
professionals in the industry, poor industry standards, high investment costs, lack
of transparency, amongst many others. Introduction of IoT to pharmaceuticals has

D. K. Sharma (B)
Department of Information Technology, Netaji Subhas University of Technology (Formerly
Known as Netaji Subhas Institute of Technology), New Delhi, India
e-mail: [email protected]
G. Kaur · M. Sharma
Department of Electronics and Communication, Netaji Subhas University of Technology
(Formerly Known as Netaji Subhas Institute of Technology), New Delhi, India
e-mail: [email protected]
M. Sharma
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 289

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_22
290 D. K. Sharma et al.

opened up a plethora of solutions to the above-mentioned problems, thus facili-

tating faster and efficient supply to the ever-increasing market demands. This section
provides a detailed study on how the application of IoT devices in this industry
ensures improved supervision of manufacturing of drugs, preventive maintenance of
machinery, and improvement in supply chain management. For example, to bring
about transparency in the manufacture of drugs, sensors can be used to monitor the
physical factors that could affect the state of the drugs. If safety levels are breached,
an alarm is triggered. In terms of Supply Chain Management, several difficulties can
present themselves before the industry such as vehicle accidents or breakdown, loss
of conveyance, etc.—problems that could prove to be a major setback for the industry
and the clients. However, the above advantages come with slight disadvantages as
well, privacy being the main area of concern. Client information must be handled
with utmost care. Data Security should be taken seriously as any lag in this area could
lead to major issues in terms of cybersecurity. The digitalisation of the healthcare
industry worldwide leading to the linkage of IoT with pharmaceuticals has proved
to be a huge step forward for society as a whole. Pharma organisations can benefit
greatly from real-time analysis of factors such as production, shipment, conveyance
management, etc. owing to improved tracking facilities and enhanced connectivities
through IoT.

Keywords IoT · IoT devices · Pharmaceutical industry · Sensors · RFID · Supply

chain management

22.1 Introduction

22.1.1 Introduction to IoT

The Internet of Things, or IoT refers to a network of computing devices, mechanical

and digital machines, objects, animals or people that are capable of transferring data
over a network without demanding any sort of interaction between humans them-
selves, or humans and the internet. IoT is a concept that is fairly easy to understand
as it deals with connecting anything and everything possible to the internet.
A thing is capable of transmitting or receiving data, or both, as soon as it is
connected to the Internet. This is why IoT technology is so useful in the modern
world [1, 2].
All the connections made to the Internet in IoT technology are categorised as
follows:
1. Collection of information and then acting on it.
2. Receiving information and then acting on it.
3. Things capable of doing both.
IoT is a result of the combination of wireless technologies microservices and
the internet, which in turn has aided the combination of operational technology
22 An Industrial Internet of Things Approach … 291

(OT) and information technology (IT), thus making it convenient for unstructured
machine-generated data to be scrutinized in order to bring about improvements. An
IoT network deals with smart devices that are web-enabled and that use embedded
processors, sensors and communication hardware to perform the above mentioned
3 steps. The information collected by IoT devices are shared via a common IoT
gateway through which data that is being sent is either analyzed locally, or is sent to
the clouds. IoT devices are also capable of communication or interacting with one
another, thus working on the data they receive from each other. Majority of the work
is done by the devices themselves, even though humans can interact with them—for
example, to transmit certain instructions or to gather certain data [3].

22.1.2 Introduction to Pharmaceutical Industry

A pharmaceutical company is basically a commercial drug company that is permitted

to research, develop, market and/or distribute drugs, mostly in the area of health-
care. The pharma industry is faced with numerous obstacles which include issues
like—lack of professionals in the industry, poor industry standards, high invest-
ment costs, lack of transparency and many more. There are 2 major reasons why a
pharma company turns towards IoT technology—for the safety of consumers and
improvement in the machine lifetimes.
IoT helps in providing pharmaceuticals with a number of efficient and effective
solutions for the above discussed problems, thus keeping up with the sky-rocketing
market demands. Onboard sensors and digital maintenance dashboards play a vital
role in helping a company acquire knowledge about the health of the machine while
keeping consumers safe at the same time.
In case of bad weather or traffic conditions, location-based IoT shipment tracking
such as GPS or cloud-based monitoring systems come in handy, especially if one
wishes to shift an item/shipment to an equipped facility temporarily. Thus, IoT tech-
nology provides us with new ways to manage, monitor and enhance some of the most
vital processes within the healthcare industry.

22.2 Background and Research on IoT

Internet of Things involves 3 stages:

1. Collection of data by sensors
2. Storage and analysis of data by an application
3. Decision making and the transfer of data to the main server.
When it comes to IoT communication, the ubiquitous connectivity provided by the
multipurpose network converts the physical world into one that can be programmed
into and is completely automated. IoT has now opened up a plethora of opportunities
292 D. K. Sharma et al.

in automation, optimization, intelligent manufacturing and smart industry, industrial

control, and is rapidly moving towards new methods of servicing customers, which
includes an on-demand service model, all of this leading to drastic transformations
in the industry. At the same time, IoT comes with such great usefulness and power
that it is capable of revolutionizing pharmaceutical manufacturing in various ways.
This is the reason why numerous leading pharma companies from around the globe
are turning to IoT technologies in their manufacturing plants to achieve optimization
and improve process efficiency [4].
An IoT network is composed of four major components: sensors/devices,
connectivity, data processing, and a user interface.
1. Sensors/Devices

Firstly, data is acquired from the environment via these sensors. Temperature reading
is a good example of this component, and a fairly simple one.
A number of sensors can be grouped together, which is why it is an attractive bet.
These devices can be a part of another device that is capable of carrying out functions
besides sensing. For instance, a phone comprises of various sensors such as the GPS,
but at the same time, it is not JUST a sensor. It has various other functions as well
[5].
2. Connectivity

Now, the data that is gathered from the surroundings is sent to the cloud. What is
a cloud? A cloud is basically a set of servers that can be accessed via the Internet.
Cloud computing eases out things for the user as he/she no longer needs to deal with
physical servers themselves. But before any of this, data must first reach the cloud.
How does that happen?
As can be seen from Fig. 22.1, it is essential for the devices to be linked to the
cloud. This is possible through a variety of methods and each of these methods
accomplishes the task in hand, reaching the cloud [5].
3. Data Processing

A software is available to process the data once it reaches the cloud.

This method of processing ranges from extremely simple to complex. Checking
if the temperature reading is within specified limits is a simple process, while the
process used to identify intruders in your house, i.e., computer vision, is a fairly
complex one. But what happens when both these processes do not give the desired
results? That involves the work of a user [5].

Fig. 22.1 Cloud computing

22 An Industrial Internet of Things Approach … 293

4. User Interface

Now, the obtained information must be made useful to the user. There could be many
ways to achieve this, such as a notification alert in case the temperature reading
shoots up or goes below acceptable levels. Also, it should be possible for the user to
check in on the video feed when required, either via an app or the web.
But, it isn’t this straightforward all the time. It is possible for the user to carry out
an action which in turn affects the system. For instance, the temperature in a cold
storage could be adjusted via an app on the phone.
Also, some automatic actions may be performed. In case the temperature in
the cold storage dips low, the system can automatically bring it up using a prede-
fined algorithm. In a similar manner, on intrusion, the system can immediately and
automatically notify the concerned authorities directly [5].

22.3 Technologies Used with IoT

The technologies that use IoT are broadly classified into the following categories:
1. Short Range IoT Network
2. Medium Range IoT Network
3. Long Range IoT Network.

22.3.1 Short Range IoT Network

• RFID

IoT demands a few essential components to facilitate communication between an

object and a device. For an object to be identified uniquely, it must be augmented
with an RFID tag. This tag also enables an object to communicate certain sets of
information, wirelessly. But what exactly is RFID technology?
Radio frequency identification system (RFID) is an automatic technology and
helps machines or computers in identifying objects, recording metadata or control-
ling individual target through radio waves. Typically, an RFID system consists of
tags (they can be transmitters or receivers) and readers (they can be transmitters or
receivers). The tag is a microchip that is connected with an antenna, that is attached
to an object. This acts as the identifier of the object. Now, communication between
the reader and the tag takes place via radio waves, as is shown in Fig. 22.2. Identifica-
tion information is transmitted to the reader through wireless communication. Using
radio frequency waves to transmit data wirelessly without any contact is what RFID
is all about. There are 3 major aspects of an RFID system: monitoring, tracking, and
supervising. Monitoring refers to observing a system for any changes or warning
signs, i.e., to be aware of the system state. Tracking refers to the act of observing
294 D. K. Sharma et al.

Fig. 22.2 Working of RFID

moving objects or persons. Supervising refers to the act of supervising the behaviour
and activities of people [6, 7].
RFID tags and readers work together to enable users to identify, track and monitor
objects.
Hospitals are now implementing technology requiring RFID so as to keep a record
of the inventory, track patients, and manage workforce. Also, in order to avoid medical
errors, verification of patient identification is done with the help of these applications.
The system is also used to collect data of hospital operations, for feedback.
• Bluetooth

Bluetooth is viewed as an important solution for the future of electronics gadgets

like wireless headphones or geolocation sensors, especially considering the fact that
every smartphone comes with the Bluetooth feature. The Bluetooth Low Energy
protocol is extremely cost effective and draws negligible power from the device. As
every coin has two sides, Bluetooth too, has its disadvantages. It is not feasible for
large chunks of data.

22.3.2 Medium Range IoT Network

• WiFi

Wifi is known to be the most popular and widespread wireless communication

protocol. In order to obtain better reliability and speedier connectivity, one must
retain high signal strength. Even though Wifi provides a strong foundation to huge
no. of IoT solutions, it still has to be managed and used in a way that it yields profits
22 An Industrial Internet of Things Approach … 295

to both the users as well as the service providers. Linkify acts as a perfect example in
this case. This is one of AVSystem’s state-of-the-art solutions which enables endless
guest Wifi customizations.
• ZigBee

This protocol is quite popular ad is widely used in several management systems,

ordinary home appliances, and machine industry. Zigbee proves to be an extremely
reliable protocol which has high security while at the same time, enables the users
to enjoy low data exchange rates. This requires low power for its working.
• Thread

Designed especially for smart homes, Thread utilises IPv6 connectivity that allows
devices connected with each other to communicate, access services in the cloud, and
interact with the users through Thread mobile apps.

22.3.3 Long Range IoT Network

• NB-IoT

An outcome of existing 3GPP technology, Narrow-band IoT is a recent radio tech-

nology standard which requires low power (battery power operation for approxi-
mately ten years) and provides connectivity with low signal strength, approximately
twenty three dB lower than that of 2G. Making use of the current network infras-
tructure, NB-IoT not only ensures universal insurance in LTE networks, but also
guarantees the quality of the signal. In many cases, these features call for imposing
NB-IoT rather than those solutions that need the construction of LoRa, Sigfox or
other local networks [8].
• LTE-Cat M1

LTE Cat M1 is a low power wide area (LPWA) connectivity standard which links IoT
with M2M devices that require medium data rates. It offers long battery life cycles
and gives more suitable in building range in comparison to mobile technologies such
as 2 G, 3 G and LTE Cat 1 [8].
Being compatible with the current LTE network, CATM1 doesn’t require the
carriers to construct new infrastructure to implement it. LTE Cat M1 proves to be
ideal for mobile use cases, as its control of hand-over between cell sites is much
better compared to NB-IoT. It is quite similar to high speed LTE [8].
296 D. K. Sharma et al.

• LoRaWAN

LoRaWAN is a low-power Long Range Wide-Area Networking protocol suitable for

low-power consumption and supporting large networks to which millions of devices
are connected. Desiring applications in wide-area network (WAN), LoRaWAN aims
to furnish low-power WANs with properties required to aid low-cost, mobile and
stable bi-directional communication within M2M, IoT and industries [8].
• Sigfox

The idea behind Sigfox is to serve an efficacious connectivity for low-power M2M
appliances that require low levels of data transfer for very short WiFi range, and
cellular range is exorbitant. Sigfox makes use of UNB technology that prepares it to
handle low data transfer speeds of 10–1000 bps (bits/second) and consumes up to
100 times less energy as compared to cellular communication solutions. It provides
a typical standby time of twenty years for a 2.5 Ah battery. Offering a vigorous,
scalable and energy-efficient network that is capable of supporting communication
between thousands of battery-operated devices across several square kilometres of
area, Sigfox is applicable for various M2M applications, such as patient monitors,
intelligent meters, environmental sensors, smart street lighting and security devices.
It is presently applied in several IoT technologies [8].

22.4 Slow Growth in the Pharma Industry

Biopharma companies worldwide are facing increasing amounts of stress and pres-
sure at the point of time when they will be required to make certain important changes
in their business models. There could be many factors affecting this decline, analysts
say. The reasons could include—poor season leading to lower offtake, deterioration
in the launching of products, regulation of prices in the market.
Pharmaceuticals may not be in a position to ensure safe development and manu-
facturing of medicines along with their distribution, mainly owing to the lack of
transparency. This could result in a number of serious repercussions—discard of
drugs, loss in revenue or even worse, mistreatment of patients. The key area of focus
for a pharma company should be on how to remain competitive, enhance marketing,
reduce or completely avoid late shipment deliveries and to reduce the amount of
waste produced. All these factors must be accounted for, while at the same time
taking care of the wider patient groups with efficiency and improving their drugs.
Logistics acts as a major challenge for a pharma company as there is poor consis-
tency in the transfer of information. Since there is a lack of timely and accurate
information transfer while transporting pharmaceutical goods, it becomes increas-
ingly challenging to trace the recent position of the goods in the supply chain. Thus,
IoT is introduced here to overcome this difficulty. What IoT does over here is that
22 An Industrial Internet of Things Approach … 297

it connects the real and virtual worlds with the internet cloud, thus facilitating live
transfer of data.

22.5 IoT in Pharma Industry

IoT supplies people and companies higher awareness in as well as command with
ninety nine percent of items plus surroundings which continue to be beyond the access
of this web. And also by doing this, IoT enables individuals and companies to become
more attached to the physical world as well as to perform much more significant,
higher level work. Right from detecting issues in production and manufacturing
to checking the situation along the supply chain, modern technology is actually
changing how medicines are created, sent, as well as consumed.
India is the biggest provider of generic drugs internationally and also bene-
fits from a crucial place in the worldwide pharmaceuticals sector. We have seen
lots of discussions on the obstacles experienced by the pharma sector including
issues like—absence of experts within the market, bad industry standards, increased
purchase expenses, insufficient transparency, amongst many others. Advent of
IoT to the pharma industry has opened up a plethora of fixes to the above
mentioned problems, thus facilitating efficient and faster supply to the ever increasing
industry demands. The pharmaceutical market is faced with numerous opportunities,
including greater command on manufacturing of drugs, proper maintenance of the
required equipments, and also enhanced supply chain management.
The overwhelming advances of the past 20 years in telecommunications
networking, analysis and data collection, along with artificial intelligence are at last
being brought to bear on the overwhelming difficulties dealing with the health care
processes, particularly the high-cost of managing an ageing public which is suffering
from much more persistent ailments. Health systems as well as entrepreneurs are
establishing brand new delivery models which use the capability of mobile devices
to constantly keep track of a person’s compliance and condition and also provide care
in a far more integrated way. The technology to evaluate huge quantities of infor-
mation is allowing more effective discovery of drugs, much more exact diagnosis,
personalized care, as well as payment designs based upon patient results.
With this changing health care atmosphere, bio pharmaceutical businesses may
not run in isolation. They are going to have to operate in cohesiveness with specialists
within advanced analytics and communication technologies. Real-world details along
with proof will have to be at the center of successful commercialization and clinical
development. A vital takeaway is the fact that biopharma businesses will have to
invest in these unique capabilities while at the same time keeping the scientific core
of theirs intact along with the pipeline of products that are new. This will be a historic
challenge faced by this industry.
Despite the fact that IoT remains in its nascent stages, the usage of smart equip-
ment as well as machine-to-machine (M2M) interaction using SMAC technologies
comes at a time when the industry is battling with patent cliffs and declining R&D
298 D. K. Sharma et al.

productivity. IoT is not only altering the individual knowledge faster, but additionally
is generating a remarkable impact on other parts of the market like R&D, medical
growth and then supply chain.
Utilization of smart devices in medical advancement, supply chain as well as
patient engagements not only bring down time-to-market for medications but addi-
tionally the real time information feeds could be ploughed to proactively identify
mistakes over the value chain as well as, therefore, enhance regulatory conformity.
Information coming from wearable products can be utilized by HCPs to recommend
personalized medications (PM) which will enhance medication effectiveness mani-
fold and can bring down therapy phase. Particularly, IoT receptors and trackers are
able to allow optimal and effective conditions for dealing with chemicals and bioma-
terials, making sure of perfect equipment operation, and also helping with medication
fraud prevention.

22.6 Advantages of IoT in the Pharmaceutical Industry

IoT supplies people and businesses higher awareness in as well as command with the
99% of items plus situations which continue to be beyond the access on the web. And
also by doing this, IoT enables people and businesses to become more attached to
the physical world around them as well as to perform much more significant, higher
level labor.

22.6.1 IoT Solves the Previously Discussed Problems

in the Following Ways

1. Enabling Preventive Maintenance of Equipment

This refers to regularly supervising the necessary information on the machinery’s

components such as:
• Multi-media filters
• Heat exchangers
• Sterilizers
• Air compressors
• pH probes
• Pressure gauges.
2. Controlling Drug Manufacturing Environment

Various sensors are enabled to supervise numerous environmental indicators in real-

time:
22 An Industrial Internet of Things Approach … 299

• CO2 level
• Radiation
• Humidity
• Light
• Temperature.
The lab supervisor or the attendant requires continuous and quick readings. So in
order to facilitate that, the dashboard needs to be regularly updated with data. Incor-
porated with an environment management process, this particular information may
also bring about automated changes. In the event of a catastrophe, like a poisonous
material leakage, the system shall notify/alert the staff persons to evacuate.
3. Facilitating Supply Chain Monitoring

IoT uses for supply chain management have grown to be preferred purchase aspects
for a lot of industries. Attached equipment, tracking men and material, sample life-
cycle management, smart packaging, along with cold chain overseeing are probably
part of the IoT programs which are especially suited for the pharma business. Ware-
housing is an important part of the pharma business. IoT programs are put into the
storage space and are also listed so as to understand as well as transmit informa-
tion that is crucial (product area, listing details) and also report back inconsistencies
or disparities, including missing products straight to factory managers’ hand held
devices/dashboards. Remedial methods occur in real time—considerably enhancing
the pace, reliability, and effectiveness on the picking operation [9].

22.6.2 Uses and Advantages of IoT in the Pharma Industry

1. Digital/Connected Factory

Equipments which are IoT compatible are able to transfer functional data in two
cases—to the associates such as original equipment manufacturing companies as
well as to the engineers belonging to technical field. This shall allow functioning
supervisors as well as factory heads to remotely deal with the factory devices and
also get the benefits of process automation and optimization [10].
2. Facility Management

The utilization of IoT receptors within production equipments allows condition based
alerts. There are lots of crucial equipments which are created to run within a specific
heat range as well as vibration ranges. IoT Sensors make an effort to check devices
as well as send out a notification whenever the tool deviates from its acceptable
range. By making sure the recommended functioning setting for machinery is set,
companies are able to save power, minimize expenses, get rid of machine downtime
as well as enhance operational effectiveness.
300 D. K. Sharma et al.

3. Plant Safety and Security

IoT blended big data evaluation is able to better the workers’ security and safety
inside the plant. By overseeing the primary key Performance Indicators (KPIs) of
safety and health, such as wounds as well as illness fees, near misses, long-term and
short-absences, automobile incidents as well as property harm or perhaps damage
during everyday activities. As a result, greater security is ensured by highly effective
monitoring systems [11].
4. Packaging Optimization

By utilizing IoT receptors in items and also product packaging, companies are able
to acquire insights into the usage patterns as well as managing of merchandise from
several buyers. Smart monitoring systems also can trace merchandise degeneration
during impact and transit of water, other environment variables and road.
5. Inventory Management

Keeping track of incidents throughout a supply chain is permitted by IoT programs.

Making use of the system, the listing is monitored as well as traced around the
world and also the consumers are informed of just about any substantial deviations
in the plans. This offers cross channel exposure directly into inventories as well
as administrators are supplied with practical estimates of accessible information,
ongoing work, plus estimated the arrival period of brand-new substances. In the end
this particular optimizes supply and also lowers costs that are shared within the value
chain.
6. Quality Control

IoT receptors accumulate certain product information along with other third party
widely distributed information from several phases of a product cycle. This particular
information pertains towards the make up of raw substances utilized, temperature
plus working atmosphere, waste materials, the effect of transportation etc. on the end
products. Every one of the inputs may afterwards be examined to recognize as well
as modify quality problems.
7. Production Flow Monitoring

In production flow monitoring, IoT makes it easier to keep track of manufacturing

right from the refining activity until the packaging of the items. Owing to this
sound monitoring and supervision of the system within real time makes it possible
to incorporate certain desired changes within operations for better and enhanced
control. Moreover, improved monitoring skills helps in getting rid of unwanted and
unnecessary workload and wastes of progress inventory.
22 An Industrial Internet of Things Approach … 301

8. Logistics and Supply Chain Optimization

The Industrial IoT (IIoT) is able to access real time supply chain data by super-
vising products, equipment, and materials as they move through the supply chain.
By networking plants with vendors, all of the people linked to the supply chain are
able to trace interdependencies, manufacturing cycle and material flow. This partic-
ular information shall aid companies foresee problems, bring down inventory as well
as possibly bring down capital demands [11].
One might come across unannounced challenges if we move from shipment to
transit and delivery into hospitals or pharmacies. The unexpected climate variations,
conveyance losses, vehicle accidents, along with any other unexpected delays can
harm pharma businesses as well as individuals [11].
To guarantee an undisturbed supply chain, pharma businesses must be concerned
about the drug’s route details. IoT is able to provide this particular remarkable expo-
sure into all procedures in real time, permitting quick actions as well as cutting short
waiting times [12].
Every program can be marked with smart RFID labels and tags for facilitated
identification, beginning with the packing of batches as well as mailing them off.
Every batch is going to be distinguished; hence their final destinations are very easily
identified [12].
The GPS embedded automobiles will constantly update their locations throughout
the transit, improving the visibility of shipment. In order to exclude hazardous
breakdowns, extra sensors are handy.
When the compound depends on temperature, smart receptors for ambience detec-
tion might be loaded also, and the exact same is followed with humidity and pressure.
Logging the present circumstances as well as evaluating them with the necessary
variable, the receptors are able to alert the carrier immediately.

22.7 Architecture of IoT

The IoT architecture is broadly categorised into 3 parts: the perception layer, the
networking or transport layer and the service or application layer.
Perception layer: It behaves as the main layer in IoT. This layer is also called data
extraction layer. Various forms of information and data is accumulated from physical
devices, by the technologies of RFID tags, sensors, reader-writers and more [13].
Network layer/Transport layer: Next, the data that is collected from the data extrac-
tion layer is then sent to the next layer, i.e. the network layer. This layer is also called a
Transport layer. Enhancement of various technologies happens via the network layer,
such as mobile communication network, GSM (global system for mobile communi-
cations), WSN (wireless sensor network), GPRS (general packet radio service) and
other communications equipment. This layer is efficient and reliable and provides
trusted networks to large scale industries [13].
302 D. K. Sharma et al.

Service layer/Application layer: The data that is collected from the network layer
is then managed and stored or kept in the service layer, which is also referred to
as the Application layer. Now this layer mainly contains two different sub-layers
viz. data management sub-layer and application service sub-layer. Complex data
and unreliable info such as screening, restructuring, cleaning and combining are
processed by the data management sublayer. A directory service such as QoS (Quality
of Service) or M2M (market to market service) is provided to this layer. A good user
information is provided for end users and applications by the collected data, which
is converted or translated into meaningful information of content [13].

22.8 Challenges and Drawbacks of IoT

Despite the profound changes in the pharmaceutical industry in the last 15 years, the
industry has also been faced with certain challenges. There has been an increasing
amount of pressure on the pharmaceuticals in terms of quality adherence, cost of
drug development, demand, competitiveness and reduced time to market. Amongst
all these factors, companies need to primarily focus on the difficulties faced in the
areas of R&D, manufacturing and distribution.

22.8.1 R&D

In order to lessen the amount of human error, waste and to increase efficiency, IoT
analytics are used in the department of R&D, as shown in Fig. 22.3. It is now easier
and faster to get approval for new drugs to be launched in the market due to improved

Fig. 22.3 Steps involved in manufacturing and selling of drugs in R&D

22 An Industrial Internet of Things Approach … 303

real-time visibility, increased availability of data during toleration and increased

efficiency. Clinical trial intervals are now shorter and optimised.

22.8.1.1 R&D Challenges

1. The processes in the R&D department are extremely people-intensive. This may
increase the possibility of human error.
2. It takes a large amount of time for developing new drugs as well as getting the
current drugs validated for other uses/purposes.

22.8.2 Manufacturing

The concept of IoT in the manufacturing department of the pharmaceutical industry

introduces certain benefits like improved stock control, quality and processes. All of
these factors are capable of increasing efficiency and thus, yield. Figure 22.4 explains
this. IoT is extremely important in identifying machines and equipments that are at a
risk of breaking down and might need potential maintenance. Maintenance can then
be scheduled for a better time [14].

22.8.2.1 Manufacturing Challenges

1. Due to potential adulteration and sourcing, the quality of raw materials is variable.
2. The quality of the end product might be affected due to damage or theft of raw
materials which could take place during storage or in transit.

Fig. 22.4 Flowchart depicting manufacture of drugs

304 D. K. Sharma et al.

Fig. 22.5 Distribution of drugs in the market

3. The quality of the product may be inconsistent due to the raw material quality
being uncertain. At the same time, this inconsistency could also be due to changes
brought about in the manufacturing processes.
4. There is a high possibility of machine downtime due to breakdowns.

22.8.3 Distribution

In this category, there is improved stock management which leads to a lower inventory
as well as less wastage. This means that the storage costs are potentially lower and
there is better space in warehouses. The concept of IoT has immensely improved
transit visibility as supply chain. Thus, supplying drugs to patients has now become
easier. Advantages of this technology also includes improved security, which makes
it tougher to counterfeit drugs. The pharma companies’ investments in the R&D
department is also protected at length (Fig. 22.5).

22.8.3.1 Distribution Challenges

1. There could be damage and/or theft of drugs in storage or transit.

2. It becomes hard to trace drugs across the supply chain due to limited visibility.
3. There could be an increase in the levels of counterfeit drugs which might in turn
reduce profits.
4. The prescribed drugs or medicines might be unavailable in pharmacies/hospitals
[15, 16].
Apart from the above mentioned challenges, there are a few more difficulties faced
by IoT. These problems are mentioned below:
• Obstacles on the technical front.
• Protection of safety and privacy.
• Standardisation.
• IoT is not internationally compatible.
• Their highly complex nature may result in failure.
• Privacy and security violation are two important factors to be considered in IoT.
• Lessened user safety.
• Job reductions become prevalent due to decrease in manual tasks.
22 An Industrial Internet of Things Approach … 305

• Increase in the AI related technology as time passes on could lead to IoT taking
over life.

22.9 Applications of IoT

Leading pharmaceutical businesses are focussing on smart gadgets to regulate medi-

cations and overlook the impact they have on patients. This includes delivering pills
consisting of medical monitors. One of the uses is to check whether or not patients,
especially the one with memory lapses, are taking their medicines on time. If the
patient forgets to do so, a reminder is sent on their gadget regarding the same. If they
fail to keep up with the schedule, it will alert the doctor to step in.
Externally wearable IoT-connected gadgets along with those that can be ingested
can concurrently gather records and transfer data to integrated software that gives
us results/outcomes which precisely help the doctors figure out the medical condi-
tions. RFID is a successful technological step that has benefited humans greatly and
is widely accepted across the world. The management system of hospitals can be
improved drastically by introducing RFID technology, which not only enhances effi-
ciency but also takes care of the safety angle. RFID technology is used to tag the
following:
1. Non-living things: things used in the medical field such as a wheelchair
2. Living things: patients, nurses and staff
3. Non-living things related to living things: this links the first two categories (e.g.
patients use wheelchairs) [17].
A perfect example of this situation is an RFID tag. These tags are made to be worn
by the patients. Also, there are bar-codes on the packets of medicines prescribed to
them which must be strictly adhered to. It is extremely important to ensure that
exactly the right dose has been consumed by the patient and only that particular
medicine which has been prescribed to him [17].
Patients’ medical information is recorded electronically. IoT-systems are used to
exchange health care systems so as to gather a patient’s medical record. This enables
identification of things based on RFID—such as medical equipment, patients and
clinical staff to retrieve medical information from numerous healthcare systems, and
show significant information to the authorised people including doctors [17].
Internet of Things has also been extremely useful in storing temperature-sensitive
products.
Active and passive temperature loggers are connected to refrigerators in ware-
houses and at other sites to record temperature in continuous intervals. With the help
of IoT technology, the devices are linked together to compare their measurements
against thermostability tables, and generate alerts in case the temperature varies [18].
306 D. K. Sharma et al.

22.10 Future of IoT

A positive growth, though at moderate levels, was expected in the performance of

the global pharmaceutical industry in the year 2019 and after. Development and
enhancement of IoT standards is of particular interest to various organizations and
countries as it is expected to bring about profound economic advantages and benefits
in the future. The uses of IoT for a brighter and better future can be seen in Fig. 22.6.
In the present day, various organizations such as the China Electronics Stan-
dardization Institute, International Telecommunication Union, IEEE and American
National Standards Institute have been trying to work on the enhancement and devel-
opment of numerous IoT standards. This process, in turn, shall speed up the outreach
of this new smart technology as along with the state-of-the-art innovations in the
pharmaceutical industry.
The network platform is evolved or developed and modified by evolving these
technologies. This platform shall fundamentally modify the manner in which our
societies will collaborate, innovate, produce and live sustainably. Right from identi-
fying difficulties on the production line to supervising conditions along the supply
chain so as to improve outcome of patients, this new technology is transforming the
manner in which production of drugs takes place. It has also brought about immense
changes and modifications in the way these drugs are delivered and consumed [19].
Numerous countries like US, South Korea and China amongst many others are
leading investors in the area of this new age technology of Internet of Things [20].

Fig. 22.6 IoT being used in various fields

22 An Industrial Internet of Things Approach … 307

1. The global giant by the name of United States of America sits atop the Global
Connectivity Index. This is due to the numerous developments and modifications
that have literally redefined the country’s digital economy, developments such as
the IIoT. This is a sub-category of IoT which refers to the data generated by the
industrial grade equipment. It has been predicted that the market shall surpass £
1.19 trillion by the year 2020. This shall create a huge window of opportunity
for businesses to utilise this tech boom to their own advantages which could be
digital, as well as economical. Businesses across the US have benefitted from IoT
as their productivity and longevity have increased and thus, lives could potentially
be changed.
2. Sweden has been declared as the world’s most cashless society as this mode of
transaction is used in less than 20% of the in-store transactions. In order to show
their support for their government’s initiative to create a safer society, shoppers
have instead opted for contactless cards or wearables. Thus, we can see that
gradually, IoT is taking over in Sweden while banknotes are being left behind.
3. In order to help commuters and travellers navigate, more than sixty Bluetooth
devices have been installed in the station of the city of Tokyo. These devices
connect to users’ smartphones with the help of an application, or app as it is
more commonly known, so as to determine their positions. With the help of
Augmented Reality (AR) digital arrows are superimposed on the app’s real-time
map and gives correct directions to the desired location to the commuters. Thus,
IoT in this capital city is proving to be not only a step for technology but a step
for city management too.
4. In South Korea, people can check the contents of their fridge, download recipes
of dishes and even give commands to their washing machines with the help
of a simple text. Irrespective of their distance from these individual machines
and homes, these commands can be given from anytime, anywhere. Thus, South
Korea is the epitome of how governmental initiatives can provide a huge window
of opportunity for IoT.
Other industries are increasingly becoming technology based and are incorpo-
rating state-of-the-art and up-to-date technologies and equipment within themselves.
In order to stay in the race and continue to be competitive and relevant, pharma indus-
tries must step up their game and move towards constant innovations and creations.
This is achievable by adopting IoT technology. Pharma companies must quicken
their implementation of IoT as the number of software products and items that are
IoT ready is only increasing [21].

22.11 Conclusion

• IoT could prove to be a game-changer in the industrial world as it is capable of

modifying orthodox and actual physical information or data which may include
308 D. K. Sharma et al.

engineering infrastructure too. This could have a direct and strong impact on
productivity and innovation [22].
• Internet of Things combines various products designed with profound social
networking abilities. In fact, sensors and actuators are becoming increasingly
popular due to their cost effectiveness and small size [22].
• The industrial world is increasingly deploying IoT products or devices so as
to contribute towards the growth of manufacturing uses such as automated
monitoring systems, management, control and maintenance [23].
• The pharmaceutical industry is highly suited for accepting IoT and this could
happen in the patient or industrial space [21].
• There is a need for innovation and utilisation of existing technologies in the most
efficient manner possible so as to increase profits, competition and relevance.
This is what the pharma industry must look into. In order to meet international
standards and fulfill market requirements and demands, pharmaceuticals must
implement the concept of IoT in their working.
• Quality and agility are amongst the many advantages offered by IoT in the indus-
trial world. This concept is also responsible for adding value to businesses and
can thus play a leading role in the new era of transformation [24].
• Implementation of smart pharma IoT technology can lead to the achievement of
lower costs, increased efficiency, better manufacturing processes and new methods
of treatment in the area of healthcare.
• Scientists are aiming towards combining or integrating RFID technologies and
multi-agent into a single IoT platform. In order to aid users in different activities
consisting this IoT platform, some agents will be developed/manufactured based
on certain preferences, especially taking safety and security into consideration
[25].
• IoT is capable of dealing with individuals’ requirements as well as those of the
healthcare suppliers by using current structures and systems, but at a fair price.

References

1. Jia, X., Feng, Q., Fan, T., Lei, Q.: In: 2012 2nd International Conference on Consumer
Electronics, Communications and Networks (CECNet), Yichang, China, Apr 2012
2. Angelopoulos, G., Kalampokis, G.T., Dasygenis, M.: An internet of things humanoid robot
teleoperated by an open source Android application. In: 2017 Panhellenic Conference on
Electronics and Telecommunications (PACET), Xanthi, Greece, 17–18 Nov 2017
3. Xia, F., Yang, L.T., Wang, L., Vinel, A.: Internet of things. Int. J. Commun. Syst. (2012)
4. Gubbi, J., Buyya, R., Marusic, S., Palaniswami, M.: Internet of things (IoT): a vision,
architectural elements, and future directions. Future Generat. Comput. Syst. 29(7), 1645–1660
5. McClelland, C.: How Does an IoT System Actually Work? 6 Jan 2019. https://www.iotforall.
com/iot-explained-how-does-an-iot-system-actually-work/
6. IoT & RFID in Inventory Management. Retrieved from https://nanoinformatics.in/iot-in-inv
entory-management-using-rfid-technology/
7. Geetha, S., Cicilia, D.: In: 2017 2nd International Conference on Communication and
Electronics Systems (ICCES), Coimbatore, India, 19–20 Oct 2017
22 An Industrial Internet of Things Approach … 309

8. Sarma, S.E., Weis, S.A., Engels, D.W.: Massachusetts Institute of Technology Cambridge, MA
9. What technologies are used in IoT—technology behind internet of things. Internet of Things.
11 June 2019. Retrieved from https://www.avsystem.com/blog/iot-technology/
10. Bhavnani, S.P., Narula, J., Sengupta, P.P.: Mobile Technology and the Digitization of
Healthcare. USA (2016)
11. Ting, S.L., Kwok, S.K., Tsang Albert, H.C., Lee, W.B.: Enhancing the information transmission
for pharmaceutical supply chain based on radio frequency identification (RFID) and internet of
things. In: 2010 8th International Conference on Supply Chain Management and Information,
Hong Kong, China
12. Pachayappan, M., Rajesh, N., Saravanan, G.: Smart logistics for pharmaceutical industry based
on internet of things (IoT). Int. J. Comput. Sci. Inf. Secur. (IJCSIS) (2016)
13. Formisano, C., Pavia, D., Gurgen, L., Yonezawa, T., Galache, J.A., Doguchi, K., Matranga, I.:
The advantages of IoT and cloud applied to smart cities. In: 2015 3rd International Conference
on Future Internet of Things and Cloud, Rome, Italy
14. Kumar, N.M., Mallick, P.K.: The internet of things: insights into the building blocks, component
interactions, and architecture layers. Procedia Comput. Sci. 132, 109–117 (2018)
15. Rantanen, J., Khinast, J.: The future of pharmaceutical manufacturing sciences. J. Pharm. Sci.
104(11), 3612–3638 (2015)
16. Mopidevi, R.: IoT use cases in pharma span the spectrum, 8 July 2019. https://blogs.sas.com/
content/hiddeninsights/2019/07/08/iot-use-cases-in-pharma-span-the-spectrum/
17. Turcua, C.E., Turcua, C.O.: Internet of things as key enabler for sustainable healthcare delivery.
Procedia Soc. Behav. Sci. 73, 251–256. The 2nd International Conference on Integrated
Information, Suceava, Romania (2013)
18. Vilamovska, A.M., Hatziandreu, E., Schindler, R., Oranje, C., Vries, H., Krapels, J.: Study on
the requirements and options for RFID application in healthcare, July 2008. https://ec.europa.
eu/information_society/activities/health/docs/studies/200807-rfid-ehealth.pdf (2008)
19. Lee, I., Lee, K.: The Internet of Things (IoT): Applications, Investments, and Challenges for
Enterprises. USA (2015)
20. Balas, V.E., Solanki, V.K., Kumar, R.: Chapter 4, 4.2.2, Internet of things-based pharmaceutics
data analysis. In: Emergence of Pharmaceutical Industry Growth with Industrial IoT Approach,
p. 103. Academic Press (2019)
21. Draft Policy on Internet of Things: Ministry of Electronics and Information Technology, New
Delhi (2015)
22. Sharma, U.: Transforming pharma logistics with IOT. 3 June 2019. Retrieved from https://
www.expresspharma.in/pharma-logistics/transforming-pharma-logistics-with-iot/
23. Medagliani, P., Duda, A., Leguay, J., Rousseau, F.: Internet of Things Applications—From
Research and Innovation to Market Deployment, Jan 2014
24. Ngu, A.H., Gutierrez, M., Metsis, V., Nepal, S., Sheng, Q.Z.: IoT Middleware: a survey on
issues and enabling technologies. IEEE Internet Things J. 4(1) (2017)
25. Mahmoud, R., Yousuf, T., Aloul, F., Zualkernan, I.: Internet of things (IoT) security: current
status, challenges and prospective measures. In: 2015 10th International Conference for Internet
Technology and Secured Transactions (ICITST), London, UK
 Part IV
Cyber Physical System Framework
and Applications
Chapter 23
Solving the Problem of Fuzzy
Transportation Using Linear
Programming and Goal Programming

Hasanain Hamid Ahmed

Abstract Many companies are interested in the process of moving materials or

equipment from the source site to the site of demand and the cost and time of arrival
and the process of transfer is called multi-objective transport, in this research we
introduced multiple linear models fuzzy objectives as both the objective functions
and variables coefficients (time) fuzzy numbers Trigonometric function was used to
convert the fuzzy numbers to the normal formula and then construct the mathematical
model and solve it using the goal programming method. A multi-objective model for
the gasoline product was achieved in which the cost and fuzzy time were reduced.
What linear programming method and programming goals and reached the most
important recommendations, including when solving the transport model using Goal
programming helps decision-makers to reach a solution.

Keyword Fuzzy transport · Objective programming · Transport problem

23.1 Introduction

Communities and countries have taken care of transport problems in the past and
have always solved these problems by planning for these deviations in transport. It
started to study transport planning and try to simulate this problem by using the best
methods that solve the transport problems which are transporting similar materials
from Asset which is the production or marketing center or any center from which
the goods are transferred to (the ends), which is the demand center, the consumption
center, or any center to which the goods are sent (with the lowest costs, increase
profits or the shortest possible time) and by identifying this problem of Transport
facing the General Company for the distribution of petroleum products and represent
it mathematically By planning to solve this problem and the best ways related to the
study, the city of Baghdad and other provinces in recent times suffers from the

H. H. Ahmed (B)
Department of Administration and Economics, College of Imam, Kadhim University of Science,
Baghdad, Iraq
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 313

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_23
314 H. H. Ahmed

frequent traffic congestion that affected the movement in the city and congestion in
the roads linking the provinces The congestion affected the movement of marketing
and the arrival of raw materials and marketing of products to and from warehouses
and refineries and therefore the impact on the Iraqi consumer, which is the goal and
basis of the planning process for this we must study and develop a transport model
with the objectives The aim of this research as well as the organization of the flow of
petroleum products (gasoline) to the warehouses and marketing the outputs of those
warehouses to the consumption centers (gas stations), and at the lowest possible cost
and as soon as possible.

23.2 Concept of the Transport Problem

The problem of transportation is important mathematical methods that help in the

process of making the right decision where.
The model aims to find an optimal method for distributing (transporting or ship-
ping) materials or goods from the sources of supply to the centers of consumption
or demand (stores or customers) so that the total cost of transport is minimal.

23.3 Methods of Solution of Transport Model

Transport can be analyzed to determine the optimal quantities to be transferred from

sources to sites and at the lowest possible cost of transport using linear programming
[1]. However, due to the special nature of transport, new methods have been developed
that have special features that make them more appropriate when analyzing the
general method [2]. For the general method in terms of mathematical processing, but
agree with it in principle, where it begins with the choice of the primary acceptable
solution (SBFS) and then develops the solution to reach the optimal solution that has
the value of the objective function at the lower end [3].
There are several ways to solve the transport model Two methods were selected
is:
3.1 Linear programming method.
3.2 Goal Programming Method.

23.4 Concept of the Goal Programming

Goal programming has been defined by (Cromero) as a mathematical method that

tends to be flexible and realistic in solving complex decision issues that take into
account several goals, variables, and constraints [4].
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 315

Blaid also defined that the goal programming model is a model that takes into
account several objectives at once and that is under the framework of choosing the
optimal solution among the possible solutions [5, 6].

23.5 Methods of Solving Goal Programming

There are two main methods used to find a solution in the issue of multi-goal
programming is:
Method of weights.
Method of Priority.
That the two methods do not necessarily lead to the same results and solution, but
each method achieves what the decision-maker wants and according to his desire and
there is no better way than the other method as the method of each method depends
on the desire of the decision-maker [7].
The first step in the classification of goals is to determine the absolute goals
and then we seek to arrange other (non-absolute) goals according to importance and
preference. Once the targets are classified, they should be grouped with a few priority
levels as possible except for absolute targets. All targets within one level must have
one unit of measurement, and weights can be specified for each target within any
group (ie, priority levels) if they are (4G) [8]. For example, targets within the same
level of priority should be assigned a special weight for each so that weights are
positive numerical values. These weighted values may be derived or can be inferred
from a particular judgment or decision. For example, we say that 3G is three times
more important than 4G, so we are interested in reducing the deviation of (G3) by
three times more than the deviation of (4G).
In other words, the method of priorities depends on the order of objectives and the
importance of the decision-maker, where the solution is based on one goal, which
has the highest priority and then beyond, and so that the goals of high priority are
not compromised to calculate the goals of low priority.
The method of priorities and the method of weights do not necessarily give the
same results because the steps of each method are different from the other and leave
the decision-maker to choose the appropriate method [9].

23.6 Fuzzy Sets

In 1965, the fuzzy sets Zadeh is defined as a class of elements with a certain degree
of affiliation. This group expresses a distinct membership function if each element
has a degree of affiliation that falls between two values (1,0). When the element takes
a degree of affiliation (1) means that the element belongs to the fuzzy group, and
when the top of the affiliation element (0) means that element does not belong to
316 H. H. Ahmed

the fuzzy group, and if the degree of affiliation element (0.5) This means that the
element belongs to the degree (0.5) This element does not belong to the fuzzy group
by the same amount Equilibrium point [10]. If the degree of affinity of the element
(0.9), this means that the element belongs to the fuzzy group (0.9) and does not
belong to the fuzzy group (0.1) and is closer to belonging. (Zimmerman) The world
has defined (zimmerman) fuzzy group and the definition is the most accurate of the
definitions and the definition indicates that if a group (A) is the group of elements
that we symbolize by the symbol (x), then the pairs arranged from the group is a
fuzzy group [11].

A = {x, μa(x) /x ∈ X }

(μ X ) represents the function of affiliation.

23.6.1 Functions Affiliation to the Fuzzy Group

The functions of affiliation are of great importance in the theory of fuzzy group, the
functions through which they are conducted [12].
Determine the degree to which elements affiliate to the fuzzy group. These
functions are used to determine how they affiliate.
An element of the fuzzy group [13]. Affiliation functions represent x-axis x repre-
sents the normal values of the variable and the y-axis, which represents the degree
of affiliation to the variable, and is their range is between zero and one. This means
that when the value of a variable is equal to one, the value certainly affiliates.
When the value of a variable is zero, the value does not affiliate to the set and the
values between zero and one, the ratio of the element to the fuzzy group [14].

23.6.2 Linear Affiliation Function

23.6.2.1 The Linear Affiliation Function Can Be Defined as Follows

⎧
⎪
⎪ 1 i f F k (x) ≥ U k
⎪
⎨
  U k − F k (x)
µ Fk (x) = i f L k < F k (x) < U k (23.1)
⎪
⎪ U k − Lk
⎪
⎩
0 other wise

where
 k 
μ F (x) represents the linear affiliation function.
Uk represent the upper limit.
Lk represent the lower limit [15].
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 317

23.6.2.2 Mathematical Model of Fuzzy Transport Problem (Linear

Affiliation Function)
Max = λ, 0 ≤ λ ≤ 1
S.to
F k (x) + λ(Uk − L k ) ≤ Uk (23.2)


n
X i j = ai , i = 1, 2, . . . , m
j=1

m
X i j = b j , j = 1, 2, . . . , n X i j ≥ 0 ∀i, j
i=1

where λ represents the satisfaction of the decision-maker to achieve the objectives

and the value between zero and one [16].

23.7 Robust Ranking Method

A mathematical method converts the fuzzy formula based on the form of human
intuition to linear formula and if it is.
ã fuzzy number The Robust ranking technique is defined.
As follows [17]:

1
R(ã) = ∫(0.5)(α L αU )dα. (23.3)
0

where
αU , α L The upper and lower limit of the fuzzy period at α [18].
The following example illustrates the robust function if we take the next fuzzy
period (5,10,15) the robust function is written as follows [19]:

1
R(ã) = ∫(0.5)(α L αU )dα, (23.4)
0

(α L αU ) = {(b − a)α + α, c − (c − b)α} (23.5)

1
R(5,10,15) = ∫(0.5)(5α + 5,15 − 5α)dα, (23.6)
0

1
R(5,10,15) = ∫(0.5)(20)dα = 10. (23.7)
0
318 H. H. Ahmed

23.8 Mathematical Model of the Problem

of Multi-Objective Fuzzy Transportation
Min = d1+ + d2+

S.T.


m 
n
Ci1j Yi j − d1+ + d1− ≤ F1 (y), (23.8)
i=1 j=1


m 
n
Ci2j Yi j − d2+ + d2− ≤ F2 (y). (23.9)
i=1 j=1


n
X i j = ai , i = 1, 2, . . . , m
j=1

m
Xi j = b j , j = 1, 2, . . . , n
i=1
m 
n
ai = bj
i=1 j=1

X i j ≥ 0 ∀i, j

where
Ci1j represents the cost and time per unit of quantities transferred [20].
Xi j represents the quantities transferred from producer i to requesting j.
ai represents the supply.
bj represents the demand.
d +, d − represents negative and positive deviations from achieving goals [21].

23.9 Data Description

The data included three warehouses for gasoline storage of the province of Baghdad
belonging to the General Company for Distribution Petroleum products and contain
petroleum derivatives (gasoline) transported by a single transport, namely(Tanks
with a load of 36 thousand liters) from the three warehouses (Dora, Karkh, Rusafa)

Table 23.1 Shows the

Rusafa Karkh Dora
capacity (m3 ) of the
warehouses (Bi) 22,342 10,000 17,574
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 319

Table 23.2 Shows the monthly demand quantities (m3 ) for the stations (Aj)
Section 77 Donya Noor Tarasban Madaen Somod Benog Jabha Mashtal Kanat Station
4034 4480 3379 4249 3236 4820 5305 4945 4387 4297 Supply

to stations Fuel filling in Baghdad on the sides of Karkh and Rusafa and Table 23.1
the following shows the names of warehouses Its carrying capacity:
Through interviews with some of the specialists in the Baghdad Distribution
Authority were obtained names the ten important filling stations on both sides of
Karkh and Rusafa, which are filling stations Table 23.2 shows the names of the
stations and the demand Monthly (gasoline).
There is a specialized committee in the General Company for the distribution of
petroleum products to measure distances between warehouses. The filling stations are
measured in kilometers and Table 23.3 shows the distances between the warehouses
Stations in Kilometers:
The cost of transporting petroleum products from warehouses to filling stations
was calculated based on The distance between them is measured in units (dinars
per m3 ), as the formula adopted in the company General distribution of petroleum
products according to the following formula:

Cost of transportation(dinar) = distance(km) ∗ price of transport in dinars

where the price of transporting gasoline per cubic meter is 51.5 (in thousands)
dinars and Table 23.4 explains the cost of transporting gasoline from warehouses to
stations.

Distance ∗ Transfer price → 17 ∗ 51.5 = 875.5

Source/with the appreciation of specialists from the transport department in the

company in addition to (Google Map). From Table 23.5 of the previous fuzzy times
1
with trigonometric numbers we apply equation R(ã) = 0 (0.5)(α L αU )dα on all
cells in the table and Table 23.6 becomes as follows:

23.9.1 Building the Mathematical Model of the Fuzzy

Transport Problem

Step 1: Eliminate fuzzy,it means apply the linear affiliation function to the cost and
time matrix as follows:
 320

Table 23.3 Shows the distances (km) between stations and warehouses
Station Kanat Mashtal Jabha Bonog Somod Madaen Tarasben Noor Donya Section 77
Distance Dora 17 23 14 31 32 36 22.5 41 12 66
Karkh 58 59 55 72 65 77 48.5 66 37 106
Rusafa 22 19 27 36 41.5 51 59.5 78 42 108
H. H. Ahmed
Table 23.4 Shows the cost of transporting gasoline (Cij ) in thousands
Station Kanat Mashtal Jabha Bonog Somod Madaen Tarasben Noor Donya Section 77
Distance Dora 8755* 11,845 721 15,965 16,480 36 18,540 21,115 618 12,875
Karkh 29,870 30,385 28,325 37,080 33,475 77 39,655 33,990 19,055 28,840
Rusafa 11,330 28,325 13,905 18,540 213,725 51 26,265 40,170 21,630 30,385
23 Solving the Problem of Fuzzy Transportation Using Linear Programming …
321
322 H. H. Ahmed

Table 23.5 Shows the time

Rusafa Karkh Dora Stations
of fuzzy transportation from
warehouses to filling stations (9, 16, 10) (9.5, 14, 11.5) (21.5, 24, 25) Kanat
(21.5, 20, 23) (8.5, 14, 9.5) (9, 14, 11) Mashtal
(17, 23, 19) (20, 23, 21.5) (25, 28, 30) Jabha
(8, 12, 9) (10, 18, 15) (28, 31, 29) Bonog
(50, 48, 53) (38, 46, 50) (49, 51, 55) Somod
(34, 39, 40) (32, 38, 40) (39, 43, 45) Madaen
(17, 18, 21) (9, 14, 12) (11, 14, 13) Tarasban
(17, 19, 18) (11.5, 15, 18) (15, 18, 20) Noor
(20, 21.5, 23) (9, 10.5, 12) (35, 41, 46) Donya
(19, 22, 21) (21.5, 25, 28) (40.5, 37, 44) Section 77

⎧
⎪ 1 k
i f F(x) ≥ Uk
⎪
⎪
 k  ⎨
Uk − F(x)
k
μ F(x) = i f L k < F(x)
k
< Uk . (23.10)
⎪
⎪ U − L
⎪
⎩
k k
0 other wise

 1  113,222,400 − F(x)
1
μ1 F(x) = (23.11)
113,222,400 − 69,536,980

 2  960,580 − F(x)
2
μ2 F(x) = (23.12)
960,580 − 395,200

Step 2: Build a mathematical model for the problem of fuzzy transport

Max = λ, 0 ≤ λ ≤ 1
S. to

Constraint of Linear affiliation function (cost constraint)

113,222,400 − F 1 (x)
≥λ (23.13)
113,222,400 − 69,536,980


113,222,400 − F 1 (x)
≥ λ ∗ 43,685,420 (23.14)
43,685,420

113,222,400 − F 1 (x) ≥ 43,685,420λ (23.15)


−F 1 (x) − 43,685,420λ ≥ −113,222,400 ∗ (−1) (23.16)

F 1 (x) + 43,685,420λ ≤ 113,222,400 (23.17)

Table 23.6 Shows the transfer time after using the robust ranking function of the gasoline product
Station Kanat Mashtal Jabha Bonog Somod Madaen Tarasben Noor Donya Section 77
Distance Dora 24 14 28 31 51 43 14 18 41 37
Karkh 14 14 23 18 46 38 14 15 10.5 25
Rusafa 16 20 23 12 48 39 18 19 21.5 22
23 Solving the Problem of Fuzzy Transportation Using Linear Programming …
323
324 H. H. Ahmed

875.5X 11 + 1184.5X 12 + 721X 13

+ 1596.5X 14 + 1648X 15 + 1854X 16
+ 1158.75X 17 + 2111.5X 18
+ 618X 19 + 1287.5X 1,10 + 0X 1,11 + 2987X 21
+ 3038.5X 22 + 2832.5X 23
+ 3708X 24 + 3347.5X 25 + 3965.5X 26
+ 2497.75X 27 + 3399X 28 + 1905.5X 29
+ 2884X 2,10 + 0X 2,11 + 1133X 31
+ 2832.5X 32 + 1390.5X 33 + 1854X 34
+ 2137.25X 35 + 2626.5X 36
+ 3064.25X 37 + 4017X 38 + 2163X 39 + 3038.5X 3,10
+ 0X 3,11 + 43, 685, 420λ ≤ 113,222,400

Constraint of linear affiliation function (Time constraint)

960,580 − F(x)
2
≥λ (23.18)
960,580 − 395,200
 
960,580 − F(x)
2
≥ λ ∗ 565,380 (23.19)
565,380

960,580 − F(x)
2
≥ 565,380λ (23.20)


−F 2 (x) − 565,380λ ≥ −960,580 ∗ (−1) (23.21)

F 2 (x) + 565,380λ ≤ 960,580 (23.22)

24X 11 + 14X 12 + 28X 13 + 31X 14

+ 51X 15 + 43X 16 + 14X 17 + 18X 18
+ 41X 19 + 37X 1,10 + 0X 1,11
+ 14X 21 + 14X 22 + 23X 23 + 18X 24 + 46X 25
+ 38X 26 + 14X 27 + 15X 28
+ 10.5X 29 + 25X 2,10 + 0X 2,11 + 16X 31
+ 20X 32 + 23X 33 + 12X 34 + 48X 35
+ 39X 36 + 18X 37 + 19X 38 + 21.5X 39
+ 22X 3,10 + 0X 3,11 + 565,380λ ≤ 960,580
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 325

Capacity constraints

X 11 + X 12 + X 13 + X 14 + X 15 + X 16
+ X 17 + X 18 + X 19 + X 1,10 + X 1,11 = 17574
X 21 + X 22 + X 23 + X 24 + X 25 + X 26
+ X 27 + X 28 + X 29 + X 2,10 + X 2,11 = 10000
X 31 + X 32 + X 33 + X 34 + X 35
+ X 36 + X 37 + X 38 + X 39 + X 3,10 + X 3,11 = 22342

Supply constraints

X 11 + X 21 + X 31 = 4297
X 12 + X 22 + X 32 = 4387
X 13 + X 23 + X 33 = 4945
X 14 + X 24 + X 34 = 5305
X 15 + X 25 + X 35 = 4820
X 16 + X 26 + X 36 = 3236
X 17 + X 27 + X 37 = 4249
X 18 + X 28 + X 38 = 3379
X 19 + X 29 + X 39 = 4480
X 1,11 + X 2,11 + X 3,11 = 6784

23.9.1.1 Solve the Mathematical Model

After solving the mathematical model using linear programming in the program
(Win-QSB) was The results are as follows:

X 12 = 4387, X 17 = 4249,
X 18 = 2154, X 1,11 = 6784, X 25 = 4295
X 28 = 1225, X 29 = 4480,
X 31 = 4295, X 33 = 4945, X 34 = 5305
X 35 = 525, X 36 = 3236,
X 3,10 = 4034, λ = 0.08,
1
F(x) = 85,202,476, F(x)
2
= 908,928
326 H. H. Ahmed

23.9.1.2 Build the Mathematical Model of the Problem of Transport

Fuzzy Multi-objectives
Min = d1+ + d2+

Goal 1: The goal of reducing the cost of transport

875.5X 11 + 1184.5X 12 + 721X 13

+ 1596.5X 14 + 1648X 15 + 1854X 16
+ 1158.75X 17 + 2111.5X 18
+ 618X 19 + 1287.5X 1,10 + 0X 1,11 + 2987X 21
+ 3038.5X 22 + 2832.5X 23
+ 3708X 24 + 3347.5X 25 + 3965.5X 26
+ 2497.75X 27 + 3399X 28 + 1905.5X 29
+ 2884X 2,10 + 0X 2,11 + 1133X 31
+ 2832.5X 32 + 1390.5X 33 + 1854X 34
+ 2137.25X 35 + 2626.5X 36
+ 3064.25X 37 + 4017X 38 + 2163X 39
+ 3038.5X 3,10 + 0X 3,11 − d1+ + d1− = 113,222,400

Goal 2: The goal of reducing the time of transport

24X 11 + 14X 12 + 28X 13 + 31X 14

+ 51X 15 + 43X 16 + 14X 17 + 18X 18
+ 41X 19 + 37X 1,10 + 0X 1,11 + 14X 21
+ 14X 22 + 23X 23 + 18X 24 + 46X 25
+ 38X 26 + 14X 27 + 15X 28
+ 10.5X 29 + 25X 2,10 + 0X 2,11 + 16X 31
+ 20X 32 + 23X 33 + 12X 34 + 48X 35
+ 39X 36 + 18X 37 + 19X 38 + 21.5X 39
+ 22X 3,10 + 0X 3,11 − d2+ + d2− = 960,580

Capacity constraints

X 11 + X 12 + X 13 + X 14 + X 15 + X 16
+ X 17 + X 18 + X 19 + X 1,10 + X 1,11 = 17574
X 21 + X 22 + X 23 + X 24 + X 25
+ X 26 + X 27 + X 28 + X 29 + X 2,10 + X 2,11 = 10000
X 31 + X 32 + X 33 + X 34 + X 35
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 327

Table 23.7 Shows the results

Results in liters Variables Results in liters Variables
of Mathematical model
parameters 460,818 X 33 4387 X 12
5305 X 34 33,662 X 13
525 X 35 4820 X 15
4249 X 37 124,618 X 16
3379 X 38 6784 X 1,11
76,682 X 39 4297 X 21
4034 X 3,10 198,982 X 26
12,975,616 d1− 371,318 X 29

+ X 36 + X 37 + X 38 + X 39 + X 3,10 + X 3,11 = 22342

Supply constraints

X 11 + X 21 + X 31 = 4297
X 12 + X 22 + X 32 = 4387
X 13 + X 23 + X 33 = 4945
X 14 + X 24 + X 34 = 5305
X 15 + X 25 + X 35 = 4820
X 16 + X 26 + X 36 = 3236
X 17 + X 27 + X 37 = 4249
X 18 + X 28 + X 38 = 3379
X 19 + X 29 + X 39 = 4480
X 1,10 + X 2,10 + X 3,10 = 4034
X 1,11 + X 2,11 + X 3,11 = 6784

23.9.1.3 Solve the Mathematical Model

After solving the mathematical model using goal programming in the program (Win-
QSB) was the results are as follows (Table 23.7):

23.10 Explanation the Results

1. After solving the mathematical model using goal programming, the results
showed that all variables are values numeric ie for example the variable (X_29)
= 3713.18 meaning that the quantity that must be transferred from the second
328 H. H. Ahmed

warehouse (refinery Karkh) to the ninth station (minimum) is (371,318) liters of

gasoline, which The cost of the company (19,055) and so on for the rest of the
variables.
2. The results showed that the value of the variable d1− = 12,975,616 million
dinars, which means that the first target (cost target) was not achieved with a
deviation of (12,975,616) million dinars from the first target, the investigator is
(100,246,784) million and not (113,222,400) million as the target.
3. The other deviating variables d2− , d2+ appeared with a zero value and this indi-
cates that the second goal (the goal of reducing the transfer time) has been fully
achieved.
4. The value of the first target function (cost reduction function) (Z1 = 100,246,784)
million dinars either value. The second target function (transfer time reduction
function) (Z2 = 985,774.42).
5. When solving the mathematical model using linear programming, the results
showed that the value of the first objective function cost reduction function) was
(85,202,476) million dinars and the second target function (time reduction func-
tion) was (908,928), the value of (λ) was (0.08) that is, the degree of satisfaction
of the decision-maker to achieve the objectives was 80% is a good percentage.
6. When solving the mathematical model using two different methods, linear
programming and goal programming the results showed that linear programming
is better than goal programming in reducing the cost and time of transporting the
gasoline product.
From the three warehouses to the filling stations, which was (11) station because
linear programming always give an optimal solution, but goal programming gives an
acceptable solution, not an optimal solution in addition to being based achieve two
opposing goals together.

23.11 Discussion and Conclusion

Discussion
1. The company relies in processing on the first wareshouse (session) by 45% and
on the warehouse The second (Karkh) by 27% As for the third warehouse was
the largest dependence of the company by 64% Of processed quantity.
2. The fuzzy multi-objective transport model is better than the normal transport
model as the fuzzy model multi-goals have greater flexibility in decision-making
as well as time and cost factor.
3. Fuzzy linear programming is one method that can be used successfully to solve
uncertainty problems.
4. The use of the optimal transport model for gasoline product achieves a profit of
(12,975,616) million dinars for the month and reduce the total time (25,194,42)
hours.
23 Solving the Problem of Fuzzy Transportation Using Linear Programming … 329

5. Achieving all demand and supply constraints as well as budgetary constraints to

the problem of transporting gasoline from warehouses to Filling stations.
6. When solving the transportation model using goal programming helps the deci-
sion maker to find the best solution as well as knowing the extent to which the
objective functions deviate from its own objectives, ie the extent to which they
are achieved.

Conclusions
1. Use goal programming to solve the transportation model.
2. Use the concepts of operations research and use it to solve the problems of the
company (transport, warehouses, production …… etc.).
3. Apply this experiment to other products (white oil, oil, etc.) as well as all
companies involved in the distribution of petroleum products in the provinces.
4. Apply this experiment to external transport (import).
5. Raise the production capacity of the warehouse (3) (Rusafa warehouse) for the
purpose of benefiting from it being a major supplier of the company as well as
the increasing demand for the product necessitates an increase in production.
6. For researchers we recommend a multi-stage transmission system.
7. Researchers have the possibility to use other objectives of the transport problem,
such as reducing distance traveled, reducing accidents, etc.

References

1. Jain, K.K., Bhardwaj, R., Choudhary, S.: A Multi-Objective transportation problem solve by
lexio graphic goal programming. Int. J. Recent Technol. Eng. (IJRTE) 1842–1846 (2019).
2. Ali Ebrahim, N.: A simplified new approach for solving fuzzy transportation problems with
generalized trapezoidal fuzzy numbers. Appl. Soft Comput. 171–176 (2014)
3. El-Wahed, W.F., Sang, S.M.: Interactive fuzzy goal programming for multi-objective trans-
portation problems. Int. J. Manage. Sci. 159–165 (2006)
4. Anukokila, P., Radhakrishnan, B.: Goal programming approach to fully fuzzy fractional
transportation problem. J. Taibah Univ. Sci. 864–874 (2019)
5. Zangiabadi, M., Maleki, H.R.: Fuzzy goal programming technique to solve multiobjective
transportation problems with some non-linear membership functions. Iran. J. Fuzzy Syst. 61–74
(2013)
6. Liu, S.-T., Kao, C.: Solving fuzzy transportation problems based on extension principle. Eur.
J. Oper. Res. 661–674 (2004)
7. Sharif Uddin, M., Roy, S.K., Ahmed, M.M.: An approach to solve multi-objective transportation
problem using fuzzy goal programming and genetic algorithm. In: AIP Conference Proceedings
(2018)
8. Hussain, R.J., Kumar, P.S.: Algorithmic approach for solving intuitionistic fuzzy transportation
problem. Appl. Math. Sci. 3981–3989 (2012)
9. Debashis, D., Satyanarayana, M.A.: Multi-choice goal programming approach for a fuzzy
transportation problem, IJRRAS 132–139 (2010)
10. Kour, D., Mukherjee, S., Basu, K.: Solving intuitionistic fuzzy transportation problem using
linear programming. Int. J. Syst. Assur. Eng. Manag. 1090–1101 (2017)
330 H. H. Ahmed

11. Díaz-Madroñero, M., Mul, J., Jiménez, M.: A modified approach based on ranking fuzzy
numbers for fuzzy integer programming with equality constraints. In: International Conference
on Industrial Engineering and Industrial Management, pp. 375–382 (2012)
12. Rani, D., Gulati, T.R., Garg, H.: Multi-objective non-linear programming problem in intuition-
istic fuzzy environment: optimistic and pessimistic view point. Expert Syst. Appl. 228–238
(2016)
13. Pramani, S. Durga, B.: Multi-objective chance constrained capacitated transportation problem
based on fuzzy goal programming. Int. J. Comput. Appl. 42–46 (2012)
14. Pramanik, S., Roy Tapan, K.: Multiobjective transportation model with fuzzy parameters:
priority based fuzzy goal programming approach. J. Transp. Syst. Eng. Inf. Technol. 40–48
(2008)
15. Ghodratnama, A., Tavakkoli-Moghaddam, R., Azaron, A.: Robust and fuzzy goal programming
optimization approaches for a novel multi-objective hub location-allocation problem: A supply
chain overview. Appl. Soft Comput. 255–276 (2015)
16. Samuel, A.E., Venkatachalapathy, M.: Modified Vogel’s Approximation Method For Fuzzy
Transportation Problems. Appl. Math. Sci. 1367 –1372 (2011)
17. Radhakrishnan, B., Anukokila, P.: Fractional goal programming for fuzzy solid transportation
problem with interval cost. Fuzzy Inf. Eng. 359–377 (2014)
18. Kumar A., Kaur, A.: Methods for solving fully fuzzy transportation problems based on classical
transportation methods. Int. J. Oper. Res. Inf. Syst. 52–71 (2011)
19. Eylul, D., University and Pace University: A supply chain distribution network design model:
an interactive fuzzy goal programming-based solution approach. Int. J. Adv. Manuf. Technol.
401–418 (2008)
20. Chahal, S.S., Gupta, G.: A new approach for solving cost minimization balanced transportation
problem under uncertainty. J. Transp. Secur. 339–345 (2014)
21. Basirzadeh, H.: An approach for solving fuzzy transportation problem. Appl. Math. Sci. 1549–
1566 (2011)
Chapter 24
Effective Solution to Integrate
and Control a Heavy Robot Driven
by Hydraulic Actuators

Chu Anh-My

Abstract This chapter addresses an effective technical solution for integrating and
controlling a heavy robot of which all joints are driven by hydraulic actuators. The
robot is designed to support workers for handling automatically hot and heavy work-
pieces among machines and components of a forging press shop floor. The controller
of the robot is designed and integrated effectively since it is constructed mainly based
on industrial PLC units, which also plays a role as a central control unit for inter-
connecting all the components of the entire forging station. The control algorithm
for the system integration is validated through functional tests and experiments that
are carried out with a real forging shop floor at The Mechanical Company Ltd. No.
83, Yen Bai Province, Vietnam.

Keywords Hydraulic robot · Forging robot · Robot design · Robot control ·

Hydraulic actuator

24.1 Introduction

In general, a conventional hot forging press shop floor is usually comprised of a

furnace for heating raw materials and a forging press center using hydraulic power. To
begin a cycle of forging process, workers must grip a raw workpiece from a workpiece
collecting area, move and place the grasped workpiece on the table of the furnace.
When, in the furnace, a workpiece is heated to a desirable temperature (e. g 1100 °C)
the workers must grip again and move the heated workpiece as quickly as possible
to a mold fixed on the bed of the hydraulic press machine. Then the workpiece is
shaped by the die of the press machine. This conventional material handling method
for a forging press shop floor causes safety risks and dangers to the direct workers
because the workers must manipulate directly hot and heavy workpieces. To support
workers in these situations, a service robotic system is highly demanded, which was
designed and implemented for an existing manufacturing shop floor in Vietnam. The

C. Anh-My (B)
Le Quy Don Technical University (LQDTU), Hanoi, Vietnam
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 331

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_24
332 C. Anh-My

detailed design and simulation of the robot arm were presented in our previous works
[1, 2].
One important feature of the designed robot arm is that all five joints of the robot
arm as well as the end-effector (EEF) of the arm are driven by hydraulic actuators
(four hydraulic cylinders and two hydraulic motors). In addition, the fixture which
was designed in order to hold and locate a hot workpiece released by the robot arm
is driven by two hydraulic cylinders as well. Moreover, the robot arm is required to
cooperate aclosely with the heating furnace and the press machine as a robotic cell.
For these reasons, to integrate and control effectively the robot arm as well as the
relevant machines of the shop floor is a critical issue.
This paper presents an effective technical solution for the integration and control of
such robotic cell. The control system of the entire forging press shop floor is designed
based on the industrial PLC units for controlling all the components of the entire
manufacturing system. In addition, a robust and efficient algorithm is constructed
and implemented to automate the manufacturing cell effectively. Functional tests and
experiments carried out with a real forging press station at The Mechanical Company
Ltd. No. 83, Yen Bai Province, Vietnam demonstrates the effectiveness and efficiency
of the control system for integration of the hydraulic actuators proposed in this paper.
It has also shown that the proposed technical solution could be extended and applied
when integrating similar robotic systems with other existing press forging shop floors
in the mechanical manufacturing industry.
In the literature, research topics related to industrial robot design, and the robot
and machine integration in industries are extensively addressed [3–10]. The general
theoretical aspects for the robot design and analysis were summarized and presented
in several books and research papers such as [11–13]. Generally, there are two main
approaches for the design and integration of a robotic cell in manufacturing industries:
(1) the use of general-purpose robots, and (2) the use of task-oriented robots.
In the first approach, based on the force/torque control method, the research work
[14] presented an technique to integrate a light weight robot with other equipment for
a drilling process. Special examples of robotic cell integratyion for assembly lines
were presented in Refs. [15, 16]. The investigation [17] showed another method for
integration of an industrial robot of low stiffness in a manufacturing cell.
In the second approach, there has been a number of efforts [18–30] focusing
on design and implementation of new innovative robot systems which can be used
effectively in manufacturing indutries. An innovative design technique of robotic
system which is based on an expert functional design model was addressed in [18].
The proposed technique is driven and oriented with the designers’ behavior and the
function-environment-structure formalism. In this context, several individual types
of robot products have been designed and published that can be found in the literature,
such as the special serial robotic manipulator [19], the cabled robot [20], the hybrid
mobile robots [21, 22], the robot with legs and wheels [23], and the assistive robotic
system [24].
Also, in the second approach, there have been several researches [25–30] that
addressed the design and implementation of parallel manipulators and hybrid parallel
- serial manipulators. The researches in [25, 26] introduced the innovative designs
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 333

of parallel robots which can be used for the heavy open die forging applications.
The works [27–30] presented some types of forging manipulators, of which the
mechanisms of the robots were synthesized with a serial architecture combined with
parallel modules. Note that the designed manipulators are capable of holding and
manipulating a hot workpiece during in the forging process. Therefore, these forging
robots are different from the robot presented in this paper. The manipulator presented
in this chapter was designed to grasp, move and release workpieces among a given
workpiece collecting area, a furnace and a forging machine that are located in a large
work space of a forging press shop floor, whereas the forging robots proposed in [27–
30] were designed to manipulate directly a hot workpiece in a forging process.Related
to the issue of dynamic modelling and analysis of a forging robot, there have also
been recent publications [27, 28].
It can be observed that, the use of the second approach to design a robot has one
particular advantage. In order to design an individual robot, designers have chances
to take full advantages of the prior knowledge of application to optimize the robot
architecture and the control system that guarantee an optimal use of the robot product
for manufacturing tasks. In this manner, more and more new and effective robotic
system have been designed and applied in manufacturing industries.
Nevertheless, it can be seen that, most of the robot products used in manufac-
turing industries are driven by electric motors. The robots driven by hydraulic actu-
ators are rare. In comparison with the robots actuated by electric motors, the robots
with hydraulic actuators have some advantages. The first one is that, when using
hydraulic cylinders to drive the active joints of a robot, the rigidity of the robot struc-
ture is significantly increased. This type of hydraulic robots is particularly suitable
with applications of heavy workload in the mechanical manufacturing industry. The
second advantage of this robot type is that no counterweights are needed to balance
the heavy payload, since the hydraulic cylinders used for driving revolute joints of
a robot play a role as additional links reacting the gravity forces of the payload and
main links of a robot. Therefore, the total mass and the overall geometric dimensions
of a hydraulic robot could be optimized. However, the hydraulic robots have also
some disadvantages such as the difficulty in designing the control system and algo-
rithm, the inaccuracy of the positioning control, the complexity in the entire system
integration, etc.
As for the robotic system under consideration in this paper, the hydraulic robot arm
was designed with a special structure of high rigidity and stability [1, 2]. In particular,
since the designed robot arm has to cooperate automatically with the heating furnace
and the forging machine when processing a workpiece, it is necessary to design
additionally a workpiece loading table and a controllable fixture mounted on the
press machine. These two additional devices play an important role as “interface”
when integrating the robot arm with other machines of the forging shop floor. In short,
the designed robotic system is composed of a robot arm, a workpiece loading table
and a controllable fixture mounted on the press machine. To integrate and control this
complex manufacturing cell, a unified control framework is desired to communicate
and control all components of the whole robotic cell.
334 C. Anh-My

The aforementioned issues motivate a cost-effective control solution to integrate

the designed robotic cell in communication with other machines and devices of a
press forging shop floor. In this paper, a unified control framework and a PLC-based
control system design are addressed. In order to integrate and control all components
of the robotic cell with respect to all required processing steps and operations of a
forging cycle, a robust control algorithm is constructed and implemented effectively.
A series of functional tests and experiments validates the integration and control of
the whole automated manufacturing cell.

24.2 Description of the Robotic Cell

Figure 24.1 presents a general diagram of the robotic cell which was designed and
presented in our previous works [1, 2]. In this manufacturing cell, the robot arm was
designed with a special serial—parallel architecture, and all active joints are driven
by hydraulic actuators. The arm includes six joints (five active joints) and seven links.
As shown in Figs. 24.1 and 24.2, the workpiece loading mechanism was designed
for the task of lifting up and down a raw workpiece. This mechanism is necessary to
lifts a raw workpiece up to a programmed position so that the gripper of the robot
arm is able to approach to and grasp the workpiece more easily and precisely. At the
beginning of any forging cycle, the workpiece loading mechanism must load a raw
workpiece that is ready for the robot arm picks it up.
Since the robot cannot release directly a grasped hot workpiece inside the forging
mold, the fixture is needed to be designed as shown in Fig. 24.2. The fixture has two
main functions. The first one is that the fixture holds the heated workpiece which
is released by the gripper of the robot arm. Then the fixture lifts the workpiece at a
position so that it is ready to be pressed by the die of the forging machine. After the
die of the machine complete its stroke to shape the workpiece inside the mold, and

Forging Press Machine Heang Furnace

Fixture

Wp Ldg Mech

Fig. 24.1 A systematic diagram of the designed robotic system

24 Effective Solution to Integrate and Control a Heavy Robot Driven … 335

Fig. 24.2 a The mechanism

Up
for loading a workpiece, and
b The hydraulic fixture
mounted on the bed of the Down
machine
Up

Down

(a) (b)

Fig. 24.3 The robot was

integrated with the forging
machine

the shaped part gets out the mold, and the fixture holds again the shaped part. That
is the second function of the fixture.
Figure 24.3 shows the robot arm prototype integrated with the forging system and
other additional devices. For a working cycle, the first task of the arm is to grasp
a raw workpiece which is ready on the workpiece loading mechanism. After the
grasping task completed, the arm moves and places the grasped workpiece on the
table of the furnace for heating. After the furnace heats the workpiece as required,
the robot approaches and picks the heated workpiece up, and transfers it quickly to
the press machine. The fixture mounted on the press machine holds and locates the
hot workpiece, in cooperation with the robot when releasing the workpiece at the
same time. Finally, the press machine stars pressing the part into the stationary mold
fixed on the bed of the hydraulic forging machine.
As discussed earlier, one important feature of the design of this robotic system is
that all active joints of the robot arm are driven by two hydraulic motors and four
hydraulic cylinders. Other two hydraulic cylinders were used to drive the fixture as
well. As shown in Fig. 24.4 , the two hydraulic motors drive joint 2 and joint 6
(Fig. 24.5 and 24.6). Joint 1, 3 and 4 are driven by three hydraulic cylinders. Joint 5
336 C. Anh-My

Fig. 24.4 All active joints of

the robot are driven by
hydraulic actuators

Fig. 24.5 The hydraulic

motor driving joint 6

Fig. 24.6 The hydraulic

gripper of the robot arm
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 337

is a kind of passive joint which does not need a drive. One more hydraulic cylinder
is used for the gripper (Fig. 24.6).
Note that, in order to provide the strength and a significant amount of grasping
force that is suitable for grasping the required heavy workpieces, the gripper of the
robot arm was designed as a kind of hydraulic gripper as shown in Fig. 24.6.

24.3 The Control System Design

In this section, the design of a control system and the contruction of a PLC-based
control algorithm for the entire robotic cell are presented.
Based on a comprehensive anlysis of the data collected with a real conventional
forging press shop floor, the main functional requirements for the design of the control
system for the robotic cell can be described as follows:
1. The robot manipulator must be able to grasp and move workpieces precisely
among the raw-workpieces collecting area, the furnace and the press machine.
2. The robot manipulator must be able to interface effectively with the furnace and
the press machine, while performing the tasks (1). In other word, the robot arm,
the heating furnace and the press forging machine must be integrated all together
as a unified robotic cell.
For every press forging cycle, the robotic cell must perform the following tasks
in Table 24.1 to meet the two described requirements.
With respect to the design specifictions of the robotic cell, the general hydraulic
diagram for the cell can be designed an shown in Fig. 24.7.
In order to design an optimal control system for the robotic cell that meets all the
functional requirements and performs effectively all the decomposed tasks, PLC—
based control solution is selected based on the following reasons.
– As compared with conventional control methods, PLC—based control method
is more flexible. This is because the conventional control methods usually use
timers and relays or other sub-system to respond to a modification of the control
system. Whereas the PLC only changes programs written on memory.
– The PLC—based control method has high standards. Using PLC—based solution
standardization, the control programming environment for a robotic cell can be
understood and used more easily. However, each unit of a conventional control
hardware has its own specifications.
– As compared with other methods, PLC—based control method is particularly
advantageous when configuring and programming because PLC units can check
a program effectively. In this manner, it fixes quickly any errors which may rise.
– The power consumption of this PLC—based control solution is low, and the
control panel is very compact and standardized.
– PLC control systems operate perfect under extreme conditions such as high
temperature and humidity.
338 C. Anh-My

Table 24.1 List of task description

Task description Component of the cell
Task 1. The workpiece loading mechanism loads a raw The workpiece loading mechanism
workpiece
Task 2. The robot arm grasps the raw workpiece loaded by The robot arm
the workpiece loading mechanism
Task 3. The robot arm moves the grasped workpiece from The robot arm
the workpiece loading mechanism to the furnace
Task 4. The robot arm releases and places exactly the The robot arm
workpiece on the table of the furnace
Task 5. The robot arm waits for the heating furnace heats The robot arm
the workpiece
Task 6. The heating furnace heats the workpiece The heating furnace
Task 7. The robot arm grasps the workpiece on the table of The robot arm
the furnace
Task 8. The robot arm moves quickly the workpiece to the The robot arm
press machine
Task 9. The robot arm releases the grasped workpiece on The robot arm
the fixture
Task 10. The fixture holds and lifts the workpiece up to the The fixture
position ready for the forging press operation
Task 11. The forging press machine operates to shape the The forging press machines
hot workpiece on the forging mold
Task 12. The fixture holds and kicks the shaped part out The fixture

– Decades ago, in order to integrate an industrial robot control system, there have
been some techniques. However, the control method using programmable logic
control PLC units has been widely used in industrial manufacturing systems.
Generally, PLC could be regarded as the best control solution to automate
manufacturing processes in industries.
For the reasons above mentioned, in this study, the control system is designed
with the use of the commercial Siemens SIMATIC PLC unit S7-1200. Figures 24.8
and 24.9 show the control box and control panel. The input/output signals of the
main PLC unit are shown in Figs. 24.10. Figure 24.11 presents the flowchart of the
algorithm for the PLC programming.

24.4 Validation and Testing

In practice, according to stringent specifications, the entire robotic system and all of
its main functions must be strictly tested. In general, all the individual components
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 339

Fig. 24.7 The hydraulic system

of the robotic cell are needed to be tested as well. In this manner, the validation of
the robotic system is a long and involved process.
As for the product prototype of the proposed robot arm, an extensive validation
of all the components and functions that is necessary to develop next version of the
robot products in the feature is desired. In practice, a number of testing steps has
340 C. Anh-My

Fig. 24.8 The robot control

system

Fig. 24.9 The panel of the robot control system

been carried out on individual subsystems and components such as the cylinders,
motors, values, control units, control programs, inputs, outputs, etc.
Based on main results of the testing series and the numerical simulation, a compar-
ison was made and it has shown that the experimental results and the simulation
results are well matched. The experiments also show that the time when the robot
moves a hot workpiece from the furnace to the mold mounted on the press forging
machine is 24.4 s. This value of time is not perfect, since the robot is required to
move a heated workpiece to the press machine as quickly as possible to keep the
temperature of the workpiece and to reduce the power consumption of the heating
process. The comparison also showed that the actual value of time (24.4 s) is still
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 341

Fig. 24.10 Diagram of input/output signals of the main PLC unit S7-1200

lower than the allowable value (25.0 s). The allowable time means that, if a heated
workpiece (1100 °C) is moved from the furnace to the press machine is greater than
25.0 s, the workpiece must be re-heated.
Another comparison for the measuring data and allowable value has been made as
well. When the gripper is controlled to place a workpiece on the table of the heating
furnace, the positioning error calculated for nine processing cycles is 4.46 mm which
is still lower than the allowable value (5.0 mm).
In a real working condition of the robotic cell, a series of tests and experiement
has been carried out. The results of the tests are described in several tables in a
specification document. Some main results are shown in Tables 24.2 and 24.3. These
tables describe the tests of the specific operation of the robotic cell. In Table 24.1,
the second column reports the critical state (must or should). In this table, all tasks
of the robotic cell have a must state since they are all the main functions of the cell.
The test results for the gripper is shown in Table 24.3.
Several functional tests were carried out and the testing results show that the
robot works well and the use of this robotic solution reduces the forging cycle time
342 C. Anh-My

Auto Mode
I0.1=1
Heating
Q1.4=1 Move joints 3&4
Move to pick a raw billet Q0.4=1, Q0.6=1
I0.4=1
Move back joints 3&4
Q0.5=1, Q0.7=1 AI1?& AI2?
Move joints 3&4
Q0.4=1, Q0.6=1
AI1?& AI2?
Release a heated billet
AI1?& AI2? Q0.4=0, Q0.6=0, Q1.3=1

11000C?
Move back joints 3&4
Grasp a raw billet
Q0.5=1, Q0.7=1
Q0.4=0, Q0.6=0, Q1.2=1
Move joints 3&4
Q0.4=1, Q0.6=1
Move back joints 3&4 AI1?& AI2?
Q0.5=1, Q0.7=1
AI1?& AI2?
Turn left joint 2
AI1?& AI2? Q0.2=0, Q0.3=1
Grasp a heated billet
Q0.4=0, Q0.6=0, Q1.2=1
I3.6
Turn left joint 2
Q0.5=0, Q0.7=0, Q0.3=1
Move back joints 3&4
Q0.5=1, Q0.7=1
Next cycle
I3.6

AI1?& AI2?
Move joints 3&4
Q0.4=1, Q0.6=1
Turn right joint 2
Q0.2=1, Q0.3=0
AI1?& AI2?

I3.6
Release a raw billet
Q0.4=0, Q0.6=0, Q1.3=1

Fig. 24.11 A flowchart of the control algorithm

by 27.7%. The robot grasps, transfers and releases workpieces realiably when serving
the furnace and the press machine.

24.5 Conclusion

In this book chapter, a cost-effective technical solution to integrate and control a

special robotic cell in the hot forging industry was presented. The testing results
show that the robotic cell operates effectively and realiably, where the robot manip-
ulator plays a central role for the shop floor automation. The arm can handle a
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 343

Table 24.2 List of all the functional tests of the robot arm
Task Critical state
The workpiece loading mechanism must load a raw workpiece Must
The robot arm must grasp the raw workpiece loaded by the workpiece loading Must
mechanism
The robot arm must move a grasped workpiece from the workpiece loading Must
table to the furnace
The robot arm must place exaxtly a workpiece on the table of the furnace Must
The robot arm must wait for the heating process of a workpiece in the furnace Must
The heating furnace must heat the workpiece Must
The robot arm must move in and grasp a heated workpiece on the table of the Must
furnace
The robot arm must move a heated workpiece to the press machine Must
The robot arm must release the grasped workpiece on the fixture Must
The fixture must hold and lift the workpiece up to the position ready for the Must
forging press operation
The forging press machine must operate to shape the hot workpiece on the Must
forging mold
The fixture must hold and kick the shaped part out Must

Table 24.3 List of tests for the gripper

Task Critical state
The gripper must not deform and drop the workpiece s Must
The gripper must place the workpiece on the table inside the furnace with Must
positioning error lower than 5 mm
Turn and turn back 900 around twist axis Must
Maintain the link posture parallel with the ground surface Must

workpiece precisely when serving a heating furnace and a hydraulic press machine.
The functional tests of the control system of the entire robotic cell were passed with
persistent results.
The experimental results show that the entire robotic system can work smoothly
and effectively that meet all the functional requirements of a hot forging press shop
floor.

Acknowledgements This book chapter was supported by British Council and Newton Fund Grant
Agreement Research Environment Links with Project Code: 528085858.
344 C. Anh-My

References

1. My, C.A., et al.: Novel robot arm design and implementation for hot forging press automation.
Int. J. Prod. Res. 57(14), 4579–4593.
2. My, C.A., Parnichkun, M.: Kinematics performance and structural analysis for the design of a
serial-parallel manipulator transferring a billet for a hot extrusion forging process. Int. J. Adv.
Robot. Syst. 12(12), 186 (2015)
3. My, C.A., et al.: Inverse kinematic control algorithm for a welding robot-positioner system to
trace a 3D complex curve. In: 2019 International Conference on Advanced Technologies for
Communications (ATC), pp. 319–323. IEEE (2019)
4. My, C.A., et al.: Modeling and computation of real-time applied torques and non-holonomic
constraint forces/moment, and optimal design of wheels for an autonomous security robot
tracking a moving target. Math. Comput. Simul. 170, 300–315 (2020)
5. My, C.A.: Integration of CAM systems into multi-axes computerized numerical control
machines. In: 2010 Second International Conference on Knowledge and Systems Engineering,
pp. 119–124. IEEE (2010)
6. My, C.A., et al.: Mechanical design and dynamics modelling of RoPC robot. In: Proceedings
of International Symposium on Robotics and Mechatronics, Hanoi, Vietnam, pp. 92–96 (2009)
7. My, C.A.: Inverse kinematics of a serial-parallel robot used in hot forging process. Vietnam J.
Mech. 38(2), 81–88 (2016)
8. My, C.A., et al.: Inverse dynamic of a N-links manipulator mounted on a wheeled mobile
robot. In: 2013 International Conference on Control, Automation and Information Sciences
(ICCAIS), pp. 164–170. IEEE (2013)
9. My, C.A., Trung, V.T.: Design analysis for a special serial-parallel manipulator transferring
billet for hot extrusion forging process. Vietnam J. Sci. Technol. 54(4), 545 (2016)
10. My, C.A., Hoan, V.M.: Kinematic and dynamic analysis of a serial manipulator with local
closed loop mechanisms. Vietnam J. Mech. 41(2), 141–155 (2019)
11. Garcia, E., et al.: The evolution of robotics research. IEEE Robot. Autom. Mag. 14(1), 90–103
(2007)
12. Khang, N.V., My,C.A.: Fundamentals of Industrial Robot. Education Publisher, Ha Noi, Viet
Nam (2010)
13. Su, J., et al.: Grasping objects: the relationship between the cage and the form-closure grasp.
IEEE Robot. Autom. Mag. 24(3), 84–96 (2017).
14. Olsson, T., et al.: Cost-efficient drilling using industrial robots with high-bandwidth force
feedback. Robot. Comput.-Integr. Manuf. 26(1), 24–38 (2010)
15. Michalos, G., et al.: ROBO-PARTNER: seamless human-robot cooperation for intelligent,
flexible and safe operations in the assembly factories of the future. Procedia CIRP 23, 71–76
(2014)
16. Cherubini, A., et al.: Collaborative manufacturing with physical human–robot interaction.
Robot. Comput.-Integr. Manuf. 40, 1–13 (2016)
17. Denkena, B., Lepper, T.: Enabling an industrial robot for metal cutting operations. Procedia
CIRP 35, 79–84 (2015)
18. Tor, S.B., et al.: Knowledge-based functional design of industrial robots. Int. J. Prod. Res.
46(16), 4501–4519 (2008)
19. Chang, P.H., Park, H.-S.: Development of a robotic arm for handicapped people: a task-oriented
design approach. Autnom. Robots 15(1), 81–92 (2003)
20. Borgstrom, P.H., et al.: Design and implementation of NIMS3D, a 3-D cabled robot for actuated
sensing applications. IEEE Trans. Robot. 25(2), 325–339 (2009)
21. Fiorini, P., Botturi, D.: Introducing service robotics to the pharmaceutical industry. Intell. Serv.
Robot. 1(4), 267–280 (2008)
22. Rea, P., Ottaviano, E.: Design and development of an Inspection Robotic System for indoor
applications. Robot. Comput.-Integr. Manuf. 49, 143–151 (2018)
23. wei, Z., et al.: Design and implementation of a leg–wheel robot: transleg. J. Mech. Robot. 9(5)
(2017)
24 Effective Solution to Integrate and Control a Heavy Robot Driven … 345

24. Kim, D.-J., et al.: System design and implementation of UCF-MANUS—An intelligent
assistive robotic manipulator. IEEE/ASME Trans. Mechatron. 19(1), 225–237 (2012)
25. Pa, P.S., Wu, C.M.: Design of a hexapod robot with a servo control and a man-machine interface.
Robot. Comput.-Integr. Manuf. 28(3), 351–358 (2012)
26. Jin, Y., Chen, I.-M., Yang, G.: Kinematic design of a family of 6-DOF partially decoupled
parallel manipulators. Mech. Mach. Theory 44(5), 912–922 (2009)
27. Yan, C., Gao, F., Zhang, Y.: Kinematic modeling of a serial–parallel forging manipulator with
application to heavy-duty manipulations#. Mech. Des. Struct. Mach. 38(1), 105–129 (2010)
28. Chen, G., et al.: Modular calculation of the Jacobian matrix and Its application to the
performance analyses of a forging robot. Adv. Robot. 23(10), 1261–1279 (2009)
29. Faguo, Y.U., et al.: Structure synthesis for forging manipulators. In: 2008 7th World Congress
on Intelligent Control and Automation, pp. 400–403. IEEE (2008)
30. Ding, H., et al.: Structure synthesis of 6-DOF forging manipulators. Mech. Mach. Theory 111,
135–151 (2017)
Chapter 25
Design and Implementation of Arabic
Plagiarism Detection System

Zahraa Jasim Jaber and Ahmed H. Aliwy

Abstract Plagiarism is a critical problem that appears in most universities and

researchers. It is a challenging task, especially in case of highly inflected and rich
languages such as Arabic. In this paper a proposed reliable approach for plagiarism
detection in Arabic language, designed and implemented using an efficient and quick
way. The main idea is to use simple Information Retrieval system (IR) to compute the
percentage of grams of plagiarized text which is difficult without using exact match
algorithm. The proposed method consists of four main stages: (i) Pre-processing,
(ii) Lexical semantic manipulation, (iii) Indexing with n-gram and (iv) Calculating
the percentage of plagiarism. The system calculates the percentage of plagiarism
directly from the intersection of its grams without using any exact match algorithm.
The extracted percentage is done in two ways: (i) percentage of plagiarism in suspi-
cious document (a file in data set) and (ii) percentage of plagiarism in a file in data set
(to suspicious file). Then Precision, Recall, F-measure and Error rate are evaluated.
The proposed method is compared with a selected method (winnowing algorithm),
using the same data set, an improvement in percentages of error rate were recorded.
In case of execution time, the proposed method is four times faster than winnowing
algorithm. Also, semantics manipulation is added to improve the proposed plagia-
rism detection system. The results of the proposed method are 0.853624, 1, 0.889991
and 0.022048 for precision, recall, f-measure and error rate respectively, while the
results of winnowing algorithm 0.666518828, 1, 0.720509265 and 0.043987806 for
precision, recall, f-measure and error rate respectively.

Keywords Arabic text analysis · Arabic text Indexing · Plagiarism detection

system

Z. J. Jaber · A. H. Aliwy (B)

Department of Computer Science, Faculty of Computer Science and Mathematics, University of
Kufa, Kufa, Iraq
e-mail: [email protected]
Z. J. Jaber
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 347

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_25
348 Z. J. Jaber and A. H. Aliwy

25.1 Introduction

Plagiarism is illegal use and a literary or intellectual theft of other person’s works
without reference to the source of information. Many people use plagiarism to copy
texts from other people or borrow their original ideas [1]. Plagiarism is a main critical
issue for universities and researchers. In case of Plagiarism, two questions appear
how to avoid plagiarism and how to detect plagiarism [2]. There are two ways to
eliminate plagiarism: the first methods are Plagiarism detection methods and the
second are plagiarism prevention methods. Plagiarism detection methods include
specialized plagiarism detection software as well as manual methods. This method
is easy and quick to implement. While the methods to prevent plagiarism, although
these methods have a positive long-term impact, but take a long time to implement
because they rely on social cooperation between universities. These methods include
measures taken to impose the sanctions on the Impostors. The two methods can
be combined to eliminate the plagiarism. The final decision of plagiarism must be
manually, in spite of the software is the most efficient way to identify plagiarism
[2]. Plagiarism is not limited to the direct copy the texts but there are some persons
rewrite the sentence by changing the words’ locations or by changing part of them
such as synonyms without referring to the original source i.e. Plagiarism can be exact
copy plagiarism, Modified plagiarism, Idea plagiarism and others. One of the most
important characteristics of plagiarism detection is the protection of the authors rights
and helps the scientific organizations to ensure that the provided scientific research
was not plagiarized from previous works and hence promotion of ethical principles
in scientific production.

25.2 Literature Survey

There are many works and researches in plagiarism detection for English language but
for Arabic is limited. Some of these works will be presented in this section: Alzahrani
and Salim [3] presented a method of detecting plagiarism in Arabic texts by using a
fuzzy set IR- model. They computed similarity between two phrases and compared it
with the value of the threshold to determine whether the two statements are similar or
different. They found that the fuzzy model did not handle cases rewording by different
synonyms\antonyms. Jadalla and Elnagar [4] used fingerprinting method, the n-gram
method for chunking the text into words and winnowing algorithm to reduce indexing
size. The data set was four different collections such as set of students’ Projects,
116,011 Arabic Wikipedia Files, the holy Qur’an and the last manually compiled
files. The fingerprints of each n-gram that are represented by hash function. The result
showed that the detection process is more accurate (Recall is 94% and Precision is
99%) and the search time was improved. Riki et al. [5] used the winnowing algorithm
with synonym recognition to detect plagiarism between Indonesian documents. The
percentage of document similarity is calculated using jaccard’s similarity coefficient.
25 Design and Implementation of Arabic Plagiarism Detection System 349

The results showed that using gram = 2, will increase the number of the used windows
and hence the process time and the percentage of similarity varies with adding the text
preprocessing and synonym recognition. Idicula [6] used fingerprinting techniques in
the detection system to identify the plagiarism. The author identified the similarity of
documents through the detection of theft of Malayalam documents and concluded that
the method of ablution improves the detection of plagiarism by changing the structure
of the sentence and replacing it with synonyms. It has obtained a high accuracy for
detecting similarities compared to the chunking based on the sentence. El Moatez
et al. [7] proposed two approaches for plagiarism detection: The first approach uses
words alignment, word-embedding, and words weighting to measuring the semantic
similarity between texts in document. The second approach is based on Machine
Learning (ML) that executed at the sentence level. They combined between lexical,
syntactic, and semantic features in the plagiarism detection process. The used data
set was the publicly available Arabic External Plagiarism Corpus (EXARA-2015).
Khan et al. [9] proposed a plagiarism detection system for the Arabic language. They
used many of similarity-measure techniques for plagiarism detection in Arabic texts.
The proposed system contains two components, firstly document-retrieval and the
second is detailed similarity computation. The results showed that in the system of
the document-retrieval level achieve about 75% accuracy in terms of f-score, while
at the detailed similarity-computation level; the f-score is about 70%

25.3 Dataset

A data set is a collection of large text files that are stored and processed electronically.
Dr. Ahmed Aliwy collected the files from Al-Sabah newspaper, consists of 54,300
text files of Arabic language and classified into 26 categories. Zena et al. [8] was
filtered these files and divided them into five categories were chosen including. Arts
and Literature, Sport, Economy, Science and Technology and Family and Commu-
nity. These categories are less than the original classifications. Some of the original
categories were merged to one category such as science and technology and have
become one category as well as the family and society … etc.

25.4 Outline of the Proposed System

The proposed system consists of four main stages: (i) preprocessing, (ii) lexical
semantic manipulation (synonyms replacement), (iii) indexing with N-gram and (iii)
matching algorithm (the proposed method). Figure 25.1 shows plagiarism detection
system and Fig. 25.2 shows corpus initialization and building inverted index stages.
350 Z. J. Jaber and A. H. Aliwy

Preprocessing
The preprocessing is done on the corpus and the suspicious document. The
proposed system has four steps: normalization, tokenization, stop words removal
and light stemming.
Normalization
Normalization, for Arabic language, is a unification of Arabic letters, diacritics
removing and Tatweel removing. Some Arabic letters has multiple forms, for example
the letter "|" have the forms ( ) which in turn should be unified.
Tokenization
Tokenization is chunking or separating texts into the smallest units either letters
or words based on the limits of a word that based on white spaces and punctuation
marks as constraints between words (“.”, ‘ ‘, ‘,’, ‘;’ …) [2]. In the proposed system this
stage is considered as an important stage for the plagiarism detection of Arabic Texts
because many words types are merged together without spaces such as prepositions
and pronouns therefore, dealing with Arabic has become more difficult than English.
Stop Words Removal
Stop words are words that have no meaning if they found alone. When it was deleted,
it does not affect the meaning of the sentence. It is used, only to complete the sentence
structure. They are the most frequent words that appear in texts such as pronouns,
prepositions, tools, etc. [8].
Light Stemming
Stemming is process of removing any affixes such as (i) prefixes that added to the
beginning of the word, (ii) infixes that added to the middle of the word, and (iii)
suffixes that added to the ending of the word, to reduce these words to their stems or
roots on the assumption that some words sharing the same stem [2].
Synonyms Replacement
Most of the plagiarized texts are modified to prevent the plagiarism detection tool
from detect it. This is done with many methods but the most important are using
synonym and reordering the phrases. Therefore, similar words, in meaning, should
be unified which make the detection process more effective. This is achieved by
converting the similar words to their most common synonym, which in turn help to
detect advanced forms of hidden impersonation (Table 25.1).
N-gram
It is a continuous series of n elements from a given pattern of speech or text. These
elements can be phonemes, syllables, letters and words according to the used appli-
cation. When elements are words, n-grams may also be called shingles. N-gram is
used extensively in authoring analysis methods, which can be defined as a series of
contiguous words called word collecting. Most IR systems use one word as base
25 Design and Implementation of Arabic Plagiarism Detection System 351

for indexing but in some cases, more than one word are used as a reference which
called n-gram. According to our knowledge, there are not any Plagiarism detection
method used n-gram for indexing and matching for exact percent but it is used as
preprocessing in fingerprint method to give the candidate one approximately. Our
methodology suggests using n-gram for indexing and extracting the percent of plagia-
rized text without any farther processing. In this paper, n-gram is used in word level
(Light stems). The used corpus (data set) is partitioned to n-gram after completing
the preprocessing and synonym replacement steps for one time only and hence each
n-gram will be sequence of n stems. These n-grams will be stored in a suitable data
structure for efficient access. Also any suspicious text will have the same processing
(n-gram). In addition, the total number of grams was calculated for each document
to be used in the calculation of the ratio of the plagiarism.
Indexing
Indexing is very important for speeding the accessing of the stored data. Usually an
index is a list of terms that appear in a document, it can be represented mathematically
as:
index: doci → {∪j keyword}, where keyword is a word in document i(doci ).
The most widely used indexing method is Inverted Index which is used to index
all the words within the data set where each posts list will contain all the documents
containing that word and possibly the location of that word as well within that
document.
Inverted index: keyword → {∪i doci }, where document i(doci ) contain keyword
(word j).
Any number of words (n-gram) can be used for indexing and not necessarily a
single word. It is possible to use two words (bigram) or three words (trigram) or any
number of words. But increasing the number of n will greatly increase the amount of
memory required and will reduce the size of posting list. Figure 25.2 shows samples
of inverted index where kw1 is keyword or n-gram and doc1, doc2 , … is posting list
of the documents contain this keyword or n-gram. Also, the posting list can contain
repetition of the n-grams.
If n = 1 (one word), the searching and matching among the query and data set
is done easily by extracting posting list of each word in the suspicious document
and then finding the posting lists of these words. This is done for small query in IR
system because the intersections of these posting lists should be extracted. If n > 1 is
taken, extracting all parts of the similarity is the complex process and the complexity
is O (2x ).
Proposed Method for Matching Algorithm
If there is a document that contains a string of stem (s1 … sm ), it can be divided into n-
gram where g1 = x1 … xn and g2 = x2 … xn+1 , …, so for the purpose of extracting all
n-grams in the documents in data set, a posting list for each n-gram will be extracted.
The intersection of these posting lists will give the closest document to the suspicious
document. If all the documents, close to this suspicious document, are needed, the
352 Z. J. Jaber and A. H. Aliwy

intersections between the posting lists will be for every two n-gram or more are done
and this will generate complexity of O (2n ) where n represent number of n-grams.
But the easiest way is to merge posting lists in single list and then sorting it so that
similar documents will be consecutive, hence all the intersections will be obtained
in a simple way and linear time. And hence, the number of shared grams between
any documents (in the data set) and the suspicious document will be known. As was
mentioned previously for a document, if one word (n = 1) is used as a basis, it is
very easy to extract the similarity percentage, not the match, that happens in the
IR systems in search engines. But when n-gram is used and n > 1, then increasing
value of n will cause a problem in extracting n-gram intersections, therefore most
researchers use n-gram to find similarity only through use fingerprint instead of the
complete document to speeding the comparison process and then traditional string
matching algorithm is used to extract identical texts parts.
In this paper, the identical text between the suspicious document and the other
documents (database) can be found without using any string match algorithm, but
directly from the posting lists information as will be shown in next paragraphs.
Suppose that the available information is inverted index where each n-gram has
a posting list that represents all the documents containing this n-gram and sorted
according to the documents number. Also, there is a suspicious document, which
we want to know the plagiarism text in it from the other documents (the database).
N-gram will be extracted from this file with their repetition. Also, for each n-gram,
a list of the interleaved n-grams with it will be recorded. For example, if we have the
following grams:

g1 = wk . . . wk+n (25.1)

g2 = wm . . . wm+n

(25.2)


Then g1 and g2 are interleaved if wk + 1 … wk + n = wm . . . wm+n−1 and g2 is
follow to g1 .
For each gram in the suspicious document, a posting list is extracted from the
inverted index with their frequency. All the posting lists are merged into a single list
with sorting it according to the value of the document. Hence all the shared grams
with a specific document will be grouped and finally the number of these grams will
be known. But this number does not reflect the real plagiarism, from that document,
because these grams are interleaved. Suppose that n = 5 and there are 10 of 5-g
shared between the suspicious document and document d. One question appears: is
the plagiarism will be summation of n multiplied by 5-g repetitions?
The shared words = 5* repetition of (5gram1 ) + · · · + 5* repetition of (5gram10 )


n
= 5∗ Repetition of (g) (25.3)
g ∈ set o f grams
25 Design and Implementation of Arabic Plagiarism Detection System 353

This percentage is not real because of the overlap between the grams and therefore
not reliable. The suggested approach is that part of the overlap gram (associated),
for the other gram, should be taken, i.e., in the previous example, each overlap gram
will be computed based on only one word and not five words. If we assume that the
follow is F (follow to) and the repetition of any gram is R (g), the number of shared
words between the suspicious document and any document will be.


n
shar ed wor ds = n*(R(g) − F) + F (25.4)
g ∈ set o f grams

For example, if we have a document with 5 of 3-g (g1, g2, g3, g4, g5 ) as shown in
Table 25.2 with their repetition, then R(g) and (F) can be extracted as in Table 25.3
where n = 3.
As can be seen from Table 25.2, g1 is followed by (g2 , g5 ), g2 is followed by g3 ,
g3 is followed by g4, g4 is followed by (g2 , g5 ), g5 is not followed by any gram, this
leads to the result that shared words can be estimated using Eq. (25.4):

the shared words = [3*(4 − 0) + 0] + [3*(2 − 2) + 2] + [3*(2 − 2) + 2]

Table 25.1 Synonyms example

Synonyms Original word (Arabic)

Table 25.2 The shared grams and their repetition

No of grams 3-words Repetition
g1 w1 w2 w3 4
g2 w2 w3 w4 2
g3 w3 w4 w2 2
g4 w4 w2 w3 5
g5 w2 w3 w5 3

Table 25.3 Grams repetition and follow to (F)

Gram g1 g2 g3 g4 g5
R(g) 4 2 2 5 3
F 0 2 2 2 3
354 Z. J. Jaber and A. H. Aliwy

+ [3*(5 − 2) + 2] + [3*(3 − 3) + 3]
= 12 + 2 + 2 + 11 + 3 = 30

It is easy to see that, the normal methods give 48 from (16 * 3) as shared words
between the suspicious and the document but the exact is 30.

25.5 Implementation and Results

As mentioned in the previous sections, the paper focuses on finding an efficient and
quick way to detect plagiarism. In our work, inverted index, which is used in search
engine, was used. The limitation of this method was solved such as limited number of
queries and getting the matching percent from the intersections with less complexity
(linear instead of exponential). The main idea in this thesis is to use simple IR
system to compute the percentage of grams which is difficult without using exact
match algorithm. Python (version 3.4) has been used, as programing language, for
all the proposed algorithms.
Preprocessing stage, (normalization, Tokenization, stop words removal and light
stemming), are implemented on the corpus and suspicious files as were explained
previously. This stage was applied on the corpus and the suspicious files in separately.
After completing the preprocessing, the synonyms are replaced by the synonyms
dictionary, which was created using WordNet, which gives each word a list of
synonyms, and then returns each synonym to the original word. The word which
has no synonym stays without replacement. Also, synonyms are replaced for the
corpus and suspicious separately. Then n-gram technique is applied after synonyms
replacement (for corpus and suspicious files). The texts are divided into five words
(n = 5) and shifted by one at a time. Figure 25.3 shows 5-g of the selected text after
preprocessing.
All the suspicious files were compared with all corpus files after the inverted
indexing process. If we suppose S is set of the suspicious files and d is set of the
corpus files then Testing of plagiarism percent is evaluated using two methodologies:
1. Percent of plagiarism in Si from dj to dj, but the result will be huge because each
suspicious file will give a result with 15,184 files; therefore precision, Recall,
and F-measure are evaluated where the average is shown in Table 25.4.
2. Percent of plagiarism in Si from dj to Si: for the same region in point (i), Precision,
Recall, and F- measure are evaluated where the average is shown in Table 25.4.
also, Error rate are estimated according to percent of plagiarism for the two
methodologies as shown in Tables 25.4.
Winnowing algorithm was applied to 5-g of words instead of characters where
Hash-Function is used to generate fingerprints for the files (a unique numerical repre-
sentation) and then determine a window size to choose the least fingerprint of each
window. The results of using winnowing algorithm for the two methodologies and
with/without using synonyms are shown in Table 25.4.
25 Design and Implementation of Arabic Plagiarism Detection System 355

Fig. 25.1 Plagiarism detection

Fig. 25.2 Inverted index

356 Z. J. Jaber and A. H. Aliwy

Fig. 25.3 5-g of the selected text file after preprocessing and synonyms

Table 25.4 Average of precision, recall, f-measure, and error rate of proposed method and
winnowing algorithm (average for all suspicious files)
The used method Precision Recall f-measure Error rate
Suggested method First methodology 0.853624 1 0.889991 0.026023
second methodology 0.853624 1 0.889991 0.022048
Winnowing with First methodology 0.72312 1 0.77566 0.06056
synonyms Second methodology 0.73672 1 0.78935 0.05345
Winnowing without First methodology 0.71342 1 0.76596 0.06113
synonyms Second methodology 0.72723 1 0.78027 0.05485

The results showed that the proposed method is better than the winnowing algo-
rithm. It is much faster than the winnowing algorithm, which takes four times the
time of the proposed method. Error average for the proposed method is less than
error average to the winnowing algorithm.
25 Design and Implementation of Arabic Plagiarism Detection System 357

25.6 Conclusion

In this paper, the suggested method was introduced for extracting percent of plagia-
rism for Arabic texts. This method extracts the percent directly without need into
other exact match methods. This method was characterized by its simplicity and fast
execution. Its execution was done by different techniques on Arabic data and it’s effi-
ciency was proved by comparing it with winnowing method which is considers the
base for content-based methods. It was evaluated using Precision, Recall, f-measure
and Error rate where the following points were concluded:
1. The proposed method was compared with the winnowing algorithm where the
results of the proposed method were much better in terms of speed and accuracy.
2. The proposed method is faster four times than winnowing in the time of execution
the proposed method.
3. The Precision, Recall, f-measure and Error ratio metrics, for the proposed method,
is much better than the results of winnowing method.
4. The time complexity, in previous methods, requires 2n to calculate intersections
while the time complexity, for the proposed method, became linear.
5. Increasing the size of n-gram will increase the memory size and reduce the
number of post list (fingerprint) but it increases the error rate and the complexity.
6. A large suspicious documents can be compared with too many stored documents
(Corpus), in linear time.
There are many works can be done as future works. Some of them can be
summarized by:
1. Trying to detect idea plagiarism for Arabic language, because it is challenge task.
2. Dealing with semantics deeply for produce intelligent plagiarism detection.
3. Introducing OCR system to assistant in plagiarism detection because some works
insert image as a text.
4. Designing a multi lingual system including Arabic language for P.

References

1. Ibrahim, R., Saeed, S., Wakil, K.: Plagiarism Detection Techniques for Arabic Script (2017)
2. El Bachir Menai, M.: Detection of plagiarism in Arabic documents. Int. J. Inf. Technol. Comput.
Sci. (IJITCS) 4(10), 80–89 (2012). https://doi.org/10.5815/ijitcs.2012.10.10.
3. Alzahrani, S.M., Salim, N.: Plagiarism detection in Arabic scripts using fuzzy information
retrieval. In: Student Conference on Research Development, Johor Bahru, Malaysia (pp. 281–
285) (November, 2008)
4. Jadalla, A., Elnagar, A.: A fingerprinting-based plagiarism detection system for Arabic text-
based documents. In: 2012 8th International Conference on Computing Technology and
Information Management (NCM and ICNIT) (vol. 1, pp. 477–482). IEEE (April, 2012)
5. Riki, R., Edy, E., Maryanto, M.: Plagiarism Detection Application Uses Winnowing Algo-
rithm with Synonym Recognition for Indonesian Text Documents. Selangor Sci. Technol. Rev.
(SeSTeR) 3(1), 35–48 (2019)
358 Z. J. Jaber and A. H. Aliwy

6. Idicula, S.M.: Fingerprinting based detection system for identifying plagiarism in Malayalam
text documents. In: 2015 International Conference on Computing and Network Communications
(CoCoNet) (pp. 553–558). IEEE (December, 2015)
7. El Moatez Billah Nagoudi, A.K., Hadda Cherroun, D.: 2L-APD: a two-level plagiarism detection
system for Arabic documents. Cybern. Inf. Technol. 18(1) (2018)
8. Abutiheen, Z.A., Aliwy, A.H., Aljanabi, K.B.S.: Arabic text classification using master-slaves
technique. J. Phy. Conf. Ser. 1032, 012052. https://doi.org/10.1088/1742-6596/1032/1/012052
9. Khan, I.H., Siddiqui, M.A., Jambi, K.M.: Towards Building an Arabic Plagiarism Detection
System: Plagiarism Detection in Arabic. International Journal of Information Retrieval Research
9(3), 12–22 (2019). https://doi.org/10.4018/ijirr.2019070102
Chapter 26
Application of Artificial Intelligence
to Asset Pricing by Vietnamese Text
Declaration

Tran Ngoc Thang, Dao Minh Hoang, Tran Thi Hue, Vijender Kumar Solanki,
and Nguyen Thi Ngoc Anh

Abstract Asset pricing problem is an interesting problem by both academic and

practical aspects. It has many practical applications such as pricing products on e-
commerce transactions, pricing mortgage assets in credit activities,... In the era of
data explosion, the application of big data analysis as well as artificial intelligence
is an inevitable trend to produce more accurate predictive results. The article has
applied artificial intelligence algorithms in asset pricing through text descriptions
of the assets in Vietnamese. The proposed method uses Named Entity Recognition
technique with a Recurrent Neural Network model in combination with Conditional
Random Field model to extract asset features, thereby building a regression model
to evaluate the price of assets based on the attribute set. The method works relatively
well with a dataset of mobile phone descriptions with high accuracy.

26.1 Introduction

As we know, the assets such as phones, watches, cars, motorcycles, real estate...
are very diverse, regularly updated and have many attributes that impact on the asset
price. Therefore, the use of an existing dataset to look up prices is not comprehensive
enough. The solution here is to use artificial intelligence to estimate the price of an

T. N. Thang (B) · D. M. Hoang · N. T. N. Anh

School of Applied Mathematics and Informatics, Hanoi University
of Science and Technology, Hanoi, Vietnam
e-mail: [email protected]
N. T. N. Anh
e-mail: [email protected]
T. T. Hue
Faculty of Management Information System, The Banking Academy of Vietnam, Hanoi, Vietnam
e-mail: [email protected]
V. K. Solanki
Department of Computer Science and Engineering, CMR Institute of Technology,
Hyderabad, TS, India

© Springer Nature Switzerland AG 2021 359

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_26
360 T. N. Thang et al.

asset based on the same information in an existing dataset. Moreover, for convenience
and flexibility in declaring assets, users often describe by a few lines of text. Thanks
to the use of artificial intelligence combined with natural language processing in
Vietnamese, the system will automatically extract the attributes of assets to serve the
price lookup. To build this model, techniques such as regression neural networks,
decision tree, and conditional random field models will be used. This article proposes
an automatic asset pricing model with input in a Vietnamese text describing asset
information and the output as asset value.
There are two key challenges of asset pricing and the application of artificial
intelligence to address those challenges. For financial aid organizations, common
assets such as phones, watches, jewelry, cars, motorcycles, real estate, etc., which
can be collected in Vietnamese websites (Suong et al. [1]), can be accepted as asset.
These assets are not only rich in type but also have very different values. Moreover,
each attribute is described by a set of attributes. Although the same manufacturer, the
same specifications, even with only a small difference in the details such as color, the
two assets have different prices. Therefore, the database of available assets is most
likely to be incomplete with information about asset types as well as attribute values.
Therefore, an accurate pricing is often not feasible. This has led to the need to use
artificial intelligence technology in general and machine learning techniques, in par-
ticular, to estimate or predict asset value based on similar information. In the current
data explosion, the application of artificial intelligence to prediction and prediction
is also an indispensable technology trend to provide more accurate predictive results
(Gans et al. [2]).
Another challenge of asset pricing is that there are a number of characteristics or
attributes of the asset described through the text. These texts are unstructured data
because they are not defined on how to organize the content of the data, depending
on the writer’s style, or the specific characteristics of each type of assets, which
leads to the complexity and costly cost of mortgaging an asset through reading and
analyzing descriptive texts manually. The traditional searching way was proposed
(Thanh et al. [3]) to rely on Google search engine and Xpath pattern, but need a
good structured built system with a little big computational cost. Our solution to this
problem is to use natural language processing techniques for Vietnamese based on
artificial intelligence technology.
In terms of implementation method, assets pricing will be divided into two phases.
The first phase extracts the attributes of an asset from the asset description paragraph.
This is done based on Named Entity Recognition (NER) thanks to a combination
of several variants of artificial neural networks (ANN) and the model Conditional
Random Fields (CRF). There have been many methods of solving the NER problem
for English and Chinese (Yadav et al. [4]), but there have been some results for
Vietnamese in recent years (Hoang et al. [5], Duong et al. [6], Phuong [7]) with
bidirectional LSTM-CRF as the core technique (Hochreiter and Schmidhuber [8],
Huang et al. [9]). In this paper, we use the same method as [5] but with a more complex
set of attributes and data. After this phase, the asset’s information will be presented
as a structured form, each attribute being an identified property of the asset. The
next phase is to build a assets pricing model. Typically, we can use classic machine
26 Application of Artificial Intelligence to Asset Pricing . . . 361

learning models [10] or variants of ANN models [11] for regression problems. As
far as we are concerned, we use a Decision Tree model, which is consistent with the
data we already have, to estimate the asset value. This method has been applied to
mobile phone pricing (Asim and Khan [12]) and car pricing (Thai et al. [13]) based
on attributes that describe asset. The main difference between the results of this paper
with [12, 13] is the construction of the network structure as well as the parameter set
for specific data and Vietnamese text data mining. To demonstrate the empiricalness
of the model proposed in this paper, we use a mobile descriptive data set consisting
of 20,000 descriptions, of which 2000 descriptions have been labelled and used to
train the model.
The paper is divided into five sections. After the introduction section, Sect. 26.2
presents the theoretical preliminary, including the main concepts and techniques used
to solve the problem. The solution to the asset pricing problem will be presented in
Sect. 26.3. The next section gives some experimental computation for evaluating the
accuracy of the proposed model. Finally, some general conclusions are given.

26.2 Theoretical Preliminary

Below are some detailed descriptions of key concepts and techniques for building an
asset pricing algorithm model. As mentioned above, the asset pricing model consists
of two phases. The first phase solves the Named Entity Recognition (NER) problem
to extract attributes of assets. This problem is described in Sect. 26.2.1. Artificial
neural network techniques and conditional random fields to solve the NER problem
will be presented respectively in Sects. 26.2.2 and 26.2.3. The last section presents
the decision tree regression model that used to estimate the price of an asset based
on the attribute set extracted in the previous phase.

26.2.1 Named Entity Recognition

Named Entity Recognition (NER) is a problem of identifying proper names or enti-

ties from documents and classifying them into a set of predefined types such as
people, organization, and location. This is one of the most important problems in
the information extraction class. The main task of this problem is to identify the
representations in the text and then classify them into predefined groups such as
the name of a person, organization, location, time, currency, etc. This problem also
includes receipt, face, extract characteristics and attributes related to the entity. For
example, extract brand attributes, manufacturers, sizes, colours, etc. from mobile
phone description.
362 T. N. Thang et al.

26.2.2 Artificial Neural Network

Artificial Neural Network (ANN) is a model that simulates the structure of the human
brain, thereby performing the same functions as the human brain, such as image
recognition, speech synthesis, reasoning, learning, planning, etc. The artificial neural
network is a typical machine learning model in the field of artificial intelligence. The
basic architecture of the ANN model consists of many neuron layers, in which each
layer consists of many nodes, each node simulating the structure of a neuron. Nodes
of different network layers may be interconnected. The deep ANN model usually has
more than two network layers, that is, in addition to the input and output layers, there
are hidden layers. In recent years, these deep learning models have revolutionized
the application of artificial intelligence because these models provide results with
greater accuracy than previous models.
This paper uses advanced deep learning models applied to the problem of natural
language processing, namely of named entity recognition problem with the input
of a text or a string of characters and outputs are the values (substring strings) of
given assets. The Recurrent Neural Network (RNN) simulates how the brain reads,
memorizes, and extracts information. One of the typical regression networks is the
Long Short Term Memory networks (LSTM). The outstanding feature of LSTM is
that it can learn long dependencies, that is, remembering and processing information
or characters from the previous states far from the current state. As a result, the model
learns more information and gives more accurate predictive results.
The Gated Recurrent Unit (GRU) is a newer generation of neural networks and
is similar to LSTM. The GRU provides the cell state and uses the hidden states
to transmit information. The GRU network is a variant of LSTM so it has similar
characteristics to LSTM. However, it can be seen that the GRU network has a simpler
structure than LSTM. It combines forgotten ports and input into an update gate. It
also has no cell state. As a result, network parameters and calculations are reduced.
So, in fact, the GRU network works faster than LSTM.
Bidirectional GRU (Bi-GRU) bidirectional regression network model, or similar
to Bi-LSTM, was developed based on the GRU network. Typically, a network is built
by connecting two GRU nodes in opposite directions with the same output, making
it possible for the network to receive information from past and future states, thereby
increasing the amount of information for the input in a network.

26.2.3 Conditional Random Field

The conditional random field (CRF) is a probability model commonly applied to

structural predictions in machine learning and pattern recognition. A typical classi-
fication model predicts a class (also called a label) for a data sample without regard
to its adjacent or related samples, while the CRF model takes into account contexts
or neighboring patterns that affect the labelling process. The CRF model has many
26 Application of Artificial Intelligence to Asset Pricing . . . 363

applications in natural language processing, such as labels prediction for input data
series. We denote that X is a random variable corresponding to the data series to be
labelled, Y is a random variable corresponding to the label string. Each component
of Y is a random variable that receives values in a finite set of state S. In the NER
problem, X can take values as sentences in natural language, Y is a random sequence
of named entities corresponds to these sentences and each component of Y has a
value set that has all entities name labels (names of people, names of places,...). X
represents the observed data series and Y represents the label or state series. Our
main task when considering the CRF model is how to rely on the model to label
the observed data series, thereby performing the attributes extraction task. This is
equivalent to maximizing the probability distribution between the state series y and
the observed data X.

26.2.4 Decision Tree

In the field of machine learning, a decision tree is a type of predictive model, that
is, a mapping from observations of an object or phenomenon to conclusions about
their objective value. Decision tree can be used for both classification problems and
regression problems called Classification tree and Regression tree respectively. When
building a decision tree, we will need to create rules to split the original data into
smaller data regions. These rules will be applied to the characteristic attributes of
the data. For each selected asset, we divide the data by its attributes into subsets
and continue repeating until the sub-datasets have the same consistency of output.
In this paper, we use the regression decision tree to build a price forecast model
with the target value of the asset price. For the regression decision tree, we use the
standard deviation to assess the uniformity of the data. After taking steps to break
down the data, we can obtain a tree structure that can be very complex, so we need
to define a stop criterion for the algorithm. At each subdivision step we will perform
the homogeneity score calculation of that data and if its score is satisfied, we will
stop the data division there.

26.3 Solution Overview

As mentioned, the problem of asset pricing with input is the text describing the assets
in Vietnamese and the output is the value of the asset. The method that solves this
problem has two phases (Fig. 26.1). Phase 1 extracts the attributes of the assets from
the descriptions paragraph by solving the NER problem. Phase 2 implements assets
pricing based on the features set of assets using the decision tree method. Below is
a detailed description of how to perform each phase.
364 T. N. Thang et al.

Fig. 26.1 Overview of asset pricing model

26.3.1 Phase 1

In the first phase, extracting attributes of a asset from the description is performed by
solving the NER problem. The input of the problem is the text describing the products
in Vietnamese. The output is the attribute set of the asset. The model solving the
problem consists of two layers of Bi-GRU stacked on top of each other (Fig. 26.2).
Two GRUs are trained on the same input sequence in different directions to form a
Bi-GRU layer. Stacking 2 Bi-GRU layers are meant to make the model learn more
complexly. The following Bi-GRU layer combines the performances learned from
the previous layer and create new representations at a high level of abstraction. So
that we hope our model has more accuracy. The result of the top Bi-GRU layer is
transmitted to a fully connected network layer followed by a softmax layer, which
makes the corresponding result is a probability distribution vector.
In the CRF layer, the evaluation score of a prediction sequence is calculated based
on two components. The first component is a probability distribution vector. The
second component is the probability of switching between two labels. This transition
probability is learned by the CRF model. We will then select the prediction sequence
that yields the highest score as a result of the model. The Bi-GRU layer in Fig. 26.2
can be replaced by a Bi-LSTM layer. Comparison results are given in the following
section.

26.3.2 Phase 2

In the second phase, for attributes-based pricing, the attributes of the assets are used
extracted from Phase 1 in combination with the price of the assets to a build price table
and automatically predict prices. We use the regression tree to estimate prices. First,
we build the base price table including extracted attributes and associated prices.
The regression price prediction model will then be developed, with the training data
being the base price table. From the regression model, we can predict the price of
the asset knowing the described attributes.
26 Application of Artificial Intelligence to Asset Pricing . . . 365

Fig. 26.2 The implemented

model of Phase 1

26.4 Computational Experiments

26.4.1 Data Description

The data used in this paper is 20,000 mobile phone descriptions (Table 26.2), includ-
ing the Numerical order, Attribute description, and Value columns. In particular, the
data for the training model in Phase 1 was 2000 phone descriptions. After performing
the data tagging, the dataset is divided into two subsets: the training set (including
1700 descriptions) and the test set (including 300 descriptions) with the number of
values for each attribute are given in Table 26.1.

26.4.2 Implementation and Evaluation of Model

26.4.2.1 Phase 1

To extract the attribute set from the descriptions, we perform the named entity recog-
nition model with 17 attributes of name, SIM, brand, accessories, application, screen
size, colour, manufacturer, model number, type, RAM, model, condition, memory,
screen, made-in by Bi-GRU model and Bi-GRU model combined with CRF, referred
to as Bi-GRU & CRF model. The output is the attribute set of 20,000 assets given in
Table 26.3.
There are many ways to evaluate the results of the NER problem, such as using
3 evaluation criteria including Precision, Recall, and F1 of each attribute. Assuming
366 T. N. Thang et al.

Table 26.1 Statistics of labels by each attribute in the training data

Name of attribute Number of train samples Number of test samples
Model number 959 154
Type 1518 284
Brand 1550 288
Model 1601 274
Name 529 184
Color 916 179
Condition 1022 216
SIM 427 86
Screen size 173 29
Application 471 36
Accessories 3592 667
Made-in 195 25
Manufacturer 162 43
Screen 216 39
Memory 95 26
RAM 14 25
Operating system 38 3

Table 26.2 Raw dataset

Numerical order Attribute description Value
1 Dien thoai di dong Nokia 410,6858.28
625H RM-943 CV VN YEL...
2 Dien thoai di dong Nokia 410,6858.28
625H RM-943 CV VN YEL...
3 Dien thoai di dong Nokia 410,6858.28
625H RM-943 CV VN YEL...
... ... ...
19,998 A00018755#&NOKIA 225 723,001.38
DS RM-1011 NV NENA 1...
19,999 A00018755#&NOKIA 225 723,001.38
DS RM-1011 NV NENA 1...
20,000 A00018755#&NOKIA 225 723,001.38
DS RM-1011 NV NENA 1...
26 Application of Artificial Intelligence to Asset Pricing . . . 367

that considering attribute A, we denote TP (True Positive) as the number of attributes

A correctly classified, FP (False Positive) is the number of non-A attributes classified
as A, FN (False Negative) is A attribute number is classified as not A. From there,
we determine the evaluation index according to the following formula:

• Precision is the percentage of entities classified correctly among the entities clas-
sified as type A, calculated by the formula:

TP
Precision = (26.1)
TP + PN

• Recall is the ratio of the number of entities classified exactly to type A among
those actually belonging to type A, calculated by the formula:

TP
Recall = (26.2)
TP + FN

• The F1 -score is the equilibrium representation of Precision and Recall, calculated

according to the formula:

2 ∗ Precision ∗ Recall
F1 -score = (26.3)
Precision + Recall

Table 26.4 gives the F1 -score for each attribute. This shows that the Bi-GRU &
CRF model outperforms the Bi-GRU model with more than half of the dataset’s
attributes. So we can conclude that the Bi-GRU & CRF model is more suitable for
the problem.

Table 26.3 Attributes set after extracting raw description

Actual price Description Color Model number ... Brand
4,106,858.28 Dien thoai di yellow 625H ... Nokia
dong Nokia
625H RM-943
CV VN
4,106,858.28 Dien thoai di yellow 625H ... Nokia
dong Nokia
625H RM-943
CV VN
4,106,858.28 Dien thoai di yellow 625H ... Nokia
dong Nokia
625H RM-943
CV VN
... ... ... ... ... ...
368 T. N. Thang et al.

Table 26.4 F1 -score based on Bi-GRU model and Bi-GRU & CRF model
Name of attribute Bi-GRU Bi-GRU & CRF
Model number 0.601 0.863
Type 0.923 0.962
Brand 0.826 0.91
Model 0.726 0.813
Name 0.520 0.722
Condition 0.810 0.947
Screen size 0.363 0.787
Accessories 0.857 0.909
Made-in 0.679 0.794

Table 26.5 NER result of Bi-GRU + CRF model

Name of attribute Precision Recall F1 -score
Model number 0.863 0.863 0.863
Type 0.956 0.956 0.962
Brand 0.900 0.919 0.91
Model 0.799 0.828 0.814
Name 0.637 0.832 0.722
Color 0.772 0.903 0.832
Condition 0.936 0.958 0.947
SIM 0.701 0.809 0.751
Screen size 0.727 0.857 0.787
Application 0.8 0.75 0.774
Accessories 0.914 0.905 0.909
Made-in 0.714 0.893 0.794
Manufacturer 0.727 0.930 0.816
Screen 0.75 0.846 0.795
Memory 0.714 0.962 0.82
RAM 0.821 0.92 0.868
Operating system 1.0 1.0 1.0

Table 26.5 shows the extracted results of all attributes using the Bi-GRU & CRF
model based on all 3 evaluation indicators. This table shows that the Bi-GRU & CRF
model gives good results.
Next, we extract the model, brand, made-in and condition attributes using the
Bi-LSTM and CRF model and compare them with the Bi-GRU & CRF model.
The graph in Fig. 26.3 shows that the F1 -score of the two models is almost equiv-
alent. However, the Bi-GRU & CRF model has a shorter training time than the
Bi-LSTM and CRF model.
26 Application of Artificial Intelligence to Asset Pricing . . . 369

Fig. 26.3 Comparison chart of Bi-LSTM + CRF and Bi-GRU + CRF models

Table 26.6 The base price table

Actual price Application Color Name Manufacture Model ... Brand
5,501,014.32 −1 BLACK −1 −1 RM-1019 ... Samsung
5,501,014.32 −1 BLACK −1 −1 RM-1019 ... Samsung
5,501,014.32 −1 BLACK −1 −1 RM-1019 ... Samsung
... ... ... ... ... ... ... ...

Table 26.7 The result with price table

Error (%) 1 2 3 4 5 6 7 10 20
Accuracy (%) 95.45 95.78 96.0 96.2 96.33 96.53 96.75 97.15 97.53

26.4.2.2 Phase 2

First of all, we have a price table of 16,000 records with columns that are asset
attributes and corresponding attribute values. A regression tree will then be built to
create a base price table (Table 26.6).
We make predictions with the regression decision tree model and get the result
given in Table 26.7.
With the accuracy rate is the ratio between the number of correct predicted records
within the allowed error level on the number of testing records, namely:
370 T. N. Thang et al.

Number of correct predicted records

Accuracy = (26.4)
Number of testing records

A record is called a correct prediction if the difference between the predicted price
and the correct value is within the allowed error levels, here from 1% to 20%.

26.5 Conclusion

In this article, we have applied artificial intelligence algorithms in asset pricing

through asset descriptions in Vietnamese. The proposed method uses the named entity
recognition (NER) technique with a Bi-GRU neural network model (Bi-LSTM) in
combination with the CRF model to extract the attributes of assets, thereby building
a regression model to evaluate the assets based on the attribute set. With this method,
the price system can intelligently predict when the input is the description of the asset
in Vietnamese, without the need for a given format. The method works relatively well
with a dataset of mobile phone descriptions. The pricing method proposed here is
applicable to many other assets such as computers, cars, and real estate.

Acknowledgements This research is supported by Hanoi University of Science and Technology

(HUST) and CMC Institute of Science and Technology (CIST).

References

1. Suong, N.H., Linh, N.V., Tai, H., Vuong, P.T. : An efficient model for sentiment analysis of
electronic product reviews in Vietnamese. arXiv:1910.13162 (2019)
2. Gans, J., Goldfarb, A., Agrawal, A.: Prediction Machines: The Simple Economics of Artificial
Intelligence. Harvard Business Review Press (2018)
3. Thanh, V.T., Dat, N.Q.: A Vietnamese information retrieval system for product-price.
arXiv:1911.11623 (2019)
4. Yadav, V., Bethard, S.: A survey on recent advances in named entity recognition from deep
learning models. arXiv:1910.11470 (2019)
5. Hoang, P.T., Phuong, L.H.: End-to-end recurrent neural network models for Vietnamese named
entity recognition: word-level versus character-level. In: International Conference of the Pacific
Association for Computational Linguistics, vol. 2017, pp. 219–232 (2018)
6. Duong, N.A., Hieu, N.K., Vi, N.V.: Neural sequence labeling for Vietnamese POS tagging and
NER. arXiv:1811.03754 (2018)
7. Phuong, L.H. : Vietnamese named entity recognition using token regular expressions and
bidirectional inference. arXiv:1610.05652 (2016)
8. Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural Comput. 9(8), 1735–1780
(1997)
9. Huang, Z., Xu, W., Yu, K.: Bidirectional LSTM-CRF models for sequence tagging.
arXiv:1508.01991 (2015)
10. Anh, N.T.N., Tu, N.D., Solanki, V.K., Giang, N.L., Thu, V.H., Son, L.N., Loc, N.D., Nam,
V.T.: Integrating employee value model with churn prediction. Int. J. Sens. Wirel. Commun.
Control (2020). https://doi.org/10.2174/2210327910666200213123728
26 Application of Artificial Intelligence to Asset Pricing . . . 371

11. Anh, N.T.N., Dat, N.Q., Van, N.T., Danh, N.N.: An: wavelet-artificial neural network model
for water level forecasting. In: 2018 International Conference on Research in Intelligent and
Computing in Engineering (RICE), San Salvador, pp. 1-6 (2018). https://doi.org/10.1109/
RICE.2018.8509064
12. Asim, M., Khan, Z.: Mobile price class prediction using machine learning techniques. Int. J.
Comput. Appl. 179(29), 0975–8887 (2018)
13. Thai, D.V., Son, L.N., Tien, P.V., Anh, N.N., Anh, N.T.N.: Prediction car prices using quantify
qualitative data and knowledge-based system. In: International Conference on Knowledge and
Systems Engineering, pp. 1–5 (2019)
Chapter 27
A Survey on Decentralized
Crowdsourcing Using Blockchain
Technology

M. Preetha, K. Elavarasi, A. Mani, E. Pavithra, P. Sudharshna,

and S. Rakshini

Abstract Many of the emerging crowd sourcing platforms find central repositories
as single point of failure, and are related to the shortcomings of conventional trust-
based models. Thanks to involvement of malicious users, they are also vulnerable to
distributed denial of service (DDoS) and Sybil attacks. Additionally, high service fees
from the community sourcing platform can hinder crowd sourcing growth. It has both
research and substantial value for tackling these potential problems. In this article,
CrowdBC, a blockchain-based cooperative crowd sourcing system through which
a community of employees may solve the requester’s challenge without investing
in some third intuition. The safety of consumers is ensured and allows less traction
costs. In specific, the new system architecture will be adopted. Implementation of a
software model on the public test network Ethereum with real-world dataset. Exper-
iment outcome reveals the viability, efficiency, and scalability of the crowd sourcing
method.

Keywords Blockchain · Smart contract · Crowdsourcing · Decentralized system

M. Preetha (B) · K. Elavarasi · A. Mani · E. Pavithra · P. Sudharshna · S. Rakshini

Department of Computer Science and Engineering, S.A. Engineering College, Chennai, Tamil
Nadu, India
e-mail: [email protected]
K. Elavarasi
e-mail: [email protected]
A. Mani
e-mail: [email protected]
E. Pavithra
e-mail: [email protected]
P. Sudharshna
e-mail: [email protected]
S. Rakshini
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 373

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_27
374 M. Preetha et al.

27.1 Introduction

27.1.1 Blockchain

It is block of chain, requesting information. The stored data in a block depends on

blockchain type. For example, A Bitcoin Block contains Sender, Receiver informa-
tion, number of bitcoins to be transferred. The first block in the chain is called genesis
block.

27.1.2 Working of Blockchain

When new data are stored in a block it is added to the blockchain. As its name,
Blockchain contains multiple blocks held together. Following are the steps to add a
block to the blockchain
• A transaction must appear.
• That transaction needs to be checked.
• That transaction has to be kept in a block.
• There must be a hash to that block.

27.1.3 Smart Contract

A smart contract is a self-execution contract whereby the terms of the agreement

between requestor and worker are written directly into lines of code. The code
and the agreements it contains exist across a network of distributed, decentralized
blockchains. The code controls execution, and trackable and irreversible transactions.

27.1.4 Crowdsourcing

Crowdsourcing is a supply chain model in which organizations procure goods and

services through a broad, relatively open and often rapidly evolving group of Internet
users, including ideas and finances; it divides work among participants in order to
achieve a cumulative result. The term crowdsourcing itself is a portmanteau for
crowdsourcing and outsourcing, and was coined in 2006.
27 A Survey on Decentralized Crowdsourcing Using Blockchain … 375

27.1.5 Consensus Algorithms

• Proof of work
• Proof of stake.

27.1.5.1 Proof of Work

It is a consensus algorithm that deters the DDOS attack and SYBIL attack. A gift is
given to first miner who solves each block problem. Miners compete to the pc power
to be the first to find the answer.

27.1.5.2 Proof of Stake

It is a type of consensus algorithm by which a cryptocurrency blockchain network

aims to achieve distributed consensus. In POS there is no reward block therefore
miners take the transaction fees. POS concurrencies can be thousand time more cost
efficient.

27.2 Literature Survey

In To et al. [1] Spatial Crowdsourcing (SC) is a platform for collecting and observing
environmental and other spatio temporal information. Spatial crowdsourcing gives
spatio-temporal tasks to a group of workers, i.e. Individual with mobile go to the
location and perform the task. In current solution the workers need to give their
privacy information to third party. In this paper, a plan for protecting location privacy
of participated worker in SC task is introduced. Differential privacy and geocasting
are two mechanism used to provide service by ensuring worker privacy. Location
privacy is guaranteed by real word experiment results.
376 M. Preetha et al.

In Zhuo et al. [2] mobile crowdsourcing is available from smartphones, the task
owners can use their sensor-rich mobile devices to crowdsource data from smart-
phone users. Collection and processing of data is a challenge when the amount of
data is large. In big data processing, data collection and set operations are used
for extracting unnecessary data. Faced with difficulties in terms of scarce computa-
tional and storage capacity, cloud-assisted solutions will work promisingly thanks to
solving the problem of big data analysis. When their data is not stored in the cloud,
the staff do not participate. In this work, cloud is used to perform the worker’s set of
operations and preserve data privacy as well. The requester can also check the output
of the set operation. We also extend our scheme to support pre-processing of data
which often excludes invalid data before analysing the data. The proposed mech-
anism reduces the computational cost by using batch verification and data update
methods.

In Christian and Roger [3] Bitcoin may be a digital currency which does not believe
in a centralized system that is unlike traditional currencies. Instead, Bitcoin relies on
a freewill network that executes and validates transactions via a distributed ledger.
Here we discuss how Bitcoin uses a multi-hop broadcast to spread transactions and
blocks to update the ledger duplicates through the network. What can be achieved
by pushing the current protocol to its limit with unilateral behavioural changes of
the client.
In Buccafurri et al. [4] the data reassurance within crowdsourcing cannot give
their time only to a single person but should be given to group. Blockchain allows
this, because the whole group receives the transaction and it will be verified by
miners. Proof of labor algorithm is used to verify the transaction. Highest-fee-first-
served policy is followed by miners. this makes the small transaction not suitable for
blockchain. To overcome this twitter is chosen to provide proof of concept.
In Ouyang et al. [5] it is very important to develop a reliable crowdsourcing
application which will detect truth from noisy and disagreed data from different
information. For big data, correct truth discovery algorithm must be scalable. Existing
algorithms are not scalable, they perform that do not scale. In this work, parallel
and streaming truth discovery algorithms are proposed. Efficiency and scalability is
guaranteed by real world experimental results.
27 A Survey on Decentralized Crowdsourcing Using Blockchain … 377

In Feng and Yan [6] the mobile crowdsourcing network (MCN) is an auspi-
cious network architecture that uses crowdsourcing architecture to perform task with
mobile devices and human involvement. It has some security and privacy issues that
make the MCN application difficult.

In this work, to overcome security and privacy issue a general architecture is

proposed that comprise of crowdsourcing sensing and computing. This work is
expected to do more survey on security and privacy solutions for MCN.
In Zhang et al. [7] Crowdsourcing is a turnup example where the users get their
task done by paying fees. The problem in crowding sourcing is that whether the
user pay or receive for what they deserve. False-reporting and free-riding make the
system weak to false users. In this work, two mechanisms is designed. First is EFF
mechanism which fires false behavior with the help of third part another is DFF
mechanism which, removes false behavior without help of third party. EFF removes
free-riding and false-reporting. DFF is semi-truthful, and misconduct such as free-
riding and false reporting will be fired. The system both guarantees transaction-wise
budget-balance and computational efficiency. Performance assessment shows that
by being false no users can be gained in our system.
In Zhang and van der Schaar [8] in recent years, crowdsourcing platforms have
arisen that enable requestors from all over the world to post tasks and receive support
from an increasingly global pool of staff. However, intrinsic problems reside in the
application of crowdsourcing as workers and as requesters are selfish and aim to
increase their own benefit. During this paper, incentives for workers were provided
to use a new game theoretical model based on repeated games to exert their efforts.
As there is still a gap within the welfare between the non-cooperative equilibrium that
occurs as workers seek their self-interest and therefore the optimal, productive Pareto
outcome. It was suggested that a relatively recent class of collectively supported
compensation protocols integrate reputation mechanisms into the dominant pricing
systems currently implemented on crowdsourcing platforms, etc. The exchanges
were first developed on a crowdsourcing network as a two-sided rivalry where the
378 M. Preetha et al.

requesters and employees are partnered and frequently play gift giving games. We
then discuss the question of the protocol developer finding an optimal and feasible
protocol that ensures the website’s very best wellbeing. We prove that the proposed
reward protocol will make the website run on the verge of productivity at Pareto. We
also look at an alternate situation where the protocol author aims to optimize website
sales and determine optimum protocol efficiency.
In Zhang et al. [9] a job that occurs during a Delay Tolerant Network (DTN) on
some node. How to publish the work pressure during pairly contacts to reduce span
A centralized polynomial-time optimal algorithm was proposed and D2 protocol was
developed, Which control r-hop neighbourhood information at a single node nodes.
Trace-driven simulations is used to show that, D2 with 1-hop Knowledge over a large
collection of environments is near-optimal. The performance gap grew smaller due
to an increase in the amount of information held by an individual.

In Cheung et al. [10] with large number of mobile users and sensors downloaded
in our mobile, the emergence of the commercial mobile crowdsensing applications
is witnessed that combine the mobile technology with crowdsourcing to give users,
time-sensitive and location dependent information. Motivated by these real-world
applications, selection of task problem for different kind of users with different loca-
tions, cost and speed. Computation leads to NP-hard problem. So ADTS algorithm is
proposed to allow users to make the selection of task by their own. The confluence of
the algorithm is proved. The ADTS scheme reaches highest the Jain’s fairness index
and coverage compared to benchmark algorithms. Finally, explained how mobile
users are formulated under the ADTS.
In Federico Ast [11] court system, mail and uncensored newspapers are three
actors built by Jeremy Bentham and Madison. The postal service and newspapers
were suffering due to doing new things because of the digital revolution. In this work,
CrowdJury was proposed. It combines the merits of crowdsourcing and blockchain
to make both a transparent and self-sustained justice system. CrowdJury is trying to
abandon the principles for a collective intelligence-based peer-to-peer court system.
In Zhu and Zhou [12] web-based equity crowdfunding could represent a new
channel for start-ups to raise money. It has low entry barriers, low cost and high
speed, thereby encouraging creativity. Recent years have seen some crowdfunding
innovations in China. However, in reality some problems still remain unresolved.
27 A Survey on Decentralized Crowdsourcing Using Blockchain … 379

Blockchain is a decentralized and distributed ledger technology designed to guar-

antee data security, accountability and honesty. Because the technology cannot be
manipulated or copied, it is considered within the finance industry to have great
potential. Based on the characteristics of the study of blockchain technologies, this
work would also pour into applications for equity crowdfunding. (1) Blockchain
low-cost development system for crowdfunding asset registration; (2) Blockchain
technology enables the sale and distribution of crowdfunding resources and thereby
promotes its circulation; (3) Blockchain technology allows peer to see transactions
between businesses and companies, and addresses compliance management issues
and financial advisory security issues; (4) Blockchain technology aims to develop an
election process for fund managers to enable them to engage in financial regulation.
It preserves the rights and interests of small capitalists; (5) Blockchain technology
helps regulators achieve economic environment and facilitates compliance practices
such as battle for risk protection and cover-up.

27.3 Real Time Examples

• Waze com is an example of spatial crowdsourcing, a free, spatially crowd sourced

GPS navigating application on android [13]. The drivers can report traffic jams,
accidents to police so that me other users on the same road are aware of the scenes
beforehand. Another aspect of this is reporting petrol bunk prices. Without any
effort the users can do those tasks. Other winning examples are Uber, GigWalk.
• Smart contracts to track stock or bond movements as they move from one owner
to the next. Stocks, especially when the financial markets are in flux, can change
hands at lightning speed [14]. But with smart contracts, traders can buy and
sell faster, as each transaction is reviewed and reported automatically on the
blockchain. This reduces cost per transaction and creates a more fluid marketplace.
• MedRec is an application that is developed to maintain records of users when
moving from one organization to another [15].
• ABRA is a crypto currency wallet that allows users to invest in crypto currencies
and monitor their financial assets [16]. Now it is available in iOS and Android.
• Reliance Industries announced that it is going to develop a blockchain-based
supply chain platform with its own crypto currency called Jiocoin. Both Jiocoin
and bit coin are 99a [17].
• OpenBazaar is decentralized market place where people are connecting directly
and securely through p2p system [18].
• Webjet is the online travel application that allows tracking the empty rooms, with
payments done fairly to the network of middle-party sites [19]. It also allows the
customers to compare, combine and book tickets.
380 M. Preetha et al.

S. No. Title Proposed work Advantages Disadvantages

1. Private location security for Differential Compromising Both worker and
employees’ databases in privacy and location privacy task is need to be
geographic crowdsourcing geocasting protected
mechanism
2. Privacy-saving verifiable Cloud is used Privacy of Computational
set operation in big data for identity is costs of the
mobile crowdsourcing maintained and system are high
assisted by the cloud the requestor
may verify the
accuracy of the
set operation
results
3. Information propagation in Consensus Duplicate No solutions are
the bitcoin network algorithm ledger are given to intrinsic
updated problems
4. Tweetchain is an alternative Proof of concept Eliminates the Formalize our
to blockchain for algorithm proof of work protocol in a
crowd-based applications result more formal
To ensure the manner, to
transaction examine it in
security mesh detail with
chain of tweets respect to health
5. Parallel and streaming truth Parallel and Truth discovery Streaming
discovery mechanism in streaming truth is effective and algorithm
large-scale measurable discovery scalable in performance has
crowdsourcing algorithm large-scale to be compared
measurable
crowdsourcing
applications
6. MCS-chain—decentralized Novel consensus Solves the issue Needs more
and trustworthy mobile mechanism of the fork and privacy in MSC
crowdsourcing based on centralization chain
blockchain of most of the
popular ledger
networks
7. Keep your promise: EFT and DFT Free riding and Nil
free-riding and mechanism inaccurate
inaccurate-reporting reporting is
mechanism in removed
crowdsourcing
8. Reputation-based incentive Incentive Free-riding and Nil
protocols in crowdsourcing protocols are incentivize
applications designed based on problem has
reputation been
mechanism overcomed
(continued)
27 A Survey on Decentralized Crowdsourcing Using Blockchain … 381

(continued)
S. No. Title Proposed work Advantages Disadvantages
9. Minimum makespan Centralized Minimum Space is limited
workload dissemination in polynomial-time makespan
DTNs: making full optimal algorithm workload
utilization of computational and distributed publish
surplus around dissemination problem has
protocol been
overcomed in
DTNs
10. Time-sensitive role Asynchronous and Determine set Users cannot
collection spread in virtual distributed task user’s role join and leave
crowd sensing selection (ADTS) range and the system
algorithm mobility plans dynamically
11. CrowdJury, CrowdJury It combines the Nil
collaboration-era crowd framework merits of
sourced justice system crowdsourcing
and blockchain
12. A crowdsourcing worker Block chain The algorithm Difficulty in
quality evaluation application is accurate and answering time
algorithm on MapReduce has high and task
for big data applications efficiency and
performance in
a big data

27.4 Conclusion

• CrowdBC’s design, a blockchain-based, decentralized crowdsourcing framework

was presented in this paper.
• On the basis of research, the current centralized crowdsourcing method suffers
from leakage of data, a central point of failure and large usage charges.
• The CrowdBC was formalized to deal with these centralized issues.
• In the meantime, we have increased the flexibility of smart contract crowdsourcing
to describe complex crowdsourcing logic here.

References

1. To, H., Ghinita, G., Fan, L., Shahabi, C.: Differentially private location protection for worker
datasets in spatial crowdsourcing. IEEE Trans. Mob. Comput. 16(4), 934–949 (2017)
2. Zhuo, G., Jia, Q., Guo, L., Li, M., Li, P.: Privacy-preserving verifiable set operation in big data
for cloud-assisted mobile crowdsourcing. IOT J. IEEE 4(2), 572–582 (2017)
382 M. Preetha et al.

3. Christian, D., Roger, W.: Information propagation in the bit coin network. In: 2013 IEEE
Thirteenth International Conference on Peer-to-Peer Computing (P2P), pp. 1–10. IEEE (2013)
4. Buccafurri, F., Lax, G., Nicolazzo, S., Nocera, A.: Tweetchain: an alternative to blockchain
for crowd-based applications. In: International Conference on Web Engineering, pp. 386–393.
Springer (2017)
5. Ouyang, R.W., Kapla, L.M., Toniolo, A., Srivastava, M., Norman, T.J.: Parallel and streaming
truth discovery in large-scale quantitative crowdsourcing. IEEE Trans. Parallel Distrib. Syst.
27(10), 1045–9219 (2016)
6. Feng, W., Yan, Z.: MCS-chain—decentralized and trustworthy mobile crowdsourcing based
on blockchain (2019)
7. Zhang, X., Xue, G., Yu, R., Yang, D., Tang, J.: Keep your promise: mechanism design against
free-riding and false-reporting in crowdsourcing. IOT J. IEEE 2(6), 562–572 (2015)
8. Zhang, Y., van der Schaar, M.: Reputation-based incentive protocols in crowdsourcing
applications. In: 2012 Proceedings IEEE INFOCOM, pp. 2140–2148, Florida, USA (2012)
9. Zhang, S., Wu, J., Lu, S.: Minimum makespan workload dissemination in DTNS: making full
utilization of computational surplus around. In: Proceedings of the Fourteenth ACM Inter-
national Symposium on Mobile Ad Hoc Networking and Computing, pp. 293–296. ACM
(2013)
10. Cheung, M.H., Southwell, R., Hou, F., Huang, J.: Distributed time sensitive task selection in
mobile crowd sensing. In: Proceedings of the 16th ACM International Symposium on Mobile
Ad Hoc Networking and Computing, pp. 157–166. ACM (2015)
11. Federico Ast, A.S.: The CrowdJury, a crowd sourced justice system for the collaboration era
(2015)
12. Zhu, H., Zhou, Z.Z.: Analysis and outlook of applications of blockchain technology to equity
crowdfunding in china. Financ. Innov. 2(1), 29 (2016)
13. Vidhyalakshmi, D., Balaji, K.: Performance of bidirectional converter based on grid application.
Indonesian J. Electr. Eng. Comput. Sci. 12 (3), 1203 (2018)
14. Kumar, S.M. and Balakrishnan, G: April. Multi resolution analysis for mass classification in
digital mammogram using stochastic neighbor embedding. In 2013 Int. Conf. Commun. Signal
Process., (pp. 101–105). IEEE (2013)
15. Hemalatha, R.J. and Vijayabaskarin, V : Histogram based synovitis scoring system in ultrasound
images of rheumatoid arthritis. J. Clin. Diagn. Res., 12(8), (2018)
16. Pendem, S. and Ramesh, G.P : 75 GHz 5G Frequency spectrum analysis. In Recent Trends
and Advances in Artificial Intelligence and Internet of Things (pp. 165–176). Sprin-ger, Cham
(2020)
17. Ramesh G.P., P. S : Design and implementation of U-Shape microstrip patch antenna for
bio-medical application. Int. J. Adv. Sci. Tec. 28(12), 364–374
18. Shahada, S.A.A., Hreiji, S.M., Atudu, S.I. and Shamsudheen, S :2019. Multilayer Neural
Network Based Fall Alert System Using IOT. International Journal of MC Square Scientific
Research, 11(4), pp.1-15
19. Badawi, W.A : UNDERGROUND PIPELINE WATER LEAKAGE MONITORING BASED
ON IOT. Int. J. MC Square Sci. Res. 11(3), pp.01–08
Chapter 28
A Study of Student’s Subjective
Well-Being Through Chatbot in Higher
Education

Shivani Agarwal and Nguyen Thi Dieu Linh

Abstract The present research discusses about the issue of student’s subjective
well-being which is very much based on one’s own cognitive judgment. This chapter
empirically investigates what effect chatbots have on subjective well-being of the
students and how subjective well-being can be measured in terms of this research
context. The major findings of the study are that students’ subjective well-being
has been measured into three dimensions. The first dimension of subjective well-
being is Life satisfaction which represented that overall students feel satisfied after
conversing with the chatbot. The second dimension is Positive affect which is also
increasing after conversing with the chatbots. The third dimension is Negative affect
which is decreasing among the students after conversing with the chatbots. So, it is
recommended to the college and university to implement chatbots in their premises
for the wellbeing of their students.

Keywords Chatbots · Subjective well-being · Students · Higher education ·

Research · Life satisfaction

28.1 Introduction

In the competitive environment and the increasing demands of twenty-first century,

the student community is facing a lot of changes in the curriculum, in classroom
teaching, in the quick or easily available information/knowledge at a single click.
No doubt the freely available knowledge is abundance in nature. But to extract that
knowledge and induced that in themselves is a major challenge among students.

S. Agarwal (B)
KIET School of Management, KIET Group of Institutions, Delhi-NCR, Ghaziabad 201206, India
e-mail: [email protected]
N. T. D. Linh
Hanoi University of Industry, Hanoi, Vietnam
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 383

V. E. Balas et al. (eds.), Further Advances in Internet of Things in Biomedical
and Cyber Physical Systems, Intelligent Systems Reference Library 193,
https://doi.org/10.1007/978-3-030-57835-0_28
384 S. Agarwal and N. T. D. Linh

In the whole world there are lakhs of university which helps the students to sharpen
their talent and further prepare them to showcase the case in the corporate world.
In the present paper we talked about the students of India. In India itself, there are
approximately 935 total university, 50 central universities, 409 state universities, 127
deemed universities, 349 private universities [15]. As, India is the most populated
country in the world, it has approximately 19.9 million students who have attended
colleges and universities in 2019.
India is also popular for the young generation which means we have the most of
the population which belongs to 18–55 years of age.
The data provided by National Crime Records Bureau in 2015, approximately,
8934 (6.7% of all suicides) students are committing suicide every year. That’s means
in an hour one student commit suicide.
In India, people generally do not talks about the psychological constructs such as,
sadness, happiness, life satisfaction, depression, anger, mood swing, stress, anxiety
and so on. Our Students/young generations have a lot of pressures of dealing with
the personal life, professional life etc. The pressures belongs to family pressure,
classroom targets, assignments, subject loads, peer pressure etc. they are undergoing
a lot of things but in real sense no one cares about their subjective well-being among
the students.
In India people are generally cateogorised as introvert they did not easily share
their issues/problems be it positive or negative with the psychologist. So, the present
chapter proposes to implement a chatbot or AI which helps the students to share
their positive and negative aspects of life with that AI and feel relieved and mentally
sound to compete in the war of good marks, placement and finally able to locate/prove
themselves in the front of the family members.
To win the war of psychological pressure among students community, the emer-
gences of chatbot play its role. In the digital era, the role of recruiter, waiter, and
analyst is very well played by Artificial Intelligence (AI). AI bots were already being
touted as the future of higher education.
In a nutshell, though it’s because AI bots contribute towards making these colleges
and future ready. There are numerous reasons why colleges and universities around
the world have chosen to integrate online chat bots for dealing with admission
enquires with their websites and social media pages.
1. Save Time:
Students are well versed with technology and can handle multiple gadgets at a
single moment. They are young and have so many curiosities in the brain. These
kinds of queries can easily be handles with the AI as for human face to face
interaction, it requires a lot of time to invest in a single students.
By intervening through AI colleges, University can save both time and efforts.
Moreover, the level of satisfaction among the student community can also be
enhanced.
2. For admission Query:
Students while searching of a good college and university have a lot of queries
regarding the faculty, location of college, fees structure, course curriculum and so
28 A Study of Student’s Subjective Well-Being Through Chatbot … 385

on. These kinds of queries are common from every student whom so ever visits the
college/University portal. To handle all the queries AI is the effective solution to
answer timely. As time is money for both the students and the university/college
authorities.
3. Hub of information for universities:
The role of chat bots is considered as a hub of information from which thousands
of students can extract the information at a time. Chatbots are considered as an
asset for the higher education authorities. Chatbot can easily predict the potential
candidates for admission. Moreover, chatbot also plays the role of counselor to
the depressed potential candidates.

28.1.1 Background About Chatbot

A chatbot popularly known as by several names such as an artificial conversational

entity (ACE), talk bot, chatterbox or chat robot. It is a computer based program which
allows communicating with human being and generally based on natural language
via text or speech. Chatbot was first invented in the year 1966 at the MIT AI Labo-
ratory, named Eliza. The main objective was to give an precise simulation of a
human conversation. There is continuos upgradation in chatbot, it is now capable to
answer complex queries and have expanded abilities for example voice interaction
and machine. Chatbots are typically categorized which include: commerce, educa-
tion, entertainment, finance, health, news and productivity. Nowadays, most chatbots
are accessed via virtual assistants such as Google Assistant and Amazon Alexa, via
messaging apps such as Facebook Messenger or WeChat.

28.1.2 Application of Chatbot in Business

The chatbot are located in all the kith and kins of business be it finance, marketing,
human resource, production, supply chain, food and restaurants etc. There are certain
applications of chatbots induced in business which are as follows:
1. Available 24 by 7:
Chatbot are vailable 24 by 7 to solve the human queries despite of the fact that,
it is early morning, late nights, rainy season etc. Chatbot can communicate with
humans without frustrating and in upcoming days, may be it easily replaces
human interaction at the first level of business.
2. Communication:
Human can communicate with one person at a time. On the contrary chatbot can
connect with thousands of people without felling tired and respond immediately.
386 S. Agarwal and N. T. D. Linh

3. Within Budget:
Unlike humans, chatbot is a onetime cost investment affair. It did not required
monthly salary, incentives, PF and other benefits which are needed by human
beings.
4. Faster onboarding:
The processing time is also less as compared to human beings. AI helps in
generating more revenues in a less time. AI continuously helps in teaching all
the employees/staff/consumer how can they perform better in lesser time.
5. Work automation:
Unlike human beings, chatbots are automated in nature; they can perform repet-
itive tasks without getting bored, and feels lethargic. Like in hotel industry, they
can serve food, take orders from the consumers and can also prepare dishes
without any productivity issues.
6. Personal assistant:
Chatbot can also play a role of personal assistant to the consumers such as
students. Students can easily discuss about fashion, about their crush, about their
peers, about teachers, about their happiness and sadness etc.

28.1.3 Scope of the Study

The current study assessed about the well-beings of students regarding the counseling
app (chatbot) in college/universities. Diener et al. [7] suggested subjective well-being
as a combination of a cognitive judgement of satisfaction with life, the frequent
experience of both positive moods and emotions and as well as the experience of
pessimistic emotions. Hence, subjective well-being may be considered “a general
area of scientific interest rather than a single specific interest construct”. The three
dimensions of well-beings are:
1. Life Satisfaction with chatbot:
According to sharing information/thoughts about overall satisfaction with chatbot
means that students really find it ease at sharing their conditions of life,
satisfaction with life, how their life is ideal or not etc.
2. Positive Affect with chatbot:
Positive Affect with chatbot has been classified into happiness, attentive,
determined, alert, enthusiastic, etc.
3. Negative Affect with chatbot:
Negative Affect with chatbot has been classified into depress, sad, anger, distress,
sorrows, etc.
28 A Study of Student’s Subjective Well-Being Through Chatbot … 387

28.2 Literature Review

S. No. Author Year Findings Conclusion

1 Young [18] 2005 The major findings of the It can be concluded that
study is that Caucasian, subjects find it more
middle-aged males, were suitable to provide
most likely to use online information and seek
counseling who are guidance keeping their
having bachelors degree anonymity intact
of four years
2 Coughlan et al. [6] 2007 The purpose is to It has highlighted that the
confront with and suggest interface of web and its
how to move from face to evaluation can stimulate
face communication to social responses in
web based systems consumers and become
deeper of the real life
information consumer
experience
3 Prasetyaningrum and 2011 In this paper, the With a thorough layout of
Gregorius [11] researchers proposed the an online application the
Interactive Multimedia researchers explored the
Approach-called possibility of one
“m-Leadership”. This unanimous counselling
application is on mobile application for the entire
phones for Guidance and country
Counseling
4 Suyoto [14] 2013 This study has been The paper looks into the
suggested a design design and layout of
application “NingBK: counselling app and
Social GC” as a means of provides an example of
giving social counseling counselling app to make it
services assistance that easier to comprehend the
run on mobile phones functioning of online
counselling applications
5 Bhakta et al. [2] 2014 The result found when the It can be concluded that
engagement time was subjects provide more
increased with chatbot, reliable and true
student shares more information to the chatbot
information on sensitive with increase in length of
topics interaction. Furthermore,
questions related to
Sensitive topics require
more focus and attention
to get correct answers
(continued)
388 S. Agarwal and N. T. D. Linh

(continued)
S. No. Author Year Findings Conclusion
6 Eliamani et al. [8] 2015 The study found that that With the intervention of
providing guidance and guidance and counselling
counseling services has service the performance of
an effect in shaping students improved and
students’ attitude towards thus online counselling
studies and career choice can help focus the students
to perform better
7 Cameron et al. [5] 2017 The purpose of the study Using a chatbot for mental
is to sketch out the design health counselling can
of a chatbot to be used provide many benefit for
within mental health the user. Research has
counselling shown users find chatbots
safe and easy to talk to
8 Srivastava [12] 2018 The paper focused on two The research puts focus on
different aspects, Firstly the importance of policy
the cut in future job framework and research in
opportunities due to rise AI and its use in various
in Artificial Intelligence spheres of life, giving an
and secondly the rising opportunity to use AI in
opportunities of AR in the the field of human
fields of health care, Counselling
finance, law, information
browsing etc
9 Yorita et al. [17] 2018 The results showed that The study highlighted that
the conversation model careful and coordinated
and Sense of Coherence efforts can help measure
model are competent the actual stress level of
enough of determining individuals and thus a
level of stress and can be well-designed counselling
utilize by the Peer app can understand the
Support model to feeling of individuals
effectively select
appropriate support
actions
10 Sturk et al. [13] 2019 The results highlighted The research went on to
that 79% of total people put light on the need and
mentioned that they will benefits of implementing a
suggest the site to clients mental health app in rural
in the future areas where people find it
inconvenient to share
mental health problems.
The same can be said
about students who often
are uncomfortable to share
their problems
(continued)
28 A Study of Student’s Subjective Well-Being Through Chatbot … 389

(continued)
S. No. Author Year Findings Conclusion
11 Armas Sánchez [1] 2019 The study highlighted the The study concluded with
changing nature of human the need of proper
life, quoting “no one has guidance for students to
a clear idea of what skills tell them of the demands of
they are going to need in future thus Counseling app
5 years’ time.” This can help fulfill that gap
unpredictability is fueled
by the astonishing speed
in which technology is
changing the nature and
form of work, among
other things
12 Wilhelm et al. [16] 2019 The study shows that how The study suggested to
can we collectively move enhance the mental health,
toward these potential patients many of times
pitfalls and give a vision faces logistics issues like
for how we might develop lack of transportation,
powerful, scalable, work issues, childcare and
precisely timed, sometimes high costs of
personalized treatments. Thus giving a
interventions to enhance clear shot towards the
global mental health development of online
counselling applications
13 Blashki and Lock [3] 2019 In future, to make better AI has some useful
decisions AI mental applications in the mental
health interventions are health field. At this stage,
intended to learn and to it is especially of value in
adjust and change based early detection, diagnosis,
on experience and treatment and
assessing prognosis
14 Cameron [4] 2019 With the emergence of People enjoy interacting
Virtual therapist, with chatbots as they find
Chatbots are significantly it simple to use. There is a
used in dealing with constant personality
mental health throughout the
conversation, and the
chatbot do well at
onboarding
(continued)
390 S. Agarwal and N. T. D. Linh

(continued)
S. No. Author Year Findings Conclusion
15 Narote et al. [9] 2019 This chat bot would help The development of
the applicant train Chatbot application in
themselves for the numerous Programming
interview purpose Language was
implemented with a
objective of making a user
Interface to start
conversation. Also work
on an idea of Integrating
Chatbot in Facebook So
that common people Can
Make Use of It
16 Prasad et al. [10] 2020 The findings were that It can be concluded from
people suffering from this research that it has
depression can obtain been a testament to know
much-needed help that depression can be
without the feeling of mitigated by means of
being judged for having technology. It also help the
such mental illness people in choosing the
best doctor for them who
can guide them and can
answer all their queries
without judging them

28.3 Methodology

28.3.1 Participants and Procedure

In this research, the participants are the students studying in B.Tech, M.Tech, MBA,
BBA, BCA, and MCA. Basically a student belongs to higher education in India.
Through, Google form data was collected from the students from different colleges
located in NCR, India. A total of 188 reply were acknowledged. Out of 188, only
163 were found to be suitable for further analysis. The reliability of questionnaire
was measured and reported as. 816. Demographical profile of respondents is shown
in Table 28.1.
28 A Study of Student’s Subjective Well-Being Through Chatbot … 391

Table 28.1 Demographical details

Demographic No. of respondents Percentage (%)
n = 163
Age (in years) Young (18–23) 100 61.35
Middle age (23–27) 63 38.65
Gender Male 111 68.09
Female 52 31.91
Education Graduation 90 55.21
Post-Graduation 73 44.79

28.3.2 Measures

For this research, the author adapted a questionnaire which has developed by Diener
et al. [7] consisted of 5 questions on a 7 point likert scale which further talks about
the overall satisfaction with life of an individual.

28.3.3 Objectives of the Study

The main objective of the research is to analyze the wellbeing of the students through
implementation of the chat bots (Counseling App) in the college and university.
The objectives were further divided into three dimensions:
1. To analyze that student’s life satisfaction with chatbot.
2. To analyze that student’s Positive Affect with chatbot.
3. To analyze that student’s Negative Affect with chatbot.

28.3.4 Hypothesis of the Study:

The main hypothesis of the study is that students well being with the shared
information/thoughts with chatbots.
The hypotheses were further divided into three dimensions:
1. Students level of satisfaction with life with chatbot.
2. Students level of positive affect with chatbot.
3. Students level of negative affect with chatbot.
392 S. Agarwal and N. T. D. Linh

28.4 Result and Discussion

The findings of the data were represented in the form of bar graph for easily under-
standing to the large number of audiences. The first finding of the result was shown
in Fig. 28.1. A majority of 33.1% students are slightly agree that, 27% agree that
their life is close to ideal life and 6.1% is strongly agree that their life is close to ideal
life. Out of 163 students, there are 108 people who are agree that their life is close to
ideal life. On the other hand, 24 students are disagree that their life is close to ideal
life and 31 students are neutral about their comparison of life with ideal life.
The second findings about “the conditions of my life are excellent” which shows
that a majority of 51.4% students are slightly agree that condition of their life are
excellent, 14.1% agree that their life is condition of their life are excellent and 1.8%
is strongly agree that their life is condition of their life are excellent. Out of 163
students, there are 79 people who are agree that condition of their life are excellent.
On the other hand, 48 students are disagree that their condition of their life are
excellent and 36 students are neutral about their condition of life are excellent which
is shown in Fig. 28.2.
The third findings about ‘I am satisfied with my life’ are shown in Fig. 28.3.
The findings about “I am satisfied with my life” which shows that a majority of
33.1% students are slightly agree that satisfied with the life, 20.9% agree that they
are satisfied with life and 9.2% is strongly agree that they are satisfied with life. Out

Fig. 28.1 In most ways my life is close to my ideal

Fig. 28.2 The conditions of my life are excellent

28 A Study of Student’s Subjective Well-Being Through Chatbot … 393

Fig. 28.3 I am satisfied with my life

of 163 students, there are 103 people who are agree that they are satisfied with life.
On the other hand, 36 students are disagree that they are satisfied with life and 36
students are neutral about they are satisfied with life or not.
The fourth findings shows that a majority of 28.2% students feel that, so far I have
gotten the important things they want in life which is represented in Fig. 28.4. The
findings about “so far I have gotten the important things they want in life” which
shows that a majority of 28.2% students are slightly agree that so far I have gotten
the important things they want in life, 23.9% agree that so far I have gotten the
important things they want in life and 9.2% is strongly agree that so far I have gotten
the important things they want in life. Out of 163 students, there are 97 people who
are agree that so far I have gotten the important things they want in life. On the other
hand, 36 students are disagree that so far I have gotten the important things they want
in life and 38 students are neutral about so far I have gotten the important things they
want in life.
The fifth findings of the result depicted that a majority of 48% students would
change almost nothing in their life which is clearly represented in Fig. 28.5. Out of
163 students, there are 48 people who would change almost nothing in their life. On
the other hand, 36 students are disagree that would change almost nothing in their
life and 90 students are neutral about would change almost nothing in their life.
The sixth findings of the results show that students feel active after conversing
with chatbots which is represented in Fig. 28.6. A majority 66.6% student feels active

Fig. 28.4 So far I have gotten the important things I want in life
394 S. Agarwal and N. T. D. Linh

Fig. 28.5 If I could live my life over, I would change almost nothing

Fig. 28.6 Positive affect: Active

after conversing with chatbots and 9.2% students disagree that they feel active after
conversing with chatbots. 24.1% feel neutral after conversing with the chatbots.
The seventh findings of the results show that students feel active after conversing
with chatbots which is represented in Fig. 28.7. A majority 72.3% student feels alert
after conversing with chatbots and 6.7% students disagree that they feel alert after
conversing with chatbots. 20.9% feel neutral after conversing with the chatbots.
The eighth findings of the results show that students feel attentive after conversing
with chatbots which is represented in Fig. 28.8. A majority 67.5% student feels
attentive after conversing with chatbots and 7.4% students disagree that they feel
attentive after conversing with chatbots. 25.2% feel neutral after conversing with the
chatbots.

Fig. 28.7 Positive affect: Alert

28 A Study of Student’s Subjective Well-Being Through Chatbot … 395

Fig. 28.8 Positive affect: Attentive

The ninth findings of the results show that students feel determined after
conversing with chatbots which is represented in Fig. 28.9. A majority 74.7% student
feels determined after conversing with chatbots and 8% students disagree that they
feel determined after conversing with chatbots. 17.3% feel neutral after conversing
with the chatbots.
The tenth findings of the results show that students feel enthusiastic after
conversing with chatbots which is represented in Fig. 28.10. A majority 76.5% student
feels enthusiastic after conversing with chatbots and 4.3% students disagree that they
feel enthusiastic after conversing with chatbots. 19.1% feel neutral after conversing
with the chatbots.

Fig. 28.9 Positive Affect: Determined

Fig. 28.10 Positive affect: Enthusiastic

396 S. Agarwal and N. T. D. Linh

Fig. 28.11 Positive affect: Excited

The eleventh findings of the results show that students feel excited after conversing
with chatbots which is represented in Fig. 28.11. A majority 68.1% student feels
excited after conversing with chatbots and 9.2% students disagree that they feel
excited after conversing with chatbots. 22.7% feel neutral after conversing with the
chatbots.
The twelfth findings of the results show that students feel inspired after conversing
with chatbots which is represented in Fig. 28.12. A majority 75.5% student feels
inspired after conversing with chatbots and 10.4% students disagree that they feel
inspired after conversing with chatbots. 14.1% feel neutral after conversing with the
chatbots.
The thirteen findings of the results show that students feel interested after
conversing with chatbots which is represented in Fig. 28.13. A majority 83.4% student
feels interested after conversing with chatbots and 4.3% students disagree that they
feel interested after conversing with chatbots. 12.3% feel neutral after conversing
with the chatbots.
The fourteen findings of the results show that students feel proud after conversing
with chatbots which is represented in Fig. 28.14. A majority 55.3% student feels
proud after conversing with chatbots and 15.4% students disagree that they feel
proud after conversing with chatbots. 29.4% feel neutral after conversing with the
chatbots.
The fifteen findings of the results show that students feel strong after conversing
with chatbots which is represented in Fig. 28.15. A majority 66.9% student feels

Fig. 28.12 Positive affect: Inspired

28 A Study of Student’s Subjective Well-Being Through Chatbot … 397

Fig. 28.13 Positive affect: Interested

Fig. 28.14 Positive affect: Proud

Fig. 28.15 Positive affect: Strong

strong after conversing with chatbots and 11.6% students disagree that they feel
strong after conversing with chatbots. 21.5% feel neutral after conversing with the
chatbots.
The sixteen findings of the results show that students feel strong after conversing
with chatbots which is represented in Fig. 28.16. About 29.5% student feels afraid
after conversing with chatbots and 41.7% students disagree that they feel afraid after
conversing with chatbots. 28.8% feel neutral after conversing with the chatbots.
The seventeen findings of the results show that students feel ashamed after
conversing with chatbots which is represented in Fig. 28.17. About 10.5% student
feels ashamed after conversing with chatbots and 72.4% students disagree that they
398 S. Agarwal and N. T. D. Linh

Fig. 28.16 Negative affect: Afraid

Fig. 28.17 Negative affect: Ashamed

feel ashamed after conversing with chatbots. % 17.2 feel neutral after conversing
with the chatbots.
The eighteen findings of the results show that students feel disstressed after
conversing with chatbots which is represented in Fig. 28.18. About 27.6% student
feels disstressed after conversing with chatbots and 41.7% students disagree that they
feel disstressed after conversing with chatbots. 41.7% feel neutral after conversing
with the chatbots.
The nineteen findings of the results show that students feel guilty after conversing
with chatbots which is represented in Fig. 28.19. About 20.3% student feels guilty

Fig. 28.18 Negative affect: Disstressed

28 A Study of Student’s Subjective Well-Being Through Chatbot … 399

Fig. 28.19 Negative affect: Guilty

after conversing with chatbots and 61.9% students disagree that they feel afraid after
conversing with chatbots. 17.8% feel neutral after conversing with the chatbots.
The twentieth findings of the results show that students feel hostile after conversing
with chatbots which is represented in Fig. 28.20. About 13.5% student feels hostile
after conversing with chatbots and 65.6% students disagree that they feel hostile after
conversing with chatbots. 20.9% feel neutral after conversing with the chatbots.
The twenty one findings of the results show that students feel irritable after
conversing with chatbots which is represented in Fig. 28.21. About 19.9% student
feels irritable after conversing with chatbots and 52.1% students disagree that they

Fig. 28.20 Negative affect: Hostile

Fig. 28.21 Negative affect: Irritable

400 S. Agarwal and N. T. D. Linh

Fig. 28.22 Negative affect: Jittery

Fig. 28.23 Negative affect: Nervous

feel irritable after conversing with chatbots. 28% feel neutral after conversing with
the chatbots.
The twenty two findings of the results show that students feel jittery after
conversing with chatbots which is represented in Fig. 28.22. About 17.2% student
feels jittery after conversing with chatbots and 49.1% students disagree that they feel
jittery after conversing with chatbots. 33.7% feel neutral after conversing with the
chatbots.
The twenty three findings of the results show that students feel nervous after
conversing with chatbots which is represented in Fig. 28.23. About 30.6% student
feels nervous after conversing with chatbots and 38.7% students disagree that they
feel nervous after conversing with chatbots. 30.7% feel neutral after conversing with
the chatbots.
The twenty four findings of the results show that students feel scared after
conversing with chatbots which is represented in Fig. 28.24. About 22.7% student
feels scared after conversing with chatbots and 54.6% students disagree that they
feel scared after conversing with chatbots. 22.7% feel neutral after conversing with
the chatbots.
The twenty five findings of the results show that students feel upset after
conversing with chatbots which is represented in Fig. 28.25. About 18.4% student
feels upset after conversing with chatbots and 58.9% students disagree that they
feel upset after conversing with chatbots. 37% feel neutral after conversing with the
chatbots.
28 A Study of Student’s Subjective Well-Being Through Chatbot … 401

Fig. 28.24 Negative affect: Scared

Fig. 28.25 Negative affect: Upset

28.5 Conclusion

The main aim of this research was to find out the level of well-being among students
about chatbots. Numerous studies have suggested that guidance and counseling is
pivotal in the university and colleges for the overall wellbeing of the communities
of students. On the contrary, there are studies which show that students feel hesitant
while conversing with the counselor. The major findings of the study are that students’
subjective well-being has been measured into three dimensions.
For the study, we analyzed subjective well-being first dimensions, which is Life
satisfaction which represented that overall students feel satisfied after conversing
with the chatbot. The second dimension is Positive affect which is also increasing
after conversing with the chatbots. The third dimension is Negative affect which is
decreasing among the students after conversing with the chatbots.
Thus, it is recommended to the college and university especially for the developing
country like India to implement chatbots in their premises for the wellbeing of their
students.Modern Chatbots are packed with various features like 24 h’ availability,
attitude sensing, conversational maturity, emotional support and SOS alarms which
observe and track the behavior of subjects and help them in the best possible ways.
With time Chatbot Technology can surely be a milestone in field of Guidance and
Counselling.
402 S. Agarwal and N. T. D. Linh

28.6 Limitations

The present study has several limitations which deserve to be mentioned. First, for
the purpose of this research, a small sample size of 1637 respondents has been used.
Significant increase in the sample size can provide an enhanced outcome. Second,
in this research, data was collected from people residing in Delhi/NCR region. A
larger demographic area under consideration can provide more variation and diversity
in data. Lastly, data was collected only through online survey; thus a Quantitative
Approach was used for collection of data. A more detailed analysis can be done with
the help of Qualitative Approach of data collection.

References

1. Armas Sánchez, P.D.: Perspectives from University Graduates facing AI and Automation in
Ireland: how do Irish Higher Education’s graduates from Maynooth University perceive AI is
going to impact them? Doctoral dissertation, National University of Ireland Maynooth (2019)
2. Bhakta, R., Savin-Baden, M., Tombs, G.: Sharing secrets with robots? In: EdMedia+ Inno-
vate Learning, pp. 2295–2301. Association for the Advancement of Computing in Education
(AACE) (2014).
3. Blashki, G., Lock, S.: Artificial intelligence and mental health (2019)
4. Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., O’Neill, S., Armour, C.,
McTear, M.: Assessing the usability of a chatbot for mental health care. In: International
Conference on Internet Science, pp. 121–132. Springer, Cham (2018).
5. Cameron, G., Cameron, D., Megaw, G., Bond, R., Mulvenna, M., O’Neill, S., Armour,
C., McTear, M.: Towards a chatbot for digital counselling. In: Proceedings of the 31st
British Computer Society Human Computer Interaction Conference, p. 24. BCS Learning
& Development Ltd. (2017)
6. Coughlan, J., Macredie, R., Patel, N.: Moving face-to-face communication to Web-based
systems (2007).
7. Diener, E.D., Emmons, R.A., Larsen, R.J., Griffin, S.: The satisfaction with life scale. J. Pers.
Assess. 49(1), 71–75 (1985)
8. Eliamani, M.P., Richard, M.L., Peter, B.: Access to guidance and counseling services and its
influence on Students”school life and career choice”. Afr. J. Guid. Counsel. 1(1), 007–015
(2014)
9. Narote, A., Dolani, A., Pandey, A., Carvalho, P., Chaudhari, R.: Visa Chat-Bot (2019)
10. Prasad, N.A., Prthaban, L., Rana, M.E.: Use of recommendation engine and chatbot for
depression consultancy platforms. J. Crit. Rev. 7(3), 2020 (2019)
11. Prasetyaningrum, T., Gregorius, R.M.: Design and implementation of mobile leadership
with interactive multimedia approach. In: International Conference on Multimedia, Computer
Graphics, and Broadcasting, pp. 217–226. Springer, Berlin, Heidelberg (2011)
12. Srivastava, S.K.: Artificial intelligence: way forward for India. J. Inf. Syst. Technol. Manage.
(JISTEM)15 (2018).
13. Sturk, H., Crowther, R., Kavanagh, D.J.: Head to health: Practitioner perceptions of the new
digital mental health gateway. Aust. J. Rural Health 27(5), 448–453 (2019)
14. Suyoto, P.: Development of mobile application social guidance and counseling for junior high
school. Int. J. Comput. Electr. Autom. Control Inf. Eng. 7, 1398–1404 (2013)
15. UGC: University-UGC. 16/5/2020. https://www.ugc.ac.in/oldpdf/Consolidated%20list%
20of%20All%20Universities.pdf (2020)
28 A Study of Student’s Subjective Well-Being Through Chatbot … 403

16. Wilhelm, S., Weingarden, H., Ladis, I., Braddick, V., Shin, J., Jacobson, N.C.: Cognitive-
behavioral therapy in the digital age: presidential address. Behav. Ther. 51(1), 1–14 (2020)
17. Yorita, A., Egerton, S., Oakman, J., Chan, C., Kubota, N.: A robot assisted stress management
framework: using conversation to measure occupational stress. In: 2018 IEEE International
Conference on Systems, Man, and Cybernetics (SMC), pp. 3761–3767. IEEE (2018)
18. Young, K.S.: An empirical examination of client attitudes towards online counseling. Cyber
Psychol. Behav. 8(2), 172–177 (2005)