Internet of Things in Modern

Computing

The text focuses on the theory, design, and implementation of the Internet of
Things (IoT), in a modern communication system. It will be useful to senior
undergraduate, graduate students, and researchers in diverse fields domains,
including electrical engineering, electronics and communications engineer-
ing, computer engineering, and information technology.

Features:

• Presents all the necessary information on the Internet of Things in

modern computing
• Examines antenna integration challenges and constraints in the
Internet of Things devices
• Discusses advanced Internet of Things networks and advanced con-
trollers required for modern architecture
• Explores security and privacy challenges for the Internet of Things-
based healthcare system
• Covers implementation of Internet of Things security protocols such
as MQTT, Advanced Message Queuing Protocol, XMPP, and DSS

The text addresses the issues and challenges in implementing communica-

tion and security protocols for the IoT in modern computing. It further
highlights the applications of IoT in diverse areas, including remote health
monitoring, remote monitoring of vehicle data and environmental charac-
teristics, Industry 4.0, 5G communications, and Next-gen IoT networks.
The text presents case studies on IoT in modern digital computing. It will
serve as an ideal reference text for senior undergraduate, graduate students,
and academic researchers in diverse fields domains, including electrical engi-
neering, electronics and communications engineering, computer engineer-
ing, and information technology.
Smart Technologies for Engineers and Scientists
Series Editor: Mangey Ram

Applications of Mathematical Modeling, Machine Learning, and Intelligent

Computing for Industrial Development
Madhu Jain, Dinesh K. Sharma, Rakhee Kulshrestha, and H.S. Hota
 Internet of Things
in Modern Computing
Theory and Applications

Edited by
Vinay Chowdary
Abhinav Sharma
Naveen Kumar
Vivek Kaundal
First edition published 2023
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742
and by CRC Press
4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
CRC Press is an imprint of Taylor & Francis Group, LLC

© 2024 selection and editorial matter, Vinay Chowdary, Abhinav Sharma, Naveen Kumar and
Vivek Kaundal; individual chapters, the contributors
Reasonable efforts have been made to publish reliable data and information, but the author
and publisher cannot assume responsibility for the validity of all materials or the consequences
of their use. The authors and publishers have attempted to trace the copyright holders of all
material reproduced in this publication and apologize to copyright holders if permission to
publish in this form has not been obtained. If any copyright material has not been acknowl-
edged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, repro-
duced, transmitted, or utilized in any form by any electronic, mechanical, or other means, now
known or hereafter invented, including photocopying, microfilming, and recording, or in any
information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.­
copyright.com or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive,
Danvers, MA 01923, 978-750-8400. For works that are not available on CCC please contact
[email protected]
Trademark notice: Product or corporate names may be trademarks or registered trademarks
and are used only for identification and explanation without intent to infringe.

ISBN: 978-1-032-39272-1 (hbk)

ISBN: 978-1-032-52582-2 (pbk)
ISBN: 978-1-003-40730-0 (ebk)

DOI: 10.1201/9781003407300

Typeset in Sabon
by SPi Technologies India Pvt Ltd (Straive)
Contents

Preface vii
Acknowledgments xi
Editors xiii
Contributors xv

1 IoT architecture and design 1

NEERAJ GUPTA

2 Application of artificial intelligence (AI) and the Internet

of Things (IoT) in process industries toward Industry 4.0 13
RASHI VERMA, RAJU YEROLLA, SUHAILAM P AND CHANDRA SHEKAR BESTA

3 A review on edge computing: Working, comparisons,

benefits, vision, instances and illustrations along
with challenges 37
PARUL DUBEY, PUSHKAR DUBEY AND KAILASH KUMAR SAHU

4 Industrial Internet of Things: IoT and Industry 4.0 53

ASMITA SINGH BISEN AND HIMANSHU PAYAL

5 Denial of Service Attacks in the Internet of Things 67

ADITI PAUL AND S SINHA

6 Extending the Unified Theory of Acceptance and

use of technology model to understand the trainees’
acceptance and usage of Internet of Things (IoT)
by skill development course 91
ANANTA NARAYANA, P. S. BIRLA AND R. K. SHASTRI

v
vi Contents

7 Intelligent approaches for disease detection

and prevention 113
SAUMYA YADAV, DEEPAK CHANDRA JOSHI, ABHISHEK JOSHI,
SANJAY MATHUR AND MANOJ BHATT

8 Interoperability in IoT-Driven Smart Buildings:

Employing Rule-based decision support systems 133
MOHAN KRISHNA S., THINAGARAN PERUMAL, SUMUKH SURYA
AND CHANDRASHEKAR

9 IoT-based parking system using Web-App 143

SAMYAK JAIN, ANKIT GUPTA, ABHINAV SHARMA AND VINAY CHOWDARY

10 A next-gen IoT-based semi-automatic mobile manipulator 155

MUKUL KUMAR GUPTA AND SHIVAL DUBEY

11 Pest identification and classification using IoT

enable technique 169
ATUL B. KATHOLE, SONALI D. PATIL, KAPIL N. VHATKAR,
DINESH CHAUDHARI AND AVINASH P. JADHAV

12 Framework for leveraging diagnostic and vehicle data

with emphasis on automotive cybersecurity 181
THRILOCHAN SHARMA, K. S. ASHWINI, PRITHVI SEKHAR PAGALA
AND KNS ACHARYA

13 An array of Fibonacci series-based wide-band printed

antennas for IoT/5G applications 195
KALYAN SUNDAR KOLA AND ANIRBAN CHATTERJEE

14 Deep learning IoT platform for dental disease detection 213

SINDHU P. MENON, PRAMOD KUMAR NAIK AND BASKAR VENUGOPALAN

Index 239
Preface

The Internet of Things (IoT) has emerged as the major contributor behind
the development of the technological infrastructure which connects physical
objects to the internet. It is evident that an examination of the role of the
IoT in the modern era is urgently needed. Therefore, the intention of this
publication is to have all the necessary information regarding the IoT in
modern computing contained within a single cover.
The organization of the book is as follows:
Chapter 1 lays the foundation stone in terms of the progress made in the
Internet of Things (IoT) field and advance the challenges in the architecture
design of IoT. The chapter also discusses both classical and emerging IoT
models.
Chapter 2 provides a bird’s-eye view of Industry 4.0 and the role of
Artificial Intelligence (AI) and the Internet of Things (IoT) in achieving this
new framework. It provides a quick overview of the history of the industrial
revolution by highlighting some significant historical events. Industry 4.0 is
a new model that strives to build an open and smart processing platform.
Chapter 3 discusses the role of edge computing and its ability to do
sophisticated calculations at the edge location rather than doing the calcula-
tions at the cloud location. In this chapter recent study, developments and
benefits in edge computing are also highlighted. There are a few illustrations
and photographs in this chapter.
Chapter 4 discusses the role of Industrial IoT (IIoT) within Industry 4.0.
IIoT, which is an integral part of IoT, is one of the most widely used terms
in the development of modern computing. This chapter offers some detailed
insights with regard to IIoT, along with an outline of its key technologies.
The chapter also focuses on the open challenges and application domain of
IoT in Industry 4.0.
Chapter 5 provides a general framework for the detection of Denial of
Service (DoS) attacks in the field of IoT. It analyses the research gaps in DoS
detection techniques and then offers a two-stage framework for DoS detec-
tion technique based on the Fuzzy Rule Manager and Neural Networks.
Chapter 6 examines the parameters that may either accelerate or hinder
the adaption of IoT by means of a skill development course. The indicated

vii
viii Preface

study extended the Unified Theory of Acceptance and Use of Technology

model through an integration of added factors, i.e., computer self-efficacy
along with relative advantages.
Chapter 7 highlights the key insights of different methods of implement-
ing IoT, AI and blockchain in the healthcare sector and also their medical
applications in disease diagnosis and treatment, sensor networks, and some
future aspects regarding data privacy, security, and safety.
Chapter 8 presents an overview of the coordination issues faced in IoT-
driven Building Energy Management Systems (BEMS) subsystems owing to
their heterogeneous nature and how interoperability tries to address the
same. It also emphasizes the need for a decision support model in the mea-
surement and management of energy usage in buildings.
Chapter 9 focuses on the role of the smart city of the future in pandemic
control. Wireless sensor networks and the Internet of Things are the back-
bone of a smart city. The chapter authors have simulated the case of a smart
city using the CupCarbon™ tool. The authors traced the movement of infec-
tious nodes and the spread has been analysed.
Chapter 10 highlights the applications of the IoT in the field of robotics.
The authors propose a remotely operated Mobile Manipulator that can
detect suspicious or explosive devices both during terrorist attacks and on
active bomb sites. The robot designed is operated wirelessly from a remote
location in order to make the bomb disposal safer.
Chapter 11 addresses a need for the IoT in agricultural areas in order to
reduce the use of chemical fertilizers and crop protection agents. In this
study the IoT devices are designated to be used for data collection and to
identify the pests in the photos. The chapter intends to create a model for
pest identification and classification.
Chapter 12 proposes a framework to utilize the On-Board Diagnostic-II
(OBD II) module to collect vehicle data and a smartphone for multiple
inbuilt sensors and connectivity employing the IoT cloud. Different data
cleaning, characterization, machine learning and analytics models are used
to present and framework and the results.
Chapter 13 proposes a printed monopole radiator-based two-element
array for the Internet of Things (IoT)-based applications and commercial
5G (3.3–3.8 GHz) connectivity. The design proposed in this chapter is suit-
able commercial 5G communications and IoT-based applications.
Chapter 14 proposes an automated system for the identification of dental
diseases using deep learning and the IoT. The technique proposed in this
study is extremely efficient and has a very low probability of error. The
experimental results of the proposed technique are also provided.
This research book is intended to cater to the requirements of a wide
domain of researchers working in the domain of AI, robotics and IoT/WSN.
This book covers a basic understanding of IoT and its architecture design
along with the application of advanced technologies such as AI, Industrial
 Preface ix

IoT, deep learning and 5G. This book serves not only academic and techni-
cal readers but also those based in industry, allowing them to gain some
insights into the latest developments in the technological framework and its
impact. Practical case studies from leading industries such as Bosch, Larsen
& Toubro, etc are also a part of this book. This book can also serve as a
valuable reference for academics, mechanical, mechatronics, computer sci-
ence, information technology and industrial engineers, and environmental
scientists as well as researchers in related subjects.
Dehradun, India

Editors:
Vinay Chowdary
Abhinav Sharma
Naveen Kumar
Vivek Kaundal
Acknowledgments

The editor acknowledges CRC Press - Taylor & Francis Group for this
opportunity and professional support. My special thanks to Gagandeep
Singh, Publisher Engineering, CRC Press, Gauravjeet Singh Reen, Senior
Editor - Engineering and Isha Ahuja, Editorial Assistant - Engineering for
the excellent support provided to us during the completion of this book.
Thanks to the chapter authors and reviewers for their availability for this
work.

Vinay Chowdary, International School of Engineering, India

Abhinav Sharma, University of Petroleum and Energy Studies, India
Naveen Kumar, Christ University, India
Vivek Kaundal, Larsen & Toubro Technology Services, India

xi
Editors

Prof. Dr. Vinay Chowdary received his B.Tech. M.Tech. degrees from MIST,
JNTUH, Hyderabad, India, in 2005 and 2012, respectively, and a Ph.D.
degree from UPES, Dehradun, India, in 2020. He is currently working as an
Associate Professor at the uGDX Institute of Technology, India. He has more
than 14 years of teaching and research experience and extensive experience
in curriculum design. His research interests include embedded systems, the
Internet of things, wireless sensor networks, and advanced microcontrollers.

Prof. Dr. Abhinav Sharma received his B.Tech. from H. N. B. Garhwal

University, Srinagar, India, in 2009 and his M.Tech. from Govind Ballabh
Pant University of Agriculture and Technology, Pantnagar, India, in 2011.
He completed his Ph.D. from Govind Ballabh Pant University of Agriculture
and Technology, Pantnagar, India, in 2016. He is currently working as
an Assistant Professor (Senior Scale) in the Department of Electrical and
Electronics Engineering in University of Petroleum and Energy Studies,
Dehradun. His fields of interest includes adaptive array signal processing,
artificial intelligence, machine learning, optimization techniques and smart
antennas.

Dr. Naveen Kumar received a B.Tech. from SVIET, Punjab, India in 2009
and an M.E. from NITTTR, Chandigarh, India in 2013. He is currently
working as a Research Engineer, FCS Railenium, France where the design
and development of antennas for railways is being carried out under the
European project ‘Shift2Rail’. He completed his Ph.D. in July 2021 from
Thapar Institute of Engineering and Technology (Deemed to be University),
Punjab, India. He has more than six years of experience in research, aca-
demics, and corporate domains. He is the inventor of a novel antenna for
which a French patent was filed in June 2020. He has published almost 50
papers in various reputable national/international journals and conferences.
He is an active member of IEEE, EuMA, and IAENG. His research interests
are antenna design, antennas for IoT, MIMO antennas, metamaterials, and
characteristic mode analysis.

xiii
xiv Editors

Dr. Vivek Kaundal received his B.Tech in Electronics & Communication

Engineering, M.Tech. in Digital communication, and a Ph.D. in Wireless
Sensor Networks (Electronics Engineering). He has more than 13 years of
wide-ranging experience in the corporate sector and academia. He is cur-
rently working for L&T Technology Services as a Specialist in LTTS, Mysore.
In this position he is responsible for designing and developing a competency
framework in alignment with business needs. His field of interests are auto-
motive and embedded systems. He has a rich experience in AUTOSAR, C4K
tool, and Mentor Graphics (CVI) tool for the configuration of AUTOSAR
modules like CAN, Diagnostics, OS etc. and in integration. He has also
worked in the area of application software components, root software com-
position file etc.
Contributors

KNS Acharya Dinesh Chaudhari

L&T Technology Services JDIET
Limited Yavatmal, Maharashtra, India
Bangalore, India
Vinay Chowdary
K. S. Ashwini uGDX Institute of Technology
Coventry University Hyderabad, India
United Kingdom
Parul Dubey
Chandra Shekar Besta
G H Raisoni College of
National Institute of Technology
Engineering
Calicut, Kerala, India
Nagpur, India
Manoj Bhatt
Pushkar Dubey
College of Technology,
Pandit Sundar Lal Sharma (Open)
G.B.P.U.A&T.
University
Pantnagar, Uttarakhand, India
Chhattisgarh, India
P. S. Birla
FOE, IGNTU Shival Dubey
Amarkantak, Madhya Pradesh, Institute of Design, Robotics &
India Optimization, University of
Leeds
Asmita Singh Bisen United Kingdom
Sharda University
Greater Noida, Uttar Pradesh, India Ankit Gupta
University of Petroleum & Energy
Chandrashekar Studies
L & T Technological Services Dehradun, Uttarakhand, India
Bangalore, India
Mukul Kumar Gupta
Anirban Chatterjee University of Petroleum and Energy
National Institute of Technology Studies
Farmagudi, Ponda, Goa Dehradun, Uttarakhand, India

xv
xvi Contributors

Neeraj Gupta Ananta Narayana

K.R. Mangalam University Panipat DHSS, MNNIT
Institute of Engineering and Allahabad, Uttar Pradesh, India
Technology
Panipat, Haryana, India Prithvi Sekhar Pagala
L&T Technology Services Limited
Avinash P. Jadhav Bangalore, India
DRGIT&R Amravati
Sonali D. Patil
Maharashtra, India
Pimpri Chinchwad College of
Engineering
Samyak Jain
Pune, Maharashtra, India
University of Petroleum & Energy
Studies Aditi Paul
Dehradun, Uttarakhand, India Banasthali Vidyapith
Rajasthan, India
Deepak Chandra Joshi
Shiv Nadar University Himanshu Payal
Dadri, Ghaziabad, India Sharda University
Greater Noida, Uttar Pradesh, India
Abhishek Joshi
Sungkyunkwan University
Thinagaran Perumal
Seoul, South Korea
University of Putra
Atul B. Kathole Selangor, Malaysia
D. Y. Patil Institute of Technology
Suhailam P
Pune, Maharashtra, India
National Institute of Technology
Calicut, Kerala, India
Kalyan Sundar Kola
National Institute of Technology
Kailash Kumar Sahu
Farmagudi, Ponda, Goa
Pandit Sundar Lal Sharma (Open)
Sanjay Mathur University
College of Technology, Chhattisgarh, India
G.B.P.U.A&T
Abhinav Sharma
Pantnagar, Uttarakhand, India
University of Petroleum & Energy
Sindhu P. Menon Studies
Professor, School of Computing and Dehradun, Uttarakhand, India
Information Technology, REVA
University Thrilochan Sharma
Bangalore, Karnataka, India L. & T. Technology Services
Limited
Pramod Kumar Naik Bangalore, India
Associate Professor, Department of
Computer Science and Engineering, S Sinha
Dayananda Sagar University CHRIST (Deemed to be University)
Bangalore, Karnataka, India Bangalore, Karnataka, India
 Contributors xvii

R. K. Shastri Rashi Verma

DHSS, MNNIT National Institute of Technology
Allahabad, Uttar Pradesh, India Calicut, Kerala, India

Sumukh Surya Kapil N. Vhatkar

Bosch Global Software Pimpri Chinchwad College of
Technologies Engineering
Bangalore, India Pune, Maharashtra, India
Mohan Krishna S.
Saumya Yadav
Alliance University
Indraprastha Institute of
Bangalore, India
Information Technology (IIIT)
Baskar Venugopalan Delhi, India
Professor, Department of Computer
Science and Technology, Raju Yerolla
Dayananda Sagar University National Institute of Technology
Bangalore, Karnataka, India Calicut, Kerala, India
Chapter 1

IoT architecture and design

Neeraj Gupta
Panipat Institute of Engineering and Technology, Panipat, India

CONTENTS

1.1 Introduction ...................................................................................... 1

1.1.1 Challenges: IoT architecture drivers ....................................... 2
1.1.2 Classical IoT architecture models ........................................... 3
1.1.2.1 oneM2M architecture ............................................... 4
1.1.2.2 IoT world forum reference model ............................. 5
1.1.3 Emerging IoT architecture models .......................................... 7
1.1.4 Conclusion ........................................................................... 10
References ���������������������������������������������������������������������������������������������� 10

1.1 INTRODUCTION

Industry 4.0 refers to the automation of the decision-making process to opti-

mize various usages of resources by enabling interactions between the physi-
cal and the digital world. The Internet of Things (IoT) is one of the critical
technologies used to facilitate the above vision and the purpose of Industry
4.0. IoT involves the participation of many technologies, including sensors,
communication protocols, edge computing, fog computing, data analytics,
and cloud computing. The design of the architecture involves specifications
of various functional modules and the interaction between these modules to
create a system. According to the Gartner glossary, “The Internet of Things
(IoT) is the network of physical objects that contain embedded technology
to communicate and sense or interact with their internal states or the exter-
nal environment” [1]. The physical objects are the sensors employed to sense
and detect the changes in the physical environment for specific applications.
The recorded parameters are then communicated and processed locally or
on cloud-based systems in order to make well-informed decisions. IoT is
an amalgamation of various technologies, including sensor networks, com-
puter networks, cloud computing, and data analytics. These technologies
form the driving force that contributes to the functionality of different IoT

DOI: 10.1201/9781003407300-1 1
2 Neeraj Gupta

layers. Regardless of disruption as a result of chip shortages, and global

events, it is estimated that there will be 27 billion IoT devices by the end of
the year 2025 [2].
The current chapter is divided into three main sections. The first section
discusses the challenges involved in IoT architecture. The second section
gives insight into two prevalent classical architecture models of IoT. The
third section outlines the current state of the art in order to enhance the
capabilities of the system.

1.1.1 Challenges: IoT architecture drivers

The deployment of trillions of sensors to capture data from billions of sys-
tems to empower millions of applications have enabled business sectors to
make informed and intelligent decisions. Global vendors are now offering
customized solutions to facilitate and automize processes across various pub-
lic and private sectors. However, multiple challenges need to be addressed to
ensure that the solutions offered remain meaningful:

1. Scale: The explosive growth of smart things has burdened the network
with enormous amounts of data. It is estimated that by 2025 total
data volume generated by IoT devices will reach 79.4 zettabytes. The
rollout of 5G is expected to further accelerate the growth of connected
cars, smart home devices, and industrial equipment. According to fore-
casts, there will be 30.5 billion IoT devices across the world by 2025.
Each of these devices will require the allocation of a unique address in
order to communicate. It is essential to strategies IPv6 addresses allot-
ment to meet the growing demand for connected devices.
2. Security: As per the report by [3], the following statistics underline the
importance of cybersecurity in the Internet of Things.
a. Cybercrime is estimated to generate about $10.5 trillion 2025.
b. 98% of IoT devices are ill-equipped to deal with cyber-attacks.
c. On average, 97% of encrypted files affected by so-called ransom-
ware attacks are recovered after companies pay the criminals.
The components in IoT need to be connected in order to share
the information. In this process the role of data security is of par-
amount importance. The basic building blocks of confidentiality,
integrity, and availability need to be ensured at various stages,
including nodes, gateways, fog, and cloud computing.
3. Privacy: With users’ increasing use of smart devices, the generated
data is more individual-specific. This may be wide-ranging in nature:
financial, health-related, location-based, and/or shopping patterns, to
list just a few. The service provider stores and processes this informa-
tion as one aspect of their offered services. The service provider must
clearly state their policies, stating the status of ownership and acces-
sibility of the information with other parties.
 IoT architecture and design 3

4. Interoperability: The solution offered by the various organizations

in the IoT domain is based on protocols that are either proprietary
solutions or belong to open standards. There are many protocols and
architectures, each with their own sets of advantages and disadvan-
tages. The need to support the legacy devices, along with new devices
with more outstanding capabilities, calls for the provision of interop-
erability solutions. It is imperative to standardize architecture at the
global level in order to ensure interoperability among solution provid-
ers so that the maximum benefits of technology can be reaped.
5. Constrained Nodes and Networks: Wireless Sensor Networks (WSN)
are an essential component of IoT. The sensor and sensor networks
are usually application-specific. The nodes or sensors are resource-
constrained devices with both limited processing capabilities and
limited power. At a different layer of IoT architecture, many efficient
protocols have been proposed and deployed successfully across the
field. Among the most prominent architectures advanced at the MAC
layer at present are IEEE 802.15.4 [4], 6LoWPAN [5], HART, and
Zigbee [6]. In addition, Message Queue Telemetry Transport Protocol,
Constrained Application Protocol, Advanced Message Queueing
Protocol, and Extensible Messaging and Presence Protocol are among
the application-based protocols.
6. Data Processing and Analysis: The massive amount of sensed and
communicated data needs to be processed and analyzed in order to
produce meaningful information. Information processing can occur
at various points of the process, including sensor nodes, edge/fog ele-
ments, and cloud servers. The criticality of the application and volume
of data define the stage of the networking element where process-
ing needs to be executed. Data parameters, including velocity, vari-
ety, variability, and veracity, need to be considered when processing
the data.
7. Heterogeneity: The concept of the IoT promotes the idea of comput-
ing anywhere anytime such that interconnected systems support device
independence, location independence, and data format independence.
To realize the principle of autonomic computing, it is essential to keep
both IP-based systems and non-IP-based systems. For example, vari-
ous devices support IPv4 instead of IPv6. There is a need to develop
and design middleware solutions that can overcome multiple hetero-
geneous issues at multiple levels of architecture design.

1.1.2 Classical IoT architecture models

There are various elements and technologies involved in realizing the
essence of IoT. Data collection, data transportation, data analysis, and
actuation based on analysis form the basic building blocks for IoT archi-
tecture. There are primarily two classical models that are mentioned in the
4 Neeraj Gupta

Figure 1.1 oneM2M architecture stack [8].

literature: oneM2M architecture (as shown in Figure 1.1) and IoT World
Forum Reference Model.

1.1.2.1 oneM2M architecture
This has developed from a project [7] founded by eight prominent ICT
standard development organizations in 2012 to address an issue related to
the interoperability and scalability issues faced by IoT technologies. The
oneM2M standard is open access in nature. The architecture is a middle-
ware technology connecting applications and devices via RESTful APIs.
In this system the architecture stack consists primarily of three layers: the
Application Layer, the Service Layer, and the Network Layer:

A. Application Layer: The Application Entity (AE) implements the appli-

cation service logic in the topmost layer. The running instances of the
application(s) should have a unique id (AE-ID). These instances can
be either virtualized or physical. The communication between the
instances is carried out by sending the request to the service layer.
B. Service Layer: The middleware layer contains Common Service
Functions (CSF) residing in the service layer. CSF can be present in
sensors, gateways, edge devices, and back-end cloud servers. There
are 14 CSFs that handle device management, registration, security,
semantic interoperability, and location services. Common Service
Entities (CSE) represent an instance of a set of CSF. Each CSE is iden-
tified by unique CSE-id. The AE invokes the CSE to utilize a particu-
lar service.
C. Network Layer: This layer includes the sensors and communication
protocols that are used to transfer the data. Network Service Entities
provide their service to the CSE. The services being offered by this later
include device location, device triggering, and various sleep modes.
 IoT architecture and design 5

Figure 1.2 Horizontal architecture [8].

The term node has a particular meaning in oneM2M standards. It is

defined in the literature as

Nodes are logical entities identifiable in the oneM2M System. oneM2M

Nodes typically contain CSEs and/or AEs. For the definition of Node
types, oneM2M distinguishes between Nodes in the ‘Field Domain’ –
i.e., the domain in which sensors/actuators/aggregators/gateways are
deployed – and the ‘Infrastructure Domain’ – i.e., the domain in which
servers and applications on larger computers reside [8].

It is important to mention that the oneM2M standard supports both IP-based

and non-IP-based systems through Interworking Proxy Entities (IPE). There
are unique AEs that facilitate communication between two incompatible
systems to communicate with each other. The horizontal architecture in
Figure 1.2 offers various advantages:

1. Reusability by offering common service layer functionality.

2. Interoperability is achieved at various layers using API.
3. The issue of heterogeneity can be addressed seamlessly.

1.1.2.2 IoT world forum reference model

IoT World Forum (IoTWF) is an industry event organized annually by
CISCO, as shown in Figure 1.3. In 2014, the IoTWF Consortium unveiled a
seven-layer architecture to simplify, clarify, identify, standardize, and orga-
nize various elements and functions of IoT systems and applications. The
terminologies defined in each layer of the reference model help standardize
the component’s scope and functionality. The most important feature of the
model is the bidirectional flow of data across various layers. The information
6 Neeraj Gupta

Figure 1.3 IOTWF reference model.

flow can be categorized as either control flow or monitoring flow. In control

flow, the information flows from the highest layer to the bottom layer; the
process is reversed in monitoring flow:

A. First Layer: This layer describes ‘things’ in IoT, which can generate
data, perform conversion from analog to digital and vice versa, and
can be queried and controlled from a remote location. A low level of
processing can be quantified in the devices.
B. Second Layer: This layer is concerned with communication and con-
nectivity protocols among the devices. The existing network protocols
can be used to transfer data
1. Between devices on the same network.
2. Between devices across different networks
3. To the device/element present in the upper layer.
To enable communication through devices that are not IP-compliant,
the model suggests using gateways. The network security and network
analytics could also be implemented at this layer.
C. Third Layer: This layer focuses on data analysis and transformation
by processing the data. The process can either be used to control the
devices based on the type of event generation or be stored at a higher
layer for storage and further processing. Edge computing and fog
computing are the related terminologies that can be referred to for
this layer.
D. Fourth Layer: The fourth layer is concerned with data accumulation
and is accessible to applications on a non-real-time basis. The acces-
sibility is converted from event-based need to query-based processing.
E. The accumulated data are further filtered and stored in database
tables.
 IoT architecture and design 7

F. Fifth Layer: The data abstraction layer handles the variety, variability,
and veracity of data. This layer’s responsibility is to ensure that con-
solidated data stored in the cloud is both reconciled and consistent.
The data generated from various sources can have different formats
and semantics.
G. Sixth Layer: Various software and algorithms are deployed to interpret
the consolidated data based on the need of the business application.
H. Seventh Layer: The IoT involves people and processes. The aim is to
collaborate with different processes and people to ease the decision-
making. Collaboration and processes are the right terms coined for
this layer.

The model recommends that security measures must

1. Secure each device

2. Provide security to process and at each level.
3. Secure communication of data across various layers.

1.1.3 Emerging IoT architecture models

Different applications have their own set of challenges that need custom-
ized solutions. Heterogeneous and voluminous data should be appropriately
processed to ensure that business decisions can be taken responsibly. There
may be a requirement to make fast decisions to drive real-time applica-
tions. Customer-oriented applications require transparency and security
while transactions are carried out. Different technologies cater to different
roles and responsibilities of ‘things’. This section categorizes IoT based on
a different level of reference model, providing appropriate solutions. The
classification is based on key technologies that paved the way for modern
computing and numerous applications.

A. Cloud-based IoT
Cloud computing has revolutionized how computing services, both
hardware and software, are being offered. Virtualization has played a
vital role in facilitating multi-tenant architecture. The storage facility
of cloud service providers accumulates the massive volume of data
generated from multiple sources. The cloud can act as a platform for
sharing typical applications among numerous ‘things’ located at vari-
ous locations. The enormous amount of processing power can be uti-
lized to process for analysis. The disadvantages of the clouds include
high latency, security, energy consumption, and billing.
B. Fog Computing-based IoT
The term fog computing was coined by Cisco Systems to reflect the
extension of the services provided by cloud computing to the edge of
the network. Fog computing provides immediate processing, storage,
8 Neeraj Gupta

and application services to the physical IoT devices. The aim of intro-
ducing the fog layer is to reduce latency and increase the quality of ser-
vice for the real-time applications to lower reaction time, for example,
in industrial applications, healthcare, and transportation. The signifi-
cant advantages offered are:
1. That it leads to reduced network congestion, thus improving the
latency period and lower bandwidth consumption.
2. That it addresses the security and privacy issues on a local level
rather than on a global level.
3. That it exhibits a high level of flexibility, scalability, and fault toler-
ance can be attained.
To supplement the requirements of IoT, fog computing environments
are required to address the challenges posed with regard to scalabil-
ity, interoperability, security, data quality, security, location-awareness,
mobility, and reliability. In order to do so, efforts have been made to
standardize the fog architecture. Open Edge Computing, Open Fog
Computing, and Mobile Edge Computing are among of the major
architectural models proposed by academia and industry.
C. SDN-based IoT
The network elements are responsible for forwarding and manag-
ing data packets in the networks. In traditional networking, each
networking component needs to be handled separately. This sepa-
ration leads to the cumbersome of managing an extensive network.
Software-Defined Networks (SDN) is an emerging architecture that
enables the separation of control logic from the networking elements.
The controller defines the rules and policies required to manage vari-
ous networking elements. The logically centralized control provides a
global view of the network and aids in framing rules to ensure vari-
ous complexities are managed well. SDN architecture has three main
components: Data Plane/Infrastructure Layer; Control Layer; and
Application Layer [9]. The layers interact with each other through an
Application Programming Interface (API) based on Representational
State Transfer (REST) architecture. In the context of IoT, SDN pro-
vides the capability to handle dynamically and control management
issues. The primary advantages of this approach are effective man-
agement, the seamless handling of mobile devices, the efficient distri-
bution of resources, and optimized energy management [10]. Some
challenges require attention from both academia and private industry
to fully realize the power of SDN [11].
1. Fault Tolerance: The resiliency of SDN networks is a major issue.
There is a need to design and implement the high fault-tolerant net-
work operating systems for controllers.
2. Scalability: The movement of voluminous data consumes resources
and constrains the network capabilities. Network Function Virtuali­
zation can provide alternate solutions, but the issue of performance
analysis based on the size of networks needs to be addressed.
 IoT architecture and design 9

3. Controller Placement: Controllers define the working of SDN ele-

ments and control the QoS parameters of the network. Depending
on the size of networks and their requirement, it is important to
decide the number of controllers, controller location, flow rules,
and assignment of devices to the controller.
4. Security: The controller’s security from both external and inter-
nal networks is of paramount importance. Northbound flow and
Southbound flow should undergo a rigorous authentication process
to maintain digital security.
D. Blockchain-based IoT
Blockchains enhance data security whenever multiple users want to
share the data without losing the control and ownership of it. It is
a kind of distributed database containing immutable data stored in
blocks. It is important to mention the data is organized using linear
data structures like a linked list. In the literature, blockchains are also
referred to as Distributed Ledger Technology (DLT). The consen-
sus algorithms, such as Proof of Work, Proof of Stake, and Practical
Byzantine Fault Tolerance, ensure the reliability and validity of the
transactions done by the users. Industrial IoT, healthcare, and the
Internet of Vehicles are among the key applications where blockchains
have been applied. The key issues that need attention are [12]:
1. Data Insertion: The data generated by the nodes need to be inserted
as a block and verified. There are primarily three ways to oper-
ate: client approach, manual approach, and via a connector. The
constrained nodes in the client mode insert the partial data. In the
manual system, the owner of the ‘things’ enters the data manually
using its blockchain address. The nodes in the fog/edge layer can
then take responsibility for adding the data to the chain, which can
be another viable option.
2. Identity Model: Most of the literature uses Public Key Infrastructure
(PKI) as the default way to uniquely identify the node in the IoT net-
work. Few solutions based on ontology and combinations of hierar-
chical things and PKI have been proposed, however. There is a need
to develop a more scalable solution for identity management.
3. Maintenance: The aspect reflects the security and privacy issues.
Forking causes a split in the chain whenever the rules governing the
blockchain protocol are changed. This situation can jeopardize the
overall security of the system. Similarly, a malicious node’s leak of
hash function can be a difficult situation to deal with.
Other issues, such as latency, energy, scalability, and flexibility, need to
be addressed [13].
E. Analytics-based IoT: Data analytics is essential to summarize the
information and knowledge from the data ‘things’ in the field. This
data is voluminous, varied, and sometimes requires fast processing
in real time. Most of the received data is unstructured. It is impor-
tant to undertake pre-processing activity to make it usable for analysis
10 Neeraj Gupta

Figure 1.4 Process flow for data analytics [14].

purposes. The analytics can be classified into five categories: descrip-

tive, diagnostic, discovery, predictive, and prescriptive. Figure 1.4
illustrates the process for analytics application

Dealing with a variety of data needs more focused attention. Predictive

and prescriptive analytics can be made more accurate by improving the
quality of input data.
Architectural models based on mobility [15], Artificial Intelligence [16],
information-centric networking [17], and 5G-IoT [18] are all emerging top-
ics that require active research participation from academia and industry.

1.1.4 Conclusion
The Internet of Things has vast potential to address the issues concerning day-
to-day human activities. Wireless sensor networks, fog computing, and data
analytics form the basic building blocks for IoT. This chapter discussed the
basic terminologies, challenges, and key IoT architectural models. Apart from
the two classical models that aim to standardize the IoT architecture, emerg-
ing paradigms, such as blockchains, VANETS, and Artificial Intelligence, are
being incorporated to enhance the capabilities of IoT. The applications based
on the IoT facilitate real-time monitoring, increase productivity, and reduce
the need for human effort. The COVID era underlines the necessity and use-
fulness of IoT applications. Future research activities need to focus on the
heterogeneity of devices, scalability, security, and energy-efficient systems.

REFERENCES

1. “Definition of Internet Of Things (iot) - IT Glossary | Gartner.” https://www.­

gartner.com/en/information-technology/glossary/internet-of-things (accessed May
01, 2022).
2. “State of IoT 2021: Number of connected IoT devices growing 9% to 12.3 B.”
https://iot-analytics.com/number-connected-iot-devices/ (accessed May 01, 2022).
3. “73 Important Cybercrime Statistics: 2021/2022 Data Analysis & Projections -
Financesonline.com.” https://financesonline.com/cybercrime-statistics/ (accessed
May 03, 2022).
 IoT architecture and design 11

4. “IEEE SA - IEEE 802.15.4-2020.”https://standards.ieee.org/ieee/802.15.4/7029/

(accessed May 04, 2022).
5. “RFC 6282 - Compression Format for IPv6 Datagrams over IEEE 802.15.4-
Based Networks.” https://datatracker.ietf.org/doc/html/rfc6282 (accessed May
04, 2022).
6. “CSA-IOT - Connectivity Standards Alliance.” https://csa-iot.org/ (accessed
May 04, 2022).
7. “OneM2M overview - OneM2M.” https://wiki.onem2m.org/index.php?title=
OneM2M_overview (accessed May 05, 2022).
8. “Using oneM2M.” https://onem2m.org/using-onem2m/developers/basics?jjj=
1647576221153 (accessed May 05, 2022).
9. E. Haleplidis, S. Denazis, J. H. Salim, O. Koufopavlou, D. Meyer, and K. Pentikousis,
“SDN Layers and Architecture Terminology,” 2015. [Online]. Available: http://
tools.ietf.org/html/draft-haleplidis-sdnrg-layer-terminology-04.
10. H. Zemrane, Y. Baddi, and A. Hasbi, “SDN-Based Solutions to Improve IOT:
Survey,” Proc. – 2018 IEEE 5th International Congress on Information Science
and Technology (CiST), Morocco, pp. 588–593, 2018, doi: 10.1109/CIST.2018.
8596577.
11. I. Deva Priya and S. Silas, A Survey on Research Challenges and Applications in
Empowering the SDN-Based Internet of Things, vol. 750. Springer Singapore,
2019.
12. S. K. Lo et al., “Analysis of Blockchain Solutions for IoT: A Systematic
Literature Review,” IEEE Access, vol. 7, pp. 58822–58835, 2019, doi: 10.1109/
ACCESS.2019.2914675.
13. R. A. Memon, J. P. Li, J. Ahmed, M. I. Nazeer, M. Ismail, and K. Ali, “Cloud-
Based vs. Blockchain-Based IoT: A Comparative Survey and Way Forward,”
Front. Inf. Technol. Electron. Eng., vol. 21, no. 4, pp. 563–586, 2020, doi:
10.1631/FITEE.1800343.
14. E. Siow, T. Tiropanis, and W. Hall, “Analytics for the Internet of Things,” ACM
Comput. Surv., vol. 51, no. 4, pp. 1–36, 2019, doi: 10.1145/3204947.
15. J. Li, Y. Zhang, Y. F. Chen, K. Nagaraja, S. Li, and D. Raychaudhuri, “A mobile
phone based WSN infrastructure for IoT over future internet architecture,”
Proc. - 2013 IEEE Int. Conf. Green Comput. Commun. IEEE Internet Things
IEEE Cyber, Phys. Soc. Comput. GreenCom-iThings-CPSCom 2013, pp. 426–
433, 2013, doi: 10.1109/GreenCom-iThings-CPSCom.2013.89.
16. S. K. Singh, S. Rathore, and J. H. Park, “BlockIoTIntelligence: A Blockchain-
enabled Intelligent IoT Architecture with Artificial Intelligence,” Futur. Gener.
Comput. Syst., vol. 110, no. xxxx, pp. 721–743, 2020, doi: 10.1016/j.future.
2019.09.002.
17. H. Yue, L. Guo, R. Li, H. Asaeda, and Y. Fang,“DataClouds: Enabling Community-
Based Data-Centric Services over the Internet of Things,” IEEE Internet Things
J., vol. 1, no. 5, pp. 472–482, 2014, doi: 10.1109/JIOT.2014.2353629.
18. H. Rahimi, A. Zibaeenejad, and A. A. Safavi, “A Novel IoT Architecture based
on 5G-IoT and Next Generation Technologies,” 2018 IEEE 9th Annu. Inf.
Technol. Electron. Mob. Commun. Conf. IEMCON 2018, pp. 81–88, 2019,
doi: 10.1109/IEMCON.2018.8614777.
Chapter 2

Application of Artificial
Intelligence (AI) and the Internet
of Things (IoT) in process
industries toward Industry 4.0
Rashi Verma, Raju Yerolla, Suhailam P
and Chandra Shekar Besta
National Institute of Technology, Calicut, India

CONTENTS

List of notations and abbreviations ......................................................... 14

2.1 Introduction .................................................................................... 14
2.1.1 1st industrial revolution ....................................................... 15
2.1.2 2nd industrial revolution ...................................................... 15
2.1.3 3rd industrial revolution ...................................................... 15
2.1.4 4th industrial revolution ...................................................... 16
2.2 Industry 4.0 .................................................................................... 16
2.2.1 Need for Industry 4.0 ........................................................... 17
2.2.2 The Significance of Industry 4.0 ........................................... 18
2.2.3 Technologies of Industry 4.0 ................................................ 19
2.2.4 Nine components that are bringing about the
transformation to Industry 4.0 ............................................. 19
2.2.4.1 Cloud computing .................................................... 19
2.2.4.2 Big data .................................................................. 19
2.2.4.3 The Internet of Things ............................................ 21
2.2.4.4 Augmented reality ................................................... 22
2.2.4.5 System integration: horizontal and vertical
system integration ................................................... 22
2.2.4.6 Cybersecurity .......................................................... 23
2.2.4.7 Simulation .............................................................. 23
2.2.4.8 Autonomous robot ................................................. 24
2.2.4.9 Additive manufacturing .......................................... 24
2.2.5 Issues and challenges faced by Industry 4.0 .......................... 25
2.3 Artificial intelligence ....................................................................... 26
2.3.1 Artificial Intelligence: a brief history .................................... 27
2.3.2 Approach of Artificial Intelligence in Industry 4.0 ................ 28
2.3.3 Key elements in Industrial AI ............................................... 29
2.3.4 Industrial AI ecosystem ........................................................ 29
2.3.5 Limitations of industrial Artificial Intelligence ...................... 30

DOI: 10.1201/9781003407300-2 13
14 Rashi Verma et al.

2.4 Examples ........................................................................................ 32

2.4.1 Automotive .......................................................................... 32
2.4.1.1 Description ............................................................. 32
2.4.2 Food and beverage ............................................................... 33
2.4.2.1 Description ............................................................. 33
2.4.3 Medical equipment .............................................................. 33
2.4.3.1 Description ............................................................. 33
2.5 Conclusion ...................................................................................... 33
Acknowledgements ................................................................................. 34
References ............................................................................................... 34

LIST OF NOTATIONS AND ABBREVIATIONS

AI Artificial Intelligence
CPS Cyber Physical System
ICTs Information & Communication Technology
CPPs Cyber Physical Production System
IoT Internet of Things
IoS Internet of Services
IoMs Internet of Manufacturing Services
IoP Internet of People

2.1 INTRODUCTION

The Industrial revolution has been deemed the most influential revolution
in history due to its pervasive impact on people’s everyday lives. The term
“industrial revolution” has been coined for a period of history that began in
eighteenth-century Great Britain and finished in the twenty-first century. As
a consequence of the acceleration of technical innovation processes, this has
resulted in a plethora of new tools and devices. Additionally, it includes sub-
tler technological advancements in areas such as labor, manufacturing, and
resource use. The industrialization, which commenced in the 18th century, was
a turning point in human history and has continued up to the current day. We
are now on the brink of a technological revolution that will impact our lives,
jobs, and relationships for the rest of our lives. The transition will be unpar-
alleled in terms of its scope, breadth, and complexity. We do not yet know
how it will play out. However, one thing is certain: a thorough and coordi-
nated response is required, which will include all stakeholders, from the public
and business sectors to education and civil society. Everyone is connected to
wider global developments. The industry has seen four revolutions to date [1].
Following each revolution, industry has upgraded with a new generation of
technology. The chronology of the industrial revolution is shown in Figure 2.1.
 Application of AI and IoT in process industries toward Industry 4.0 15

Figure 2.1 Historical perspective of industrial revolutions.

2.1.1 1st industrial revolution

The beginnings of the First Industrialization in the eighteenth century were
ushered in by the discovery of steam engines and the mechanization of pro-
duction. What had previously been done mechanically on a basic spinning
wheel with threads now produced eight times the size in the same time-
frame. The most important development in terms of increasing human pro-
ductivity was the development of steam power. Where muscle power had
previously been used to operate weaving looms, steam engines would be
employed. This led to revolutionary breakthroughs such as the introduction
of steam-powered railways, which reduce the time for individuals and items
to travel over vast distances.

2.1.2 2nd industrial revolution

The Second Revolution began in the mid-nineteenth century along with
the development of electricity and the introduction of production lines, the
twentieth century was transformed. In one of the most significant innova-
tions Henry Ford adapted the concept of a conveyor belt originating from
the slaughterhouse to automotive manufacturing, significantly altering the
process and ushering in the age of mass production. The car was now manu-
factured in partial phases on conveyor belts, which both accelerated the
procedure and lowered the costs of production.

2.1.3 3rd industrial revolution

The rapid spread of memory-programmable controllers and computers to
automate tasks in the 1970s was the driving force of the Third Industrial
Revolution. Thanks to the advancement of these technologies, we can now
manage the whole manufacturing process with a minimal need of human
assistance. This is demonstrated by robots that follow pre-programmed
orders without the need for human input.
16 Rashi Verma et al.

2.1.4 4th industrial revolution

At the time of writing, the Fourth Industrial Revolution is well underway.
This is referred to in the literature as “Industry 4.0” and is distinguished
by the use of information and communication technology in manufactur-
ing. These initiatives builds on the achievements of the Third Industrial
Revolution Production System, which has already integrated computer tech-
nology. This is now strengthened by a network connection and has Internet-
connected digital twins in certain respects. These enable systems to interact
and to generate information about one another. This is the next level of
industrial computerization. All systems are connected, resulting in a “cyber-
physical production system” and, consequently, automated systems in which
manufacturing systems, elements, and people interact over a network and
output is almost self-sufficient.
Industry 4.0 has the potential to dramatically alter the manufacturing
environment if all of these enablers function collectively. Examples include
machines that can detect malfunctions and begin repair operations on their
own, as well as self-organized logistics that can adapt to unanticipated
changes in production.
Industry 4.0 is a huge success for the manufacturing sector. Manufacturing
will be impacted in a variety of ways as a result of digitization, including the
way in which things are created and distributed, as well as how items are
managed and enhanced.

2.2 INDUSTRY 4.0

Since the first Industrial Revolution manufacturing has transitioned from

water and steam power to mechanize the production machinery to electric
power to create mass production. Then information technology to automat
production, thereby increasing the complexity, automation, and sustainabil-
ity of manufacturing processes while allowing workers to operate machin-
ery simply, efficiently, and consistently.
In today’s environment, industrial processes must be both computerized
and sophisticated. In the industrial sector, mass manufacturing is increas-
ingly being phased out in favour of personalized items. Significant advance-
ments in manufacturing technology and applications have all contributed to
increases in productivity. Industry 4.0 is characterized by a higher level of
organization and supervision throughout the whole value chain of a prod-
uct’s life cycle, with an emphasis on more personalised customer expecta-
tions. Industry 4.0, which integrates the Internet of Things, intelligent
manufacturing, and cloud-based manufacturing, remains a future, albeit
feasible concept. Figure 2.2 depicts the transition of conventional industries
to Industry 4.0. Industry 4.0 refers to the tight integration of humans into
manufacturing processes in order to promote continuous improvement and
a higher focus on value-added activities while reducing waste.
 Application of AI and IoT in process industries toward Industry 4.0 17

Figure 2.2 Industry 4.0 will lead to higher profitability and productivity © CC
BY 3.0 [2].

2.2.1 Need for Industry 4.0

Industry 4.0’s objective is to turn conventional machines into self-aware,
self-learning machines capable of optimising their general efficiency and
upkeep control via interaction with their environment. The purpose of
Industry 4.0 is to create a platform for data-driven industrial applications
that is open and cognitive. Industry 4.0’s main requirements include actual
data monitoring, recording the status and location of things, and retaining
instructions to control industrial operations [3].
18 Rashi Verma et al.

2.2.2 The Significance of Industry 4.0

Industry 4.0 principles are intended to allow businesses to work more flexi-
bly and to analyse large volumes of data in real time, resulting in better stra-
tegic and operational decisions. This new industrial stage has been enabled
by the extensive usage of ICTs in industrial settings, as well as the declining
cost of electronics [4], which has resulted in the installation of more sensors
in physical objects. These advances in technology have opened the path for
the embedded system design and networked setup. This setup seeks to moni-
tor and control equipment, conveyors, and products using a feedback loop
that gathers a vast quantity of large amount of data and modifies the digital
model by integrating data from physical processes. The outcome is a smart
factor, as seen in Figure 2.3.
As a consequence, several technologies have evolved and been integrated
into manufacturing systems since the 1980s. These have included cloud
technology for on-demand manufacturing services, modelling for calibra-
tion, and rapid prototyping for a flexible production system. Businesses that
use a Cyber Physical System (CPS) for information processing benefit from
increased decision support and can respond more quickly to a range of con-
ditions, such as breakdowns in manufacturing lines. As a consequence,
through the integration of manufacturing with smart grids for energy effi-
ciency, these systems may increase corporate productivity through the opti-
misation of resource use.

Figure 2.3 Concept of smart factories © CC BY 3.0 [2].

 Application of AI and IoT in process industries toward Industry 4.0 19

In addition, Industry 4.0 offers opportunities and benefits for company

development. Collaborative networks of enterprises use the horizontal inte-
gration concept to pool resources, split risks, and respond swiftly to market
changes, enabling them to grasp new opportunities. Figure 2.4 depicts the
Industry 4.0 structure, along with its various functionalities.

2.2.3 Technologies of Industry 4.0

Industry 4.0, according to the German Federal Government, is an evolving
framework where logistics and shipping networks are concerned, including
structure such as the Cyber Physical Production Systems (CPPS) shown in
Figure 2.5. This makes widespread use of the widely accessible information-
sharing system for the automated exchange of data and in which production
and business systems are consolidated [5].
Industry 4.0’s nine elements, as outlined in Figure 2.7 [6], will change
everything from segregated and optimised cell manufacturing into a com-
pletely automated, optimised, and integrated process. As a consequence, tra-
ditional manufacturing linkages between suppliers, producers, and consumers,
as well as between people and robots, will improve and evolve [7].

2.2.4 Nine components that are bringing about the

transformation to Industry 4.0
2.2.4.1 Cloud computing
The technical infrastructure that connects and communicates the sector is a
cloud-based IT platform and many parts of the 4.0 Application Centres [8].
With Industry 4.0, enterprises will be required to communicate more data
across locations and businesses, resulting in response times of a millisecond
or less [7]. “Digital production” is a term that refers to the process of con-
necting several devices to a common cloud for the purpose of exchange
data, and also the collection of equipment both on a production floor and
throughout the whole operation (Marilungo et al. 2017).

2.2.4.2 Big data
To allow instantaneous decision-making, the gathering and detailed assess-
ment of large amounts of data is required. A collection of data from various
origins such as industrial machines and technology, as well as corporate and
client solutions, will become commonplace [7]. Big data, as per Forrester, is
made up of four factors: the amount of data, the variety of data, the rate
at which new data is collected and processed, and data value. To find out
more, collected information from previously acquired data is used find the
risks that happened during different manufacturing processes in the past in
 20
Rashi Verma et al.
Figure 2.4 The Industry 4.0 paradigm.
 Application of AI and IoT in process industries toward Industry 4.0 21

Figure 2.5 CPS – Cyber Physical Systems © CC BY 3.0 [2].

addition to anticipate challenges that may develop in the future, as well as

alternative remedies to avoid them from reoccurring [10].

2.2.4.3 The Internet of Things

The Internet of Things (IoT) is a universal system of connected and evenly
identified items that correspond via customary protocol [11]. The complete
architecture of the IoT includes the Internet of Services (IoS), the Internet
of Manufacturing Services (IoMs), the Internet of People (IoP), integrated
devices, and technology and communication integration (IICT) [12]. Core
parts of this component include context, limitlessness as well as effective-
ness with context referring to an item’s capacity to interact in a sophis-
ticated manner with its present surroundings and respond quickly when
anything changes. Optimization demonstrates the notion that today’s items
are greater than they used to be when simply interfaces to a human-operated
system [13]. Omnipresence communicates information about an entity’s
location, physical or atmospheric factors, and optimizing shows that pres-
ent goods are far more than just the linkages to a system of human input at
the interaction point between humans and machines.
The production process ought to be smart, flexible, and linked via the
mixing of materialistic goods, human determinants, inventive equipment,
sensors of high intelligence manufacturing processes, and across the assem-
bly line’s organisational confines. Metrics and technology will, in future, be
critical components of smart machine planning and control [12]. Integrated
stacking and containers, for example, will become the motivating force
underlying inventory management and storage. When it comes to freight
22 Rashi Verma et al.

Figure 2.6 Internet of things, Internet of services, Internet of data © CC BY 3.0 [2].

transit, monitoring becomes more accurate, quicker and more secure.

Figure 2.6 shows the basic elements of Industry 4.0.

2.2.4.4 Augmented reality
Industry 4.0 research is highlighting best practices for enhancing quality in
assembly processes via the use of augmented reality. Augmented reality solu-
tions might help with various tasks, such as gathering components in a ware-
house and conveying maintenance procedures to mobile devices. The industry
may use augmented reality to provide employees with accurate information,
helping them to make better choices and carry out tasks more quickly. During
the time analyzing the system in need of repair, operators may be able to
obtain restoration information on how to rebuild a specific item [7].

2.2.4.5 System integration: horizontal and vertical system

integration
In the field of production, the two primary strategies involved are incorpo-
ration and self-optimization [14]. Vertical integration, networked manufac-
turing systems, and collaborative manufacturing are the three integration
aspects outlined by the Industry 4.0 paradigm: (a) horizontal integration
across the whole value creation network; (b) vertical integration throughout
 Application of AI and IoT in process industries toward Industry 4.0 23

the full network of value generation; and (c) engineering from beginning to
conclusion from start to finish throughout the product life cycle.
Vertical and horizontal digital integration and computerization of indus-
trial operation need the automation of communication and collaboration,
most notably along established procedures [15].

2.2.4.6 Cybersecurity
Alongside the expanding accessibility and utilization of conventional com-
munications protocols associated with Industry 4.0, the demand to protect
crucial manufacturing technologies and production lines from cybersecurity
risks has increased substantially. Consequently, secure, reliable communica-
tions are vital, as are enhanced machine and user identification and access
control [7]. Closely integrating the physical, service, and digital worlds may
improve the quality of information required for the design, optimization,
and operation of manufacturing systems [8]. The phrase “computing, com-
munication, and control systems” (CPS) refers to “Computing, communica-
tion, and control mechanisms are all intertwined in natural and man-made
systems (physical space)” [10]. The primary features of CPS are decentral-
ization and process autonomy.
Supply networks, which are categorized as Collaborative Cyber Physical
Systems since they’re commonly utilized in production system in addition to
other cyber physical systems like municipal traffic control and control sys-
tems [16], are important parts of CPS development. The use of cloud tech-
nology to automatically link cyber physical systems in real time allows for
continuous data flow [17]. The digital shadow of manufacturing represents
a physical thing in a virtual or information realm. Massive cyber physical
systems problems are used to solve the basic demand of manufacturing
operations and system optimization in real settings [12]. By incorporating
relevant sensors into CPS, it should be possible to identify machine failure
and prepare CPS automatically for fault repair tasks. The cycle time required
to conduct the task at that station is also utilized to establish the appropriate
allocation of resources at each workstation [18]. Cloud computing is used
in the 5C architecture to link machines (machine-to-machine or human-to-
machine) [19]. For example, the smart vehicle is an example of integrated
Cyber Physical System (CPS) manufacturing that represents Industry 4.0
innovation [20]. This production employs an information-mining technique
to perform prognosis that attains a precision of 80% [21].

2.2.4.7 Simulation
Simulations will be increasingly widely utilized in production systems that
harness real-time data to create a virtual model of the actual environ-
ment, which may include machines, goods, and workers, therefore reduc-
ing machine setup times and improving quality [7]. Virtual commissioning
24 Rashi Verma et al.

may be accomplished using 2D and 3D simulations, as well as models of

cycle times, power consumption, and production plant layouts. During the
startup phase, simulation may help decrease downtime and changes, as well
as addressing production issues [21]. Additionally, simulation enhances deci-
sion-making by simplifying and expediting it [14].

2.2.4.8 Autonomous robot
On a daily basis, robots become increasingly self-sufficient, flexible, and
coordinated, and they will unavoidably connect with one another and oper-
ate safely among others while learning from experience [7]. An autonomous
robot is used to improve the accuracy of autonomous manufacturing pro-
cesses and to operate in locations where there are few people. Autonomous
robots are capable of performing jobs accurately and rationally in a certain
amount of time [22] while also prioritizing safety, adaptability, diversity, and
participation, Table 2.1 shows some of the examples.

2.2.4.9 Additive manufacturing
Additive manufacturing is primarily used in Industry 4.0 to create more
personalized items with complex and lightweight constructions. If addi-
tive manufacturing is decentralized, the cost of production will decrease,
resulting in a greater performance rate [7]. Production should be quicker
and less costly through the use of different manufacturing methods such as
fused deposition modeling (FDM), selective laser melting (SLM), and selec-
tive laser sintering (SLS) [8]. Numerous businesses confront the challenge
of expanding product customization and decreasing time to market as cus-
tomer demands evolve on a regular basis. Digitization, IT sector involve-
ment, and correct processes were able to resolve the aforementioned issues.
A more efficient organizational structure was required to accommodate
growing demand while also extending the life of the product.
For instance, identical model automobiles are offered with a variety of
engine, bodywork, and equipment choices, all designed to satisfy the demand
of more savvy clients who are selective [13]. Figure 2.7 depicts the various
technologies of Industry 4.0.

Table 2.1 Example of autonomous robots

Serial Manufacturing
Robot name Functioning
No. company
1 Kuka LBR Kuka A lightweight robot is designed for critical
iiwa industrial tasks
2 Baxter Rethink Robotics Packaging robot with interactive capabilities
3 BioRob Arm Bionic robotics Used in areas close to human
 Application of AI and IoT in process industries toward Industry 4.0 25

Figure 2.7 Industry 4.0 technologies.

2.2.5 Issues and challenges faced by Industry 4.0

Mechanical systems have been used since the dawn of time, and today’s
highly automated assembly lines are, to some extent, a continuation of this
process. New technologies have benefited industry development by enabling
it to be flexible and adaptive to shifting market needs. Embed ability, pre-
dictability, adaptability, and robustness are just a few of the unforeseen con-
ditions that may arise. Numerous complications and basic concerns develop
with the program’s implementation of “Industry 4.0” [12].
Intelligent Decision-Making and Negotiation Mechanisms: The two most
critical characteristics of a smart factory are autonomy and sociability.
However, since the majority of systems adhere to the fundamental 3C para-
digm (Company-Client-Competitor), they lack these two features. Further
research is necessary to develop an autonomous manufacturing architecture
in lieu of the hierarchical structure [12].

• High-Speed IWN Protocols: Industry needs a robust and reliable com-

munication network capable of transporting large amounts of data.
However, the IWN (Industrial Wireless Protocols) standards WIP-PA
and Wireless HART do not offer sufficient bandwidth, making them
inefficient and slow [12].
• Production-specific Big Data and Analytics: While every industry gen-
erates a significant quantity of data, assuring the quality and integrity
26 Rashi Verma et al.

of data that focuses on the usable characteristics of data captured by

production systems is a challenging challenge. By including just the
unique characteristics of data, it may be possible to ascertain the man-
ufacturing process’s quality and efficiency.
• System Modeling and Analysis: Any system may be modeled as a self-
organized manufacturing process using the appropriate dynamic equa-
tion and control model. However, this is only viable for basic systems;
larger systems remain a hot focus of research [12].
• Cybersecurity: Industry 4.0 requires robust cybersecurity prior to
going live. Any kind of cyber attack might result in a significant finan-
cial loss for the business. Although some smart factories use encryp-
tion and authorization mechanisms as a form of cybersecurity, this
may prove insufficient [7].
• Modularize and Flexibility Physical Artifacts: In order to manufacture
a product, machining and testing equipment ought to be shared and
cooperated on in order to facilitate dispersed decision-making. As a
consequence, what is needed is a modular and intelligent conveying
device capable of changing production routes dynamically [12].
• Investment Issues: Investment is a relatively broad problem that affects
the majority of new technology-based manufacturing endeavors.
Significant expenditure is necessary to adopt industry standards. 4.0
are originally a SME (Structural Equation Modelling). Implementing
all of Industry 4.0’s pillars requires a significant investment from an
industry [23].

2.3 ARTIFICIAL INTELLIGENCE

Artificial Intelligence (AI) is a vast subject of computer science focused

on the development of intelligent computers capable of undertaking tasks
that would typically need human intelligence. While Artificial Intelligence
(AI) has been defined in a variety of ways over the past few decades, John
McCarthy recommends the following definition in this 2004 study: “It
is the science and engineering of creating intelligent devices, particularly
intelligent computer programs.” This is comparable to the analogous chal-
lenge of utilizing computers to comprehend human intellect, although
AI is not limited to physiologically observable techniques. The Artificial
Intelligence issue was founded, however, by Alan Turing’s landmark 1950
paper “Computing Machinery and Intelligence.” In this paper, Turing, who
has been called the “Father of Computer Science,” answers the topic “Can
Machines Think?” Artificial Intelligence was the answer to this very ques-
tion. Artificial Intelligence is an area of computer science whose purpose
is to answer Turing’s question in the affirmative. Researchers working on
Artificial Intelligence are attempting to mimic or emulate human intellect
in robots.
 Application of AI and IoT in process industries toward Industry 4.0 27

2.3.1 Artificial Intelligence: a brief history

Smart robots and artificial beings made their first appearance in Greek
mythology. Aristotle’s invention of the syllogism and its application of ratio-
nal thinking marked a turning point in civilization’s quest to understand its
own mind. Despite its lengthy and illustrious history, Artificial Intelligence
as we comprehend it now has only been around for about a century. Here’s
a rundown of some of the most significant AI events in recent years.

1940s
• “A Logical Calculus of Concepts Intrinsic in Nervous Action,” by
Warren McCullough and Walter Pitts, was published in 1943. This
was the first mathematical formalism for building a learning algorithm
was described in this study.
• The Organization of Behavior: A Neuropsychological Theory, written
by Donald Hebb in 1949, proposed that experiences alter brain cir-
cuits and that connections between neurons become stronger as they
are utilized more frequently.

1950s
• In 1950, Alan Turing publishes “Computing Machinery and Intelligence,”
in which he proposes the Turing Test, a technique for determining
whether or not a computer is intelligent.
• In 1950, two Harvard freshmen, Marvin Minsky and Dean Edmonds,
build SNARC (Stochastic Neural Analog Reinforcement Calculator),
the first neural network computer.
• In 1950, Claude Shannon releases “Programming a Computer for
Chess Playing.”
• Isaac Asimov wrote “The Three Laws of Robotics” in 1950.

1960–2010
• In 1963, John McCarthy founds the Stanford Artificial Intelligence
Laboratory.
• In 1972, PROLOG, a logic programming language, is created.
• Japan’s Ministry of International Trade and Industry (MITI) launched
the Fifth Generation Computer Systems project in 1982.
• In response to Japan’s FGCS, the US government launched the Strategic
Computing Initiative in 1983, which funds advanced computing and
Artificial Intelligence research via DARPA. DART, an automated logis-
tical planning and scheduling program, was used by US troops during
the 1991 Gulf War.
• In 1997, Gary Kasparov, the world chess champion, is beaten by IBM’s
Deep Blue.
• In 2005, STANLEY, a self-driving car, wins the DARPA Grand Challenge.
28 Rashi Verma et al.

• In 2008, Google advances voice recognition technology and makes it

accessible through its iPhone app.

2010–2020
• Apple introduced Siri, an Artificial Intelligence-powered virtual assis-
tant, as part of their iOS operating system in 2011.
• In 2014, Google’s self-driving vehicle passes the state’s driving test for
the first time.
• Amazon launched Alexa, a virtual assistant, in 2014.
• In 2016, Hanson Robotics introduces Sophia, the world’s first “robot
citizen,” a humanoid robot capable of face recognition, linguistic com-
munication, and facial expression.
• Baidu will make its Linear Fold AI algorithm accessible to scientists
and physicians working on a SARS-CoV-2 vaccine during the early
phases of the pandemic in 2020. In only 27 seconds, the algorithm
can predict the virus’s RNA sequence, which is 120 times quicker than
previous approaches.

While Artificial Intelligence is a diverse discipline with a variety of method-

ologies, advances in algorithms are causing a paradigm shift in practically
every aspect of the IT sector. AI systems now have a wide range of real-
world applications.

2.3.2 Approach of Artificial Intelligence in Industry 4.0

Artificial Intelligence (AI) is a branch of machine learning that involves
image analysis, text classification, automation, and advanced analytics.
Broadly speaking, machine learning and Artificial Intelligence have been
seen as “black-art” methods, with a dearth of convincing evidence to per-
suade business that these techniques would work regularly in a consistent
manner while offering a financial return on investment simultaneously, the
efficiency of learning algorithms is extremely reliant on the programmer
expertise and choices. As a consequence, the success of Artificial Intelligence
in industrial applications is restricted. Industrial AI, on the other hand, is a
field devoted to developing, verifying, and implementing varied algorithms
are developed for industrial applications with lengthy productivity.
It’s a useful method and discipline for producing answers for industrial
implementation, providing a link between academic and industry AI devel-
opment. Automation powered by Artificial Intelligence has a substantial
quantitative influence on the rise of productivity. In addition, today’s indus-
tries face new market demands and competitive pressures.
They need a paradigm shift dubbed Industry 4.0. Integrating AI with new
technologies such as the Industrial Internet of Things (IIoT) [24], Big Data
analytics [20, 24–26], cloud computing [27, 28], and cyber physical systems
[10, 10, 12] will enable businesses to function more efficiently, effectively,
and sustainably.
 Application of AI and IoT in process industries toward Industry 4.0 29

Due to the fact that industrial AI is still in its infancy, its structure, tech-
niques, and difficulties must be defined in order to provide a foundation for
its use in industry. To do this, we developed an Industrial AI ecosystem that
incorporates the field’s most critical components and acts as a road map for
increased knowledge and application. The methods and technologies on
which an Industrial AI system may be constructed are also discussed.

2.3.3 Key elements in Industrial AI

ABCDE: The acronym ‘ABCDE’ may be used to define the critical compo-
nents of Industrial AI. The important variables are analytics technology (A),
Big Data technology (B), cloud or cyber technology (C), domain expertise
(D), and evidence (E) [29]. While research is at the basis of AI, it is only ben-
eficial when combined with other components. Both Big Data technology
and cloud computing are key components that serve as both a data source
and a foundation for Artificial Intelligence in industry. While they are key
components, domain expertise and proof are also critical concerns that are
sometimes ignored in this scenario. Domain expertise is critical in the fol-
lowing areas:

1. Recognizing the issue and directing Industrial AI’s capabilities toward

resolving it;
2. Comprehending the system in order to acquire data of the appropriate
quality;
3. Understanding the technical meanings of the metrics and how they
correlate to the fundamental properties of a system or processes;
4. Recognizing how these factors vary amongst systems.

Evidence is also required to validate Industrial AI models and to include

learning through time capabilities into them. We can only develop the AI
model over time by amassing trends in data and the information (or label)
that goes with them. Figure 2.8 illustrates how AI may take us from visible
to invisible space, and from resolving difficulties to preventing them from
occurring in the first place.

2.3.4 Industrial AI ecosystem
The proposed Industrial AI ecosystem, as depicted in Figure 2.9, is a pro-
gressive approach for thinking for determining industry demands, issues,
technologies, and strategies in order to build revolutionary AI systems. This
diagram may be used as a logical framework for developing and imple-
menting an Industrial AI strategy. This ecosystem highlights common unmet
business demands such as self-awareness, self-comparison, self-prediction,
self-optimization, and resilience.
When seen through the lens of the Cyber Physical Systems (CPS) proposed
in [2], these four technologies become more intelligible. The above mentioned
30 Rashi Verma et al.

Figure 2.8 I ndustrial AI’s impact: from resolving apparent issues to avoiding prob-
lems those aren’t evident.

four in the 5C era, technology is the key to success. (Connection, Conversion,

Cyber, Cognition, and Configuration), as shown in Figure 2.10.

2.3.5 Limitations of industrial Artificial Intelligence

Industrial AI is accompanied with a plethora of expectations; even if only
some of these expectations are met, it will result in a “one-of-a-kind” experi-
ence and substantial difficulties in deploying AI to industries. [29] Among
the present obstacles and complications, the following concerns and com-
plexities are more important and sorted:

1. Interactions between machines: While AI algorithms can properly

convert a set of inputs to a set of outputs, they are vulnerable to min-
ute input discrepancies caused by machine variances. It must guaran-
tee that future implementations of specific Artificial Intelligence (AI)
solutions do not interfere with the operation of other systems.
2. Data integrity is one of the most important factors to consider. Large,
clean data sets with low errors are required for AI systems. The down-
stream repercussions of learning from erroneous or insufficient data
sources may be contaminant
3. Information security: The smart manufacturing system is becoming pro-
gressively prone to cyber attack as more linked technologies are used.
The scope of this vulnerability is currently underestimated, and the
industry is unprepared to deal with the security threats that ensue [11].
Application of AI and IoT in process industries toward Industry 4.0 31
Figure 2.9 Industrial AI eco-system.
32 Rashi Verma et al.

Figure 2.10 Technology for CPS introduction in the industrial sector.

2.4 EXAMPLES

2.4.1 Automotive
2.4.1.1 Description
Automobile manufacturers use a number of technologies and components
to enhance their vehicles’ utility and cost. Hydraulic systems are important
because they govern vehicle braking and the steering wheel, two signifi-
cant aspects of driving. These hydraulic systems need high-pressure pipes
or hoses to operate properly. Due to the wide range of vehicle designs, a
plethora of hydraulic hose variations are made every day to serve busy auto-
motive manufacturing lines.

Challenges
• A wide range of those options.
• Manual inspection is inefficient and expensive.

Solution
• Computer vision models have been taught to reliably recognise and
categorise different types of defects (cuts, gouges, drag marks, and so
on) at a rate of 99% or higher.
• When compared to a fully human method, an automated end-of-line
inspection using computer vision achieves almost 100% coverage and
improves categorization uniformity.
 Application of AI and IoT in process industries toward Industry 4.0 33

2.4.2 Food and beverage

2.4.2.1 Description
While making bread or pastries at home may seem to be a simple process
capable of being handled by the majority of people, producing baked goods
on a large scale offers a number of challenges. The aesthetic of the baked
product is essential in the client’s eyes, and many returns are based on how
a product appears rather than how it tastes.

Challenges
• One-on-one examination by a worker is prohibitively expensive.
• Escaping faults is considerably more expensive than non-escaping
defects due to increased expenses in packing, shipping, and inventory
management.

Solution
• The rapid training of computer vision models for the purpose of iden-
tifying damaged items.
• Side cameras with an edge device and stream processing provide for
real-time computing and 100% inspection at line speeds.

2.4.3 Medical equipment
2.4.3.1 Description
Pharmaceutical products are complicated and are frequently mass-produced
on automated lines capable of outputting 100 million or more units per year.
To increase profitability, several pharmaceutical businesses are adopting digi-
tal transformation. The use of cameras and streaming analytics tools to help
with fault identification on high-throughput lines is one aspect of this trend.

Challenges
• Manual inspection is time-consuming and ineffective in detecting
some sorts of problems.
• High rejection product cost.
• Limited line availability and throughput.

Solution
• Computer vision enables the detection of problems that are not readily
apparent to the naked eye.
• 99% accuracy in identifying and classifying defects.

2.5 CONCLUSION

Quicker computers, smarter machines, smaller sensors, and less expen-

sive data storage and transmission may allow machines and other things
34 Rashi Verma et al.

to interact with and adapt to one another. The nine components of

Industry 4.0 are described in depth, along with examples to help under-
stand their relevance and identify the obstacles and issues associated
with their implementation. As Industry 4.0 adoption grows, additional
research streams, such as accessible and manufacturing administration
and a well-structured distribution network, information gathering from
manufacturing lines, and optimization of that input for an application in
appropriate equipment, energy efficiency, and strategic planning, should
be produced.
Industry 5.0 is a term that refers to research disciplines considered part of
the next industrial age. However, it also refers to a broader transition that
includes effects on social order, government and institutions, human person-
alization, and economic and industrial effects. This section explores the
industrial applications of Artificial Intelligence and the Internet of Things.
Since Artificial Intelligence and the Internet of Things have advanced from
science fiction to the forefront of game-changing innovations, there is an
urgent need for systematic Artificial Intelligence and the Internet of Things
research and implementation to track their true impact on Industry 4.0,
which refers to the next wave of technological systems. This chapter explains
how AI and IoT fit into the Industry 4.0 concept. By providing an insight
into the Industrial AI and IoT ecosystem in current production, this chapter
also intends to give a foundation for conceptualizing steps toward adopting
Industrial AI and IoT technologies.

ACKNOWLEDGEMENTS

We want to thank our faculty-in-charge of the National Institute of Technology,

Calicut, for guiding us in writing this book chapter. We would also like to
thank the reviewer for doing a detailed review and giving useful comments.

REFERENCES

1. E. Oztemel and S. Gursev, “Literature review of Industry 4.0 and related

technologies,” J. Intell. Manuf., vol. 31, no. 1, pp. 127–182, Jan. 2020, doi:
10.1007/s10845-018-1433-8.
2. M. Crnjac, I. Veža, and N. Banduka, “From Concept to the Introduction of
Industry 4.0,” Int. J. Ind. Eng. Manag., vol. 8, no. 1, pp. 21–30, 2017.
3. F. Almada-Lobo, “The Industry 4.0 revolution and the future of Manufacturing
Execution Systems (MES),” J. Innov. Manag., vol. 3, no. 4, Art. no. 4, 2015,
doi: 10.24840/2183-0606_003.004_0003.
4. A. Schumacher, S. Erol, and W. Sihn, “A Maturity Model for Assessing Industry
4.0 Readiness and Maturity of Manufacturing Enterprises,” Procedia CIRP,
vol. 52, pp. 161–166, Jan. 2016, doi: 10.1016/j.procir.2016.07.040.
 Application of AI and IoT in process industries toward Industry 4.0 35

5. M. A. K. Bahrin, M. F. Othman, N. H. N. Azli, and M. F. Talib, “Industry 4.0:

A Review On Industrial Automation AND Robotic,” J. Teknol., vol. 78, no.
6–13, Art. no. 6–13, Jun. 2016, doi: 10.11113/jt.v78.9285.
6. S. Parhi, K. Joshi, T. Wuest, and M. Akarte, “Factors Affecting Industry 4.0
Adoption – A Hybrid SEM-ANN Approach,” Comput. Ind. Eng., vol. 168,
p. 108062, Jun. 2022, doi: 10.1016/j.cie.2022.108062.
7. M. Rüßmann, M. Lorenz, P. Gerbert, M. Waldner, J. Justus, and M. Harnisch,
“Industry 4.0: The Future of Productivity and Growth in Manufacturing
Industries,” vol. 9, p. 14, 2015.
8. M. Landherr, U. Schneider, and T. Bauernhansl,“The Application Center Industrie
4.0 - Industry-driven Manufacturing, Research and Development,” Procedia
CIRP, vol. 57, pp. 26–31, Jan. 2016, doi: 10.1016/j.procir.2016.11.006.
9. E. Marilungo, A. Papetti, M. Germani, and M. Peruzzini, “From PSS to CPS
Design: A Real Industrial Use Case Toward Industry 4.0,” Procedia CIRP, vol.
64, pp. 357–362, Jan. 2017, doi: 10.1016/j.procir.2017.03.007.
10. B. Bagheri, S. Yang, H.-A. Kao, and J. Lee, “Cyber-Physical Systems Architecture
for Self-Aware Machines in Industry 4.0 Environment,” IFAC-Pap., vol. 48,
no. 3, pp. 1622–1627, Jan. 2015, doi: 10.1016/j.ifacol.2015.06.318.
11. P. Leitão, S. Karnouskos, L. Ribeiro, J. Lee, T. Strasser, and A. W. Colombo,
“Smart Agents in Industrial Cyber–Physical Systems,” Proc. IEEE, vol. 104,
no. 5, pp. 1086–1101, May. 2016, doi: 10.1109/JPROC.2016.2521931.
12. J. Qin, Y. Liu, and R. Grosvenor, “A Categorical Framework of Manufacturing
for Industry 4.0 and Beyond,” Procedia CIRP, vol. 52, pp. 173–178, Jan. 2016,
doi: 10.1016/j.procir.2016.08.005.
13. K. Witkowski, “Internet of Things, Big Data, Industry 4.0 – Innovative
Solutions in Logistics and Supply Chains Management,” Procedia Eng., vol.
182, pp. 763–769, Jan. 2017, doi: 10.1016/j.proeng.2017.03.197.
14. G. Schuh, T. Potente, C. Wesch-Potente, A. R. Weber, and J.-P. Prote,“Collaboration
Mechanisms to Increase Productivity in the Context of Industrie 4.0,” Procedia
CIRP, vol. 19, pp. 51–56, Jan. 2014, doi: 10.1016/j.procir.2014.05.016.
15. S. Erol, A. Jäger, P. Hold, K. Ott, and W. Sihn, “Tangible Industry 4.0: A
Scenario-Based Approach to Learning for the Future of Production,” Procedia
CIRP, vol. 54, pp. 13–18, Jan. 2016, doi: 10.1016/j.procir.2016.03.162.
16. D. Ivanov, B. Sokolov, and M. Ivanova, “Schedule Coordination in Cyber-
Physical Supply Networks Industry 4.0,” IFAC-Pap., vol. 49, no. 12, pp. 839–
844, Jan. 2016, doi: 10.1016/j.ifacol.2016.07.879.
17. T. Stock and G. Seliger, “Opportunities of Sustainable Manufacturing in
Industry 4.0,” Procedia CIRP, vol. 40, pp. 536–541, Jan. 2016, doi: 10.1016/j.
procir.2016.01.129.
18. D. Kolberg and D. Zühlke, “Lean Automation Enabled by Industry 4.0
Technologies,” IFAC-Pap., vol. 48, no. 3, pp. 1870–1875, Jan. 2015, doi: 10.1016/
j.ifacol.2015.06.359.
19. D. S. de Dutra and J. R. Silva, “Product-Service Architecture (PSA): Toward a
Service Engineering Perspective in Industry 4.0,” IFAC-Pap., vol. 49, no. 31,
pp. 91–96, Jan. 2016, doi: 10.1016/j.ifacol.2016.12.167.
20. J. Shi, J. Wan, H. Yan, and H. Suo, “A survey of Cyber-Physical Systems,”
in 2011 International Conference on Wireless Communications and Signal
Processing (WCSP), Nov. 2011, pp. 1–6. doi: 10.1109/WCSP.2011.6096958.
36 Rashi Verma et al.

21. S. Simons, P. Abé, and S. Neser, “Learning in the AutFab – The Fully Automated
Industrie 4.0 Learning Factory of the University of Applied Sciences Darmstadt,”
Procedia Manuf., vol. 9, pp. 81–88, Jan. 2017, doi: 10.1016/j.promfg.2017.04.023.
22. S. Vaidya, P. Ambad, and S. Bhosle, “Industry 4.0 – A Glimpse,” Procedia
Manuf., vol. 20, pp. 233–238, Jan. 2018, doi: 10.1016/j.promfg.2018.02.034.
23. T.-T. Shi, P. Li, S.-J. Chen, Y.-F. Chen, X.-W. Guo, and D.-G. Xiao, “Reduced
Production of Diacetyl by Overexpressing BDH2 Gene and ILV5 Gene in Yeast
of the Lager Brewers with One ILV2 Allelic Gene Deleted,” J. Ind. Microbiol.
Biotechnol., vol. 44, no. 3, pp. 397–405, 2017, doi: 10.1007/s10295-017-1903-6.
24. L. D. Xu, W. He, and S. Li, “Internet of Things in Industries: A Survey,” IEEE
Trans. Ind. Inform., vol. 10, no. 4, pp. 2233–2243, Nov. 2014, doi: 10.1109/TII.
2014.2300753.
25. J. Lee, E. Lapira, B. Bagheri, and H. Kao, “Recent Advances and Trends in
Predictive Manufacturing Systems in Big Data Environment,” Manuf. Lett.,
vol. 1, no. 1, pp. 38–41, Oct. 2013, doi: 10.1016/j.mfglet.2013.09.005.
26. J. Lee, H. D. Ardakani, S. Yang, and B. Bagheri, “Industrial Big Data Analytics
and Cyber-physical Systems for Future Maintenance & Service Innovation,”
Procedia CIRP, vol. 38, pp. 3–7, Jan. 2015, doi: 10.1016/j.procir.2015.08.026.
27. D. Wu, M. J. Greer, D. W. Rosen, and D. Schaefer, “Cloud Manufacturing:
Strategic Vision and State-of-the-Art,” J. Manuf. Syst., vol. 32, no. 4, pp. 564–
579, Oct. 2013, doi: 10.1016/j.jmsy.2013.04.008.
28. L. Zhang et al., “Cloud Manufacturing: A New Manufacturing Paradigm,”
Enterp. Inf. Syst., vol. 8, no. 2, pp. 167–187, Mar. 2014, doi: 10.1080/17517575.
2012.683812.
29. J. Lee, H. Davari, J. Singh, and V. Pandhare, “Industrial Artificial Intelligence for
Industry 4.0-Based Manufacturing Systems,” Manuf. Lett., vol. 18, pp. 20–23,
Oct. 2018, doi: 10.1016/j.mfglet.2018.09.002.
Chapter 3

A review on edge computing

Working, comparisons, benefits, vision,
instances and illustrations along with
challenges
Parul Dubey
G H Raisoni College of Engineering, Nagpur, India

Pushkar Dubey and Kailash Kumar Sahu

Pandit Sundar Lal Sharma (Open) University, Bilaspur, Chhattisgarh,
India

CONTENTS

3.1 Introduction .................................................................................... 37

3.2 Literature review ............................................................................. 38
3.3 The working of edge computing ..................................................... 40
3.4 Comparison—edge computing, cloud computing and fog
computing ....................................................................................... 43
3.5 The benefits of edge computing ....................................................... 46
3.6 Instances and illustrations of edge computing ................................. 46
3.7 Edge maintenance ........................................................................... 48
3.8 Opportunities for 5G, IoT with edge computing ............................. 49
3.9 Challenges of edge computing ......................................................... 49
3.10 Conclusion ...................................................................................... 50
References ............................................................................................... 50

3.1 INTRODUCTION

Edge computing, also known as distributed information technology archi-

tecture (ITA), is an unique advantage in which client data is processed close
to the point where it was received by the server.
Currently, research is an important aspect of every organization, provid-
ing critical business information and allowing the real-time administration
of critical company operations and activities. built-in sensors and Internet of
Things (IoT) devices that function in real time from remote locations or in
difficult working circumstances have the potential to acquire massive
amounts of data from today’s businesses on a daily basis, allowing them to
become more efficient.
DOI: 10.1201/9781003407300-3 37
38 Parul Dubey et al.

As a result of these developments, businesses are reevaluating their

approach to computer use and management. Conventional computing para-
digms focused on central data centers and the internet as we know it now are
incapable of coping adequately with the massive volumes of data to be found
in the real world. These efforts may be thwarted by network constraints such
as bandwidth restrictions, latency restrictions, and unpredictability restric-
tions. Enterprises are benefiting from the introduction of edge computing
architecture to achieve address these data challenges.
Essentially, edge computing is the practice of storing information closer to
the place of origination rather than in a centralized data center. It is found
to be more efficient to process and analyze data locally as it is produced on
the ground than to having it transmitted to a centralized data centre for
processing and analysis. An example: As soon as edge computing is done,
the results, which include real-time business insights and equipment mainte-
nance estimates, are routed back to the main data centre for review and
human involvement. It should come as no surprise that edge computing is
gaining attention since it is transforming IT and business computing.
In order to process data as near to an Internet of Things device as feasible,
edge computing is utilized. As a consequence, time, efficiency, budget, and
security may all be enhanced for organizations’ IT departments. There are a
number of other essential technologies that operate in conjunction with
edge computing, including hybrid cloud and 5G. Devices and applications
that are part of the Internet of Things (IoT) obtain significant benefits from
this technology. In truth, the Internet of Things and the edge are more than
just good combinations; they are becoming increasingly intertwined.

3.2 LITERATURE REVIEW

In the wake of the rise of the Internet of Things (IoT) and the popularity of
cloud-based services, a new computing paradigm, known as edge comput-
ing, has emerged. As even the Internet of Things (IoT) and cloud services
grow more widespread, edge computing is becoming increasingly popular.
Some of the challenges that edge computing may address include security
and privacy, reaction speed and battery life to mention a few [1] Numerous
examples of edge computing in operation are provided throughout the
chapter, spanning beyond cloud dumping to smart urban and collaborative
edge computation. Edge computing difficulties and potential are explored in
this chapter, which experts believe will catch the attention of the community
and stimulate more study in this area.
Mobile edge computing (MEC) has the potential to offload Internet of
Vehicles (IoV) applications while also allocating resources. The majority of
current offloading methods are useless because they consider the applica-
tion as a whole when determining whether or not it should be offloaded. A
more sophisticated solution is presented in the paper [2], which is an
 A review on edge computing 39

application-centric architecture with a finer-grained offloading method

based on the separation of apps. Nodes in the framework represent sub-
tasks, and edges between nodes indicate data flow relationships between
subtasks; this is how we build applications in our framework. It is possible
to find offloading nodes in both automobiles and MEC servers that are
within range of the network. As part of the offloading process, subtasks are
allocated to various computing nodes, and the procedure is repeated.
Furthermore, offloading takes care of the time constraints connected with
each subtask. It has been shown via experimentation that the proposed
offloading approach outperforms alternative non-partitioning methods in
terms of both execution time and throughput.
For consumers who want to access data quickly, multi-access edge com-
puting (MEC) is an additional storage option. It is not possible for individual
edge servers to store huge volumes of data due to their physical dimension
and storage capacity limitations. As a result of the nature of their operation,
the edge servers often have to communicate data to other servers in order to
deliver services to end users. They operate in a somewhat given environment
and found that they are administered by a range of edge network operators.
One of the most significant obstacles to collaborative edge storage is the lack
of incentives and trust among the participants [3]. These difficulties are
addressed in this work by the proposal of a new decentralised system,
CSEdge, which uses blockchain technology to facilitate collaborative edge
storage. To encourage competition among CSEdge servers, edge servers may
submit data offloading requests to other CSEdge servers. Prior successes and
goodwill are taken into consideration when selecting the prize winners. Their
ability to store offloaded data will be recognised, and they’ll be rewarded for
the successful completion of a difficult task. Their reputation will be assessed
in the future on the basis of their achievements, which will be recorded on the
blockchain via a distributed consensus process. This infrastructure is built on
Hyperledger Sawtooth and has been tested in a simulated MEC against a
baseline system as well as against two current state-of-the-art systems. Based
on the results, CSEdge is capable of facilitating edge server cooperation on
storage in a cost-effective and efficient manner.
Because of the prevalence of the Internet of Things (IoT) and cloud ser-
vices, edge computing allows data processing to take place at the edge of the
network rather than solely in the cloud [4]. Edge computing may be able to
handle issues such as latency, power consumption on mobile devices, band-
width charges, security, and privacy, along with many other things.
One developing architecture, Mobile Edge Computing (MEC), extends
cloud computing capabilities to the network edge via the use of mobile base
stations [5]. It may be used in cellular, wireless, and wired applications since
it makes use of software based at the network’s edge. Mobile subscribers,
enterprises, and other vertical industries all benefit from MEC’s flawless
integration of multiple application service providers and suppliers, which is
available to all customers. It is a vital component of the 5G architecture,
40 Parul Dubey et al.

since it allows for the implementation of a broad variety of cutting-edge

applications and services that need very rapid response times. The paper
covered recent research and technological advancements in the area of
microelectromechanical systems. There is an explanation of what MEC is,
what its advantages are, how it may be implemented, and some of the prob-
able applications.
Because of the extraordinary advancements in embedded systems-on-a-
chip technology, many commercial devices now have sufficient processing
power to run full operating systems. This has led to an increase in the poten-
tial of the Internet of Things. Early Internet of Things (IoT) devices were
only capable of gathering and transmitting data for the purpose of analysis.
Because of the increasing processing capacity of current devices, however, it
is now feasible to do computations on-site, a process which is referred to as
edge computing [6]. Edge computing, which allows cloud computing to pro-
vide a greater variety of services and applications by bringing services in
response to the network’s edge, is expanding its capabilities. Edge comput-
ing technologies are currently the subject of our research; as part of this
inquiry, researchers highlight and report on recent breakthroughs in this
field. Because of this classification and categorization, we have been able to
determine the most significant and supporting characteristics of several edge
computing paradigms for the Internet of Things.

3.3 THE WORKING OF EDGE COMPUTING

It is the only part of edge computing that takes into account the physical
location of the device. In traditional corporate computing, data is usually
made at the client port, such as a person’s computer, before it is sent out to
other people. In this case, the data is distributed throughout a wide area
network (WAN), namely the internet. It is then kept and analysed by a com-
mercial application on the company’s network. Following this, the output
is sent back to the person who hired us [7]. This technique of client-server
data processing has been in use for most traditional business applications
for a considerable period.
However, conventional data centre infrastructures are unable to cope
with the rapid rise in the count of web devices and the resulting increase in
data density.
IT architects have increasingly switched their emphasis away from the
actual data centre and toward the virtual edge of their infrastructures as a
function of the transfer of computing systems from the data centre to the
area where the data is created. In most cases, data centres should be posi-
tioned as close to their customers as feasible; however, where this is not
possible, clients should be located as close as possible to the servers. As a
result, it was decided that computing resources should be located in the
neighborhood of where they are needed, rather than depending on previous
 A review on edge computing 41

generations’ understandings of distant computation, such as satellite sites

and regional offices, to house computer resources.
In order to collect and analyse data locally, edge computing puts storage
and servers on a remote LAN, sometimes requiring as little as a half-rack of
equipment in order to carry out the task. For protection against exposure to
unfavorable environmental conditions such as temperature fluctuations,
moisture incursion, and so on, many computer systems are placed in shielded
or hardened enclosures that are designed to withstand impact. It is common
for processing to include normalization and analysis of the data stream in
order to discover business insight. The results of the analysis are the only
ones that are communicated directly to the main data centre.
It can be observed that architecture that is most suitable for one kind of
computer activity may not be the framework that is suitable for other types
of computing activities. As a result of the emergence of the edge computing
architecture, it is now feasible to locate computational resources closer to
the data source, ideally at a location that is physically adjacent to the data
source. Distant location offices, service centers, data centre hosting, and
cloud computing are examples of concepts of distributed computing that
have been around for a long time and have a proven track record.
While the move away from the usual centralized computer paradigm may
require high levels of management and surveillance, decentralization may
prove to be a challenging challenge to accomplish. Large volumes of data are
generated and consumed by today’s organizations, and edge computing offers
a realistic solution to the network issues associated with transmitting that data.
To illustrate, consider the increasing popularity of self-driving vehicles.
The only way they’ll be able to travel around will be via computerized traffic
lights. Real-time data creation and processing will be required for driverless
cars. It becomes more obvious how serious the situation is as the number of
self-driving automobiles increases. We need a network that is both speedy
and responsive in this situation. Edge computing addresses a variety of net-
work issues, including bandwidth, latency, congestion or reliability.

• Bandwidth: In computing, bandwidth refers to the amount of data

that may be sent through a network at any one time. All networks
have a limit on the amount of bandwidth they may use, but wireless
networks have significantly tougher limits. This signifies that there is
a limit on the amount of data that can be sent over a network or the
number of devices that can send data over a network. Increased net-
work capacity can support more devices and data, even if doing so
comes with a price that do not provide a solution to other challenges.
• Latency: The amount of time it takes for data to be sent from one place
to another is referred to as network latency. Despite the notion that
data should be able to travel over networks at the velocity of light,
physical distances, network congestion, and network failures may
cause information to move more slowly than it should. When analytics
42 Parul Dubey et al.

and decision-making are postponed, the system’s ability to respond in

real time is reduced. In the instance of the autonomous automobile, it
even resulted in the loss of life.
• Congestion: The internet may be solely described as a worldwide “net-
work of networks” Many everyday computer tasks, such as file trans-
fers and rudimentary streaming, may benefit from general-­purpose
data exchanges via the internet. Because of the sheer number of devices
and data being sent over the internet, it’s possible that the network
may become overburdened, necessitating lengthy information retrans-
missions and significant levels of congestion. The Internet of Things is
rendered worthless if the network fails, since this results in even more
congestion and perhaps excludes certain internet users.

When servers and storage are placed near to the source of data creation,
edge computing reduces latency and congestion, enabling a much smaller
and cheaper LAN to support many more devices. Edge computing is becom-
ing more popular. When data is kept and safeguarded locally, edge servers
may perform critical edge analytics or, at the very least, pre-process and
limit the data, allowing decision-makers to make decisions in real time prior
to uploading findings or merely required data to the cloud or centralized
data centre. Figure 3.1 explains the working of cloud computing and edge
computing and shows the region where the computation occurs actually.

Figure 3.1 Working of cloud computing and edge computing.

 A review on edge computing 43

3.4 COMPARISON—EDGE COMPUTING, CLOUD

COMPUTING AND FOG COMPUTING

“Cloud computing” and “fog computing,” two terms that are often used
in conjunction with one another when referring to edge computing, are
occasionally used synonymously. However, although there are some simi-
larities between the two notions, they are not the same concept and cannot
be used frequently in the very same paragraph. Comparing and contrasting
the two notions to see how they vary from one another is an interesting
exercise:

• A emphasis on the real deployment of power and memory resources

in close proximity to the data that is being created are the common
traits that run across edge computing, cloud computing, and fog
computing, to mention a few of the technologies that are being used
today. Understanding the distinctions between these three points
of view is a simple issue of observation and comparison. The loca-
tion of resources has a significant influence on the efficacy of those
resources.
• Edge computing: In the computing world, “edge computing” refers to
the deployment of server and storage facilities close to the site of data
generation. In an ideal network, data storage and computing should be
positioned at the network’s edge. This may be achieved, for example,
by placing a small container on top of the turbine that contains many
processors and storage for the purpose of collecting and analyzing
data collected by sensors. The installation of a small amount of com-
puter and storage capacity to collect and analyze data from a broad
variety of railway and rail traffic sensor data is conceivable, for exam-
ple, at a railway station’s ticket counter. Following that, the results of
such processing may be transmitted to another data centre for human
inspection, archiving, and larger-scale analytics.
Following that, the results of such processing may be transmitted to
another data centre for human inspection, archiving, and larger-scale
analytics.
• Cloud computing: In addition, there are several places across the globe
(regions) where power and memory resources may be deployed, which
makes cloud computing incredibly scalable and readily accessible. Well
before the era of the Internet of Things, services from cloud service
providers were also making the cloud a much more enticing central
platform for Internet of Things deployments in the cloud, according to
Gartner. Cloud computing provides more than enough resources and
services to handle complex analytics, but the nearest geographic cloud
facility may be sited hundreds of miles away from the point where
data is collected, and connections rely on the same unreliable internet
connectivity that is used to support traditional data centers. Although
44 Parul Dubey et al.

cloud computing is not intended to be a replacement for traditional

data centers, it is often utilized in combination with these facilities.
Despite the fact that the cloud has the capacity to move centralized
processing closer to the source of data, it is hard for the cloud to reach
the network’s edges.
• Fog computing: Although there are many different options for deploy-
ing computer and storage resources, there seems to be no “one-size-
fits-all” approach. Where there is no access to a cloud data centre,
the edge implementation may be too resource-constrained and/or geo-
graphically scattered to be economically sustainable. An emerging idea
known as fog computing could be useful as a consequence.

Consequently, fog computing systems may generate massive amounts of sen-

sor or Internet of Things data that cannot be specified by an edge. Advanced
technologies, smart cities, and smart utility systems all offer examples of
how technology is being used. In a smart city, data may be used to monitor,
assess, and enhance public transit, municipal utility systems, and civic ser-
vices, among other things. Fog computing may operate numerous fog node
deployments inside the range of the environment in order to collect, analyze,
and interpret information across a big enough region.
Even among technology experts, the terms fog computing and edge com-
puting are often used interchangeably because of their almost identical defi-
nitions and designs, and the terms are sometimes used interchangeably when
referring to cloud computing.
Starting with the recognition that cloud computing and edge computing
are two unique concepts that cannot be swapped for one another, it is vital
to understand how they differ [8]. In essence, edge computing is used to
handle data that must be processed in a timely fashion, while cloud comput-
ing is used to process data which does not need immediate attention.
Figure 3.2 explains the architecture in detail.
Edge computing surpasses cloud computing in terms of latency when used
in remote places with inadequate or no connectivity to a centralized data
centre or server. Edge computing is the ideal solution for the internal storage
that is needed at these locations, which may be thought of as a small data
centre. Table 3.1 explains the comparison in detail.
Edge computing may also be beneficial to devices that are extremely spe-
cialized or intelligent. Despite the fact that these devices look and operate
in a similar way to PCs, they are not general-purpose desktops. Because
they are intelligent, specialized computers behave in a certain way when
they encounter certain other devices. Edge computing, by contrast, has a
downside in some industries where rapid responses are required for
success.
Edge computing may also be beneficial to devices that are extremely special-
ized or intelligent. Despite the fact that these devices look and operate in a
 A review on edge computing 45

Figure 3.2 Architecture of interaction between Edge, Fog and Cloud.

Table 3.1 Comparison of cloud, fog and edge computing

Parameter Cloud computing Fog computing Edge computing
Data Cloud server Inside a fog node or an Device itself
processing IoT gateway
site
Objective Long-term in-depth Real-time and immediate Real-time and
investigation immediate
Working Independent Fog works in integration It can be defined
along with the cloud without cloud
Latency High Low Very low
Security Best Better Good
Architecture Centralized with Hierarchical and flat Separate nodes
regions and architecture with
availability zones several layers that form
a network

similar way to PCs, they are not general-purpose desktops. Because they are
intelligent, specialized computers behave in a certain way when they encounter
certain other devices. Edge computing, on the other hand, has a downside in
some industries where rapid responses are required for success.
46 Parul Dubey et al.

3.5 THE BENEFITS OF EDGE COMPUTING

There are a number of significant benefits of this kind of computation that

may make the technology attractive in some situations, such as the reduc-
tion of bandwidth constraints, the elimination of excessive delay, and the
elimination of network congestion:

1. Autonomy: Edge computing is advantageous in circumstances where

the internet connection is intermittent, or when the bandwidth is lim-
ited. The oil rigs, ships on the high seas, and other remote locations are
all instances of this. Only when a network connection is available can
data be preserved for transmission to a central location. For example,
water quality sensors on filtration systems in rural villages may be
used to save data for transmission. A significant amount of bandwidth
or time is conserved by minimizing the amount of information that
has to be sent and by processing data on the local network.
2. Data sovereignty: Whenever it comes to transferring massive amounts
of data, the problem is more than only technical in nature. When trav-
elling beyond national and regional boundaries, data protection, con-
fidentiality, and other legal concerns may become more difficult to deal
with. Edge computing allows data to be retained close to the place of
origin while remaining within the constraints of current data sover-
eignty requirements, such as the General Data Protection Regulation
(GDPR), which specifies how data is kept, processed, and exposed.
Access to a local processing of raw data helps to protect private infor-
mation before it has been sent to the cloud or to a central data centre,
which may be placed in remote locations across the world.
3. Edge security: Finally, edge computing has the ability to make the pro-
cess of creating and maintaining data security more straightforward.
However, despite the fact that cloud providers provide Internet of
Things services and expertise in complicated evaluations, organiza-
tions are still concerned with the safety and protection of information
after it has left their network perimeter. In situations where Internet
of Things devices are prone to hacking and other destructive actions,
security may be addressed by using computing at the edge in order to
allow encryption and protecting the deployment itself against such
threats.

3.6 INSTANCES AND ILLUSTRATIONS OF EDGE COMPUTING

In theory, data may be processed and evaluated “in-place” at or close to

the network edge adopting edge computing approaches, which allow data
to be processed and examined in real time. A powerful way to make use of
data that cannot be transferred to a central location at the outset—usually
 A review on edge computing 47

due to the sheer volume of data making such migrations economically dif-
ficult, technically tricky, or otherwise infringing specifications, such as data
­sovereignty—is to store it in a decentralized location. There are several
examples of how this phrase has been used in the real world, including:

• Industrial production: In order to discover production flaws and

improve product quality, one industrial business deployed edge com-
puting to analyze manufacturing operations in real time, with the goal
of improving product quality. Environmental sensors were planted
throughout the manufacturing plant to evaluate how each manufac-
turing component is created and maintained, as well as how often
the components are kept in stock for each product. These sensors
were made possible by edge computing. This allows the corporation
to make decisions about its manufacturing plant and operations with
increased speed and accuracy as a consequence of the new information
technology.
• Farming: It could be possible to create an indoor farm that does not
need sunlight, soil, or pesticides. Such an approach could make time
savings of more than 60%. Sensors enable the firm to keep track of its
resource use, nutrition density, and harvesting efficiency, among other
things. Crop production algorithms are continually being updated
based on the data collected and assessed in order to ensure that crops
are harvested in the best possible condition.
• Optimization of the network: Edge computing may be used to help
enhance network performance by analyzing the performance of users
across the internet and then finding the most dependable, low-latency
network channel for each user’s data to be sent. The use of edge com-
puting for time-sensitive traffic has the effect of acting as a “steering”
technique for congestion.
• Safety at work: Edge computing, which makes use of on-site cameras,
worker safety systems, and other sensors, can assist organizations in
monitoring working conditions and ensuring that employees abide to
defined safety rules, particularly in remote or dangerous workplaces
such as construction sites or oil rigs.
• Better healthcare: A significant increase in the amount of patient
data collected by gadgets, sensors, and other hospital equipment has
occurred in the healthcare industry. Edge computing will have to use
automation and machine learning to obtain information, disregard
“normal” data, and detect issue data in order for doctors to take quick
action to help patients in preventing health crises. This will allow doc-
tors to take immediate actions to assist patients in preventing health
crises.
• Transportation: On a daily basis, autonomous vehicles consume and
create anywhere between 5 TB and 20 TB of data, gathering informa-
tion about their own location, speed, and other factors, as well as
48 Parul Dubey et al.

information about other drivers and vehicles on the road. It is critical

that all of this information be gathered and processed while the auto-
mobile is going forward. As each autonomous vehicle evolves into a
“edge,” this necessitates the use of a significant amount of onboard
computational capacity. According to the real-world situations,
authorities and companies may also benefit from the data in order to
better manage existing vehicle fleets and operations.
• Retail: Large volumes of data may be obtained via CCTV, stock moni-
toring, sales data, and other real-time business information, all of
which can be used by retail organizations. Edge computing may be
used to identify a successful elevated platform or campaign. It can also
be used to predict sales and optimize vendor purchasing decisions.
Because local retail environments might differ significantly, edge com-
puting can be a cost-effective choice for storing and processing at each
location.

3.7 EDGE MAINTENANCE

Edge implementation would be incomplete if it did not also take into con-
sideration edge maintenance:

• Technologies that are primarily concerned with the management of

risks as well as the preventive threats should be given serious consid-
eration. Because every device is a network component that may be
compromised, there are an overwhelming number of distinct threat
vectors for hackers to take advantage of.
• It is necessary to provide access to control and monitoring even if the
real data is not available, due to the fact that connectivity is another
issue. As a fallback for both connectivity and management reasons,
some edge installations make use of a supplementary link as part of
their design.
• Edge deployments need remote provision and management due to the
fact that they are located in inaccessible and usually hostile areas. The
ability to watch what is going on outside of their company and to
restrict deployments when necessary is essential for information tech-
nology (IT) management professionals.
• It’s hard to overstate the significance of physical care and mainte-
nance. The usage of Internet of Things (IoT) devices has been shown
to dramatically diminish their usable life when batteries and devices
are replaced on a regular basis. It is inevitable that equipment will
wear out and that they will need to be fixed or updated at some time.
Maintenance must also take into consideration the practical chal-
lenges associated with the site.
 A review on edge computing 49

3.8 OPPORTUNITIES FOR 5G, IoT WITH EDGE

COMPUTING

Edge computing is continually evolving and enhancing its functionality via

the application of modern techniques. It is expected that edge services would
be widely available by 2028, making it the most important development.
Edge computing, rather than being situation-specific, is projected to become
more prevalent and transform the way the internet is used in the future,
offering more abstraction and potential applications for edge technology.
Recent years have seen a rise in the number of edge computing-specific
processing, memory, and connectivity devices. In order to increase product
interoperability and flexibility, more multi-vendor coalitions are needed. To
illustrate this, Amazon Web Services (AWS) and Verizon have collaborated
to boost connection at the network’s edge.
Increasingly widespread use of new wireless systems, such as 5G and
Wi-Fi 6, is expected over the next few years, with a substantial influence on
edge installations and usage in the process. As a result of these technologies,
previously unimaginable virtualization and automation capabilities, such as
greater vehicle autonomy and workload migration to the edge, will be made
possible.
Edge computing has become more popular as a result of the growth of the
Internet of Things and the huge quantity of data it creates. Although edge
computing is still in its early stages, the proliferation of Internet of Things
devices will have an impact on its future development. Future ideas include
the production of micro modular data centres (MMDCs), which are small
data centres that can be moved around. As the name implies, it is a mobile
data centre that may be installed in strategic areas, such as around a town
or state, to bring computers much closer to the data that they are processing
without jeopardising the protection of data being processed.

3.9 CHALLENGES OF EDGE COMPUTING

Although edge computing has the potential to benefit a broad variety of

applications, the technology is not without its drawbacks. Furthermore, in
addition to conventional network limits, there are a number of significant
problems that may have an influence on the deployment of edge computing:

• Capacity is limited: The extensibility of cloud computing’s resources and

services are two of the factors that make edge computing, also known
as fog computing, so appealing. An edge architecture may be beneficial,
but only if the scope and purpose of the infrastructure are well specified.
Even a large-scale solution of edge computing performs a specific role
by using limited resources and providing a limited number of services.
50 Parul Dubey et al.

• Connectivity: Any edge deployment, no matter how tolerant, will need

some type of connectivity, even under the most permissive of circum-
stances. When developing an edge deployment, it is critical to con-
sider what happens if a connectivity is lost at the edge of the network.
The independence, artificial intelligence, and elegant failure planning
required for successful edge computing are all essential in the case of
connection challenges.
• Security: When building an edge computing deployment, it’s vital to
consider elements such as policy-driven configuration enforcement,
security in the computing and storage and confidentiality in both the
data at rest and in flight, with a particular emphasis on the latter. Even
if cloud-based IoT services enable secure communications, if you’re
constructing an edge location from the ground up, this isn’t normally
included in the service fee.
• Data lifecycles: There is a recurring problem with today’s data overload: a
significant amount of it is completely meaningless. Think about a medical
monitoring equipment, for example: only the problem data is important,
and preserving days of normal client records is not essential. Because it
depends primarily on short-term data that is not even stored for the long
haul, real-time analytics is inefficient. After completing an analysis, it is
up to the company to decide which information to preserve and which
information to reject. Consequently, companies and regulators must take
steps to guarantee that the information they possess is protected.

3.10 CONCLUSION

In the commercial world, edge computing has become increasingly preva-

lent. However, edge computing is not the only approach available to busi-
nesses. When faced with computing challenges, cloud computing is still a
decent option for IT vendors and organizations. They may choose to use
it in combination with edge computing to provide a more comprehensive
solution. In addition, the direction of all data to the edge is not a smart strat-
egy. Accordingly, public cloud providers have started to integrate Internet of
Things strategies and tools stacks with edge computing.
As we’ve discussed in this review, edge computing and cloud computing
are not regarded as interchangeable concepts. Indeed, many organizations
have embraced edge computing technology, drawn by its promise to over-
come some of the basic difficulties connected with cloud computing.

REFERENCES

1. W. Shi, J. Cao, Q. Zhang, Y. Li, and L. Xu, “Edge Computing: Vision and
Challenges,” IEEE Internet of Things Journal, vol. 3, no. 5, pp. 637–646,
October 2016.
 A review on edge computing 51

2. L. Wu, R. Zhang, Q. Li, C. Ma, and X. Shi, “A Mobile Edge Computing-Based

Applications Execution Framework for Internet of Vehicles,” Frontiers of
Computer Science, vol. 16, no. 5, January 2022.
3. L. Yuan et al., “CSEdge: Enabling Collaborative Edge Storage for Multi-Access
Edge Computing Based on Blockchain,” IEEE Transactions on Parallel and
Distributed Systems, vol. 33, no. 8, pp. 1873–1887, August 2022.
4. N. Abbas, Y. Zhang, A. Taherkordi, and T. Skeie, “Mobile Edge Computing: A
Survey,” IEEE Internet of Things Journal, vol. 5, no. 1, pp. 450–465, February
2018.
5. B. Varghese et al.,“Realizing Edge Marketplaces: Challenges and Opportunities,”
IEEE Cloud Computing, vol. 5, no. 6, pp. 9–20, November 2018.
6. S. Wang, Y. Zhao, J. Xu, J. Yuan, and C.-H. Hsu, “Edge Server Placement in
Mobile Edge Computing,” Journal of Parallel and Distributed Computing,
vol. 127, pp. 160–168, May 2019.
7. Stephen J. Bigelow, “What Is Edge Computing? Everything You Need to
Know,” SearchDataCenter, December1, 2021. https://www.techtarget.com/
searchdatacenter/definition/edge-computing
8. Shivam Arora, “Edge Computing Vs. Cloud Computing: Key Differences
[2022 Edition],” Simplilearn.com, January 16, 2018. https://www.simplilearn.
com/edge-computing-vs-cloud-computing-article
Chapter 4

Industrial Internet of Things

IoT and Industry 4.0
Asmita Singh Bisen and Himanshu Payal
Sharda University, Greater Noida, India

CONTENTS

4.1 Introduction .................................................................................. 53

4.2 Streamlined work .......................................................................... 54
4.3 Internet of Things .......................................................................... 54
4.4 The Industrial Internet of Things ................................................... 55
4.5 Industry 4.0 ................................................................................... 56
4.6 Key technologies for the Industrial Internet of Things ................... 56
4.6.1 Blockchain technology ........................................................ 56
4.6.2 Cloud computing ................................................................ 57
4.6.3 Artificial intelligence and cyber physical systems ................. 57
4.7 Intelligent manufacturing in the context of Industry 4.0 ................ 57
4.8 Open research issues ...................................................................... 58
4.9 Application domains ..................................................................... 60
4.9.1 Healthcare .......................................................................... 61
4.9.2 Smart cities ......................................................................... 61
4.9.3 Smart environments ............................................................ 61
4.9.4 Industry .............................................................................. 62
4.10 Challenges ..................................................................................... 62
4.11 Conclusion .................................................................................... 62
References ............................................................................................... 63

4.1 INTRODUCTION

Today, as a result of rapidly changing technological advances, smart devices

can connect to each other. In addition, over the years, the capacity of these
devices has decreased, and their depurational capacity and repertory resources
have increased. Modern smart devices are bedecked with embedded systems
and have the ability to communicate, detect, activate and search, collect,
cached and process data in real time. The Internet of Things (IoT) is an inno-
vative and fast-growing technology with a variety of applications, features
and services. The IoT allows us to penetrate our required environment and

DOI: 10.1201/9781003407300-4 53
54 Asmita Singh Bisen and Himanshu Payal

its goals, connect the physical world with the digital world, and “connect
people and devices anytime, anywhere, with anything and everyone” [1, 2].
The digitization and automation of industrial production combines ancestral
manufacturing and industrial practices with large-scale machine-to-machine
communications, IoT, Cyber-Physical Systems (CPS) and many other innova-
tive technologies [3].

4.2 STREAMLINED WORK

Many sophisticated studies have been conducted presenting and analyzing

the concepts, characteristics, applications, and key implementation tech-
nologies of IoT. They reviewed various IoT phenomena and paradigms and
discussed key processing technologies. In addition, the domains of the IoT
application are analyzed and grouped in the areas of transport and logistics,
health, personal and social, smart environment and futures [4, 5].
In his research [6], he​introduced a concentrated “cloud-based” vision for
IoT organizations around the world. They likewise examined an IoT out-
line, as well as key working advances and application areas, and some IoT
patterns and scientific categorizations. Moreover, they defeated clear diffi-
culties and future patterns in the IoT business and introduced an informa-
tion examination and contextual analysis on the Aneka/Azure cloud stage.
Finally, the requirement for combination of the remote sensor organization,
the Internet and disseminated registering was featured [7, 8].
In an effort to clarify the need for high-end IoT hardware, [9] conduct
research on existing IoT media opportunities, challenges and supporting
technologies. Advanced centralized solutions for implementing IoT applica-
tions are provided, as well as a complete analysis of problems and support
technologies in IoT-centric software development. IoT is a dynamic and
global ubiquitous computing methodology which enables the quality of life
and ease of end user.

4.3 INTERNET OF THINGS

The Internet of Things (IoT) is one of the most widely used terms in the
development of modern computing. Although often referred to as technol-
ogy, it is more accurately described as a platform for integrating objects
(through recognition) so that the information collected about them can be
used to analyze, interpret, define and execute this and other related informa-
tion [10, 11].
The term Industry 4.0 comes from an undertaking in the German govern-
ment’s cutting-edge strategy. Such an undertaking upholds robotization.
This depends on innovative ideas for the Internet of Things (IoT), which
empowers the production line representing things to come [12, 13].
 Internet of Things in modern computing 55

Since then, companies have presented solutions to this concept, with the
help of governments, especially Europe (especially Germany), but also coun-
tries like the United States, Japan and China, indicating that this was an
industrial and strategic era.
In this, field gear, machines, other items comprise a digital actual frame-
work that independently trades data, starts activities and drives each other
freely. Changing production lines into a brilliant climate that overcomes any
issues between this present reality and the computerized world [14]. The
solid tendency of the electrical and various leveled universes of plant mecha-
nization will move toward savvy industrial facility networks that empower
dynamic re-designing cycles and the capacity to respond deftly to interrup-
tions and disappointments.
The vital parts of this is that it can be characterized by three ideal mod-
els: a brilliant item, a savvy machine and high-level administrator. The
directing thought behind the Smart Product is to make the section a func-
tioning piece of the framework. Items get a memory where organization
information and prerequisites are put away straightforwardly as isolated
building plans. The smart machine worldview portrays the cycle by which
machines become Cyber-Physical Manufacturing Systems (CPMS).
Expanded Operator, the third worldview referenced above, centers around
the mechanical help of laborers in a perplexing climate of profoundly mea-
sured creation frameworks.

4.4 THE INDUSTRIAL INTERNET OF THINGS

Modern IIoT is a particular kind of IoT that spotlights applications in the

cutting-edge industry and savvy production. With regards to Industry 4.0,
IIoT should be visible as a complicated framework with numerous frame-
works and gadgets. Specifically, IIoT consolidates a few significant cutting-
edge innovations to make a framework that works more productively than
a simple combination of its parts [15, 16].
With authentic assets, network innovation, applications, sensors, pro-
gramming, media and storehouses, IIoT conveys arrangements and capaci-
ties that empower and control the business cycles and resources. IIoT
administrations and applications give basic answers for better preparation,
booking and the board of items and frameworks [17, 18].
Besides, through the different associated gadgets that can and do speak
with one another and with additional brought together regulators, IIoT will
decentralize investigation and navigation, generating ongoing input and
reactions [19]. Therefore, both enterprises’ general accessibility and support-
ability and their presentation have improved, creation has been accelerated,
item time-to-showcase is more limited at low rates, and their general exhibi-
tion is enhanced, arriving at huge capacities at a level up until recently never
accomplished financial development and efficiency effectiveness [20, 21].
56 Asmita Singh Bisen and Himanshu Payal

4.5 INDUSTRY 4.0

In recent years, the global business environment has changed dramatically

due to the advent of technology, innovation and innovation. The fourth
variant (Industry 4.0) is its newest application and function, multiple inte-
grated devices, as well as its new product [22].
It is controlled by the accompanying four innovation groups:

1. PC information, power and availability,

2. examination and knowledge,
3. human–machine cooperation, and
4. computerized-to-actual transformation [23].

Likewise it joins the force of conventional enterprises with trend-setting

innovations that empower savvy items to be incorporated into interweaved
advanced and actual cycles. These cycles are associated with one another
[24, 25].
The primary plan standards for different parts of Business 4.0 are:

1. reconciliation,
2. virtualization,
3. decentralization,
4. in the nick of time stacking,
5. administration accessibility [26].

Industry 4.0 addresses the change from “concentrated” to “decentralized”.

Because of the circulation of knowledge, clever gadgets are created in free
cycle control frameworks. A significant new part of endlessly fabricating
processes is the mix of the genuine and virtual universes, where fabricating
hardware both produces items and communicates with them. Accordingly,
ventures, creation expenses, and plans of action become more astute,
prompting the development of “savvy” industrial facilities [27].

4.6 KEY TECHNOLOGIES FOR THE INDUSTRIAL INTERNET

OF THINGS

IIoT support is intangible and combines a wealth of epic editing, including

IoT computing, comprehensive evaluation information, artifact informa-
tion, real-world structural mechanisms, and extended operations.

4.6.1 Blockchain technology
Considering the related handling plant circumstance, Internet of Things con-
traptions help steady data collection and incitation. As the most fundamental
 Internet of Things in modern computing 57

part of IIoT, these contraptions track the creation line assets across the globe
[28]. The devices in a totally related IIoT framework are sent across all the
assembling plant workplaces, going from stockrooms to creation workplaces
and dispersion focuses. Regardless, the configuration, sending, checking, and
upkeep of these contraptions is a troublesome errand and require extraordi-
narily qualified specific staff.

4.6.2 Cloud computing
The dramatic development of information in IIoT requires a system that
requires a first-order system to manage, manage, display, and store informa-
tion. A distributed distribution system provides statistics, membership, and
control over all online jobs and production systems. All content and acces-
sories attached to it are attached to the remains. Cloud sites are arranged as
private, public (only security for cloud-based clients) or ivory (using both
types of auxiliaries) [29]. Since the underpinning of waiter ranches and the
enlistment of specific staff require high spending thusly confidential cloud
organization models are not a reasonable decision for as of late members
as well as nearly nothing and medium-level undertakings. However, large
and well-established overall undertakings favor the sending of private fogs
to ensure the wellbeing, security, and assurance and adjust to current covert
work for high ground.

4.6.3 Artificial intelligence and cyber physical systems

Man-made information advances guarantee that the IIoT construction
ought to run uninhibitedly and magnificently to limit the thoughtful inter-
cessions and further encourage efficiency. The AI movements make IIoT free
by utilizing complex AI advancements, for instance, multi-master frame-
works and conversational AI [30]. In addition, the data is implanted at lay-
ers in the Industrial Internet of Things structures from sensors to gadgets to
edge servers and cloud server farms by empowering different chase, optimi-
zation, and guess calculations. To confine human undertakings and media-
tions, IIoT structures attract different advanced certifiable frameworks like
gathering designs and present-day robots. The substance of computerized
genuine structures lies in locally accessible embedded IoT contraptions
which empower different sensors and actuators to work in the state of the
art conditions.

4.7 INTELLIGENT MANUFACTURING IN THE CONTEXT

OF INDUSTRY 4.0

Fabricating involves a central modern part which fundamentally affects

individuals’ vocation and a country’s economy. Fully intent on upgrading
58 Asmita Singh Bisen and Himanshu Payal

the general creation, efficiency and item-quality administration all through

the different phases of the lifecycle of items, IoT offers applications and
administrations which incorporate high-level observing and following, exe-
cution and practicality enhancement and human–machine connection [31].
Subsequently, it makes sense that IoT can give a ton of answers for the
assembling space which is described by its intricacy and expansiveness of
utilizations, its different Cyber Physical Systems and its assembling activity
and the board philosophies.
Clever assembling utilizes the consolidated knowledge of individuals,
cycles and machines in order to increment creation. It offers remarkable
answers for the location and checking of possible harm, glitches and break-
downs. Additionally, it upgrades control and the executives, further devel-
ops practicality and accessibility and streamlines asset the board and sharing.
Moreover, it applies state-of-the-art advancements to different customary
frameworks, administrations and items [32]. Thus, clearly keen assembling
definitely affects the general capacity and financial aspects of undertakings
and will make ready for the headway of current ventures [33].
Shrewd assembling targets growing continuous, independent and human-
like insightful dynamic frameworks that decrease the requirement for human
contribution and intercession. In order to achieve this, man-made brain-
power, AI, hereditary calculations and other cutting-edge innovations,
approaches and methods are utilized. This reality contains a significant dis-
tinctive element between canny assembling and customary assembling [34].
In any case, the objective of both assembling spaces continues as before, or,
at least, to fulfill clients’ prerequisites and market needs as well as amplify-
ing benefits while at the same time limiting conceivable expense and waste.

4.8 OPEN RESEARCH ISSUES

Clearly with regards to the use of these can upgrade and change the ongoing
enterprises and yield a great deal of advantages because of its cutting-edge
innovations, applications and administrations. It is likewise fundamental to
highlight that IoT does not focus solely on changing businesses and expand-
ing their efficiency. In addition, it enhances the central motivation behind
ventures and relieving the shortcomings brought about by inheritance
frameworks. Thus, it ought to be viable with existing gadgets, frameworks
and foundations and have the option to implant knowledge into them [35].
Accordingly, undertakings that are going through advanced change will be
worked with to take on and execute IoT and take advantage of its vari-
ous advantages and arrangements without having to straightforwardly put
resources into absolutely spic and span hardware as cost would far offset
the prompt advantages. In any case, for this to be completely carried out
and for IoT to be embraced and completely used by businesses and under-
takings, a great deal of difficulties and open issues ought to be investigated.
 Internet of Things in modern computing 59

Many elaborate investigations, which break down indispensable IoT dif-

ficulties, coordination and execution issues and open exploration issues,
have been directed. All the more explicitly, open exploration issues, includ-
ing normalization exercises tending to and organizing as well as security and
protection were examined by [36]. Key IoT challenges, for example, interop-
erability and normalization, information and data secrecy, encryption and
protection, the naming and character of the board, IoT greening as well as
article and organization security were depicted [37].
Correspondence and recognizable proof advancements, dispersed frame-
work innovations and insight and accentuated security issues, for example,
information secrecy, protection and trust were aspects of the primary exami-
nation challenge that were investigated [38]. Information the executives and
mining, security and protection were viewed as the fundamental difficulties
which undertakings face in IoT advancement zeroed in on security and pro-
tection moves in IIoT and their weakness to an assortment of cyberattacks
[39, 40].
Many elaborate investigations, with respect to imperative difficulties,
combination and execution issues and the open exploration issues of
Industry 4.0, have been done. These have identified a few difficulties and
principal issues in different segments that happen all through the execution
of the advancements of the IoT [41, 42].
The recognized areas in their examinations were:

1. Shrewd direction and exchange component,

2. Fast modern remote organization conventions, fabricating explicit
large information and its scientific,
3. Framework demonstrating and investigation,
4. Digital security and
5. Modularized and adaptable actual ancient rarities,
6. Investment issues.

A few producers and endeavors wonder whether or not to proceed like this
because of specific worries and obstructions. These incorporate vulnerabili-
ties about monetary advantages, the absence of systems of planning across
various authoritative units, missing ability, abilities and capacities, wavering
to go through extremist change and concerns with respect to the outsider
suppliers’ security [43, 44].
To summarize, in the light of the previously mentioned examinations, the
most huge and normal difficulties and open exploration issues which busi-
nesses and undertakings ought to know about are:

a. Availability, unwavering quality, portability and other QoS models;

b. Security, protection and privacy of information;
c. Interoperability and versatility;
d. Fault resilience and usefulness for wellbeing;
60 Asmita Singh Bisen and Himanshu Payal

e. Management of tasks, assets, energy and information;

f. Networking tending to and ID;
g. Architecture, conventions and normalization exercises
h. Trouble in planning activities across various hierarchical units, like
exploration and improvement such as Research and Development
(R&D), IT, assembling, deals, and money divisions, because of unfor-
tunate cooperation between them; [45]
i. Uncertainty about adopting as opposed to rethinking and the absence
of information about specialist co-ops;
j. Concerns about online protection while including outsider innova-
tion/programming and execution suppliers;
k. Concerns about information proprietorship while working with out-
sider suppliers;
l. Challenges with coordinating information from unique sources to
empower Industry 4.0 applications.

4.9 APPLICATION DOMAINS

Albeit the new advances in universal registering and the possibilities pre-
sented by the IoT render the improvement of countless applications practi-
cal, advances are already visible in a couple of areas [46]. IoT applications
target further developing quality for the end-client local area and supporting
framework and universally useful tasks [47].
The intricacy and the size of the issue to be addressed, as well as the par-
ticularities, prerequisites and attributes of the particular areas in which they
will be carried out, ought to be thought about as there is no “one-size-fits-
all” arrangement. Also, they ought to be planned circumspectly in order to
fulfill different goals and prerequisites while at the same time expanding the
nature of involvement and nature of administration levels [48–50]:

i. Ecological space: This includes applications that secure, screen and

foster every single regular asset, natural administration administra-
tions, energy the board, reusing, farming and so on
ii. Industrial area: Applications of this area include monetary or business
exchanges between ventures, associations and different elements. Also,
they allude to assembling, planned operations, banking, monetary leg-
islative specialists and so forth
iii. Social area: This includes applications with respect to the turn of
events and the incorporation of social orders, urban communities, and
individuals as well as administrative administrations toward residents
and other society structures [51].

There are a variety of applications, including the following:

 Internet of Things in modern computing 61

4.9.1 Healthcare
The clinical area is quite possibly the earliest business to have adopted the
IoT and in which it can be seen to have had the most significant impact.
Advances in this area can open new doors, administrations and applica-
tions to work on the medical care and clean area. In addition, IoT stages
and administrations improve current living arrangements and work with the
acknowledgment of ubiquitous medical services vision that is “medical ser-
vices to anybody, whenever, and anyplace by eliminating area, time and dif-
ferent restrictions while expanding both the inclusion and nature of medical
care” [52]. This reality will improve and computerize the method involved
with gathering information, thereby producing a phenomenal amount of
information that can be used to pursue additional logical and clinical inves-
tigations. With the end goal of forestalling the beginning of medical issues
all the more proficiently, IoT advancements work on persistent therapy and
prosperity.

4.9.2 Smart cities
Rapid metropolitan development is now overwhelming the current foun-
dation and utilities and featuring the requirement for more manageable
metropolitan preparation and public administrations. The applications and
administrations are being taken advantage of for these new necessities to be
fulfilled and the cultural changes proportionate with this rapid development
to be answered. In addition, autonomous city and home organizations will
be keen and fit for detecting, observing and adjusting to ecological pliancy
as well as responding to human exercises. All the more explicitly, shrewd
innovations and gadgets are interconnected and, subsequently, they can
improve and upgrade the quality and way of life for city tenants as well as
guarantee that their fundamental administrations are given [53]. Moreover,
by making shrewd urban communities that mix the generally settled city
administrations and utilities that occupants cooperate with consistently,
advancing the use of city foundation, assets and offices and upgrading city
inhabitants’ life quality.

4.9.3 Smart environments
By using completely interconnected mechanical gadgets and implanted
frameworks, IoT targets invade our regular climate and its items and make
better approaches to associate with these brilliant conditions [54].
Through the involvement of IoT in a blend with mechanized programming
specialists for ongoing following and observing, savvy conditions become a
mechanical environment of different interconnected gadgets. These brilliant
gadgets can safely convey and cooperate as well as recover, cycle, store and
trade information on a continuous basis. By incorporating these heterogeneous
62 Asmita Singh Bisen and Himanshu Payal

information into applications, the variation cycle to inhabitants’ and natural

persistently changing requirements is worked with. Thus, overall, their prereq-
uisites are speedily and sufficiently met. In addition, IoT applications in this
area target working on the ongoing natural wellbeing by lessening and reliev-
ing the likely effect of harm and catastrophe. IoT advancements permit the
improvement of creative constant checking and dynamic emotionally support-
ive networks and applications with regard to ecological issues, for example,
early expectation and recognition of cataclysmic events, weather patterns
and so on.

4.9.4 Industry
A particular classification of IoT centers around its cases in current ventures
and savvy production. Being an intricate arrangement of a wide assortment
of systems is thought of. Besides, it includes a critical part to modern space
and is firmly connected with the fourth modern transformation (Industry
4.0). It joins a few creative key innovations in order to deliver a framework
that capacities more successfully than the amount of its parts.

4.10 CHALLENGES

The variegated and labyrinthine nature of its structure has faced numerous
challenges, in areas such as safety and security, adaptability, heterogeneity
and the quality of real tools.
To cope with these difficulties, efficient information the executives models
are required. These information the board models ought to be fit for effi-
ciently taking care of the tremendous measure of crude information gener-
ated by gadgets [55, 56]. These models should in like manner give the data
to the board administrations with fast data taking care of, trustworthy and
secure data accumulating, recuperation and speedy data flow.

4.11 CONCLUSION

IoT is an inventive and rapidly developing innovation which offers differ-

ent novel applications, administrations and arrangements and connects
the physical to the advanced world. In addition, it targets changing the
ongoing businesses into shrewd ones using the powerful organization of
interconnected gadgets. Improving their activity and usefulness, expand-
ing their efficiency and lessening their expenses and waste are among the
many advantages and benefits that undertakings can acquire by utilizing
IoT. Further, ventures that completely embrace IoT will be in front of their
rivals, become more agile, adjust to the ceaselessly evolving market, make
results greater that fulfill clients’ necessities and prerequisites.
 Internet of Things in modern computing 63

Additionally, with regards to Industry 4.0, IoT, and all the more explicitly
IIoT, can be used in blend with other creative advances, for example, huge
information, distributed computing, CPSs and so on to improve and change
the ongoing assembling frameworks into smart ones. It considers machines
to become autonomous substances that can gather and break down infor-
mation and offer guidance upon it without requiring any human interces-
sion as it presents self-viability, self-advancement, and self-perception. It
tries to adapt effectively to the worldwide serious nature of the present busi-
ness sectors and enterprises in accordance with the clients’ steadily changing
necessities and prerequisites.
In spite of the fact that IoT offers an abundance of answers for ventures as
well as a large number of contemporary and high-level applications and
administrations, it is currently at an early phase of improvement, reception
and execution. Accordingly, for the different momentum moves and open
issues to be experienced and settled, further exploration ought to be done. All
things considered, the total execution and brief reception of IoT alongside
fitting usage of its original advances, applications and administrations cannot
further develop life quality; however, it can likewise yield huge individual,
proficient and financial open doors and advantages sooner rather than later.

REFERENCES

1. Perera, C., Liu, C. H., & Jayawardena, S. (2015). The emerging internet of
things marketplace from an industrial perspective: A survey. IEEE Transactions
on Emerging Topics in Computing, 3(4), pp. 585–598.
2. Vermesan, O., Friess, P., Guillemin, P., Gusmeroli, S., Sundmaeker, H., Bassi, A.,
… & Doody, P.(2011). Internet of things strategic research roadmap. Internet
of Things-Global Technological and Societal Trends, 1(2011), pp. 9–52.
3. Li, J., Huang, Z., & Wang, X. (2011, May). Notice of Retraction Countermeasure
research about developing Internet of Things economy: A case of Hangzhou
city. In 2011 International Conference on E-Business and E-Government
(ICEE), pp. 1–5. IEEE.
4. Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey.
Computer Networks, 54(15), pp. 2787–2805.
5. Sundmaeker, H., Guillemin, P., Friess, P., & Woelfflé, S. (2010). Vision and
challenges for realising the internet of things. Cluster of European Research
Projects on the Internet of Things, European Commision, 3(3), pp. 34–36.
6. Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of
things (IoT): A vision, architectural elements, and future directions. Future
Generation Computer Systems, 29(7), pp. 1645–1660.
7. Al-Fuqaha, A., Guizani, M., Mohammadi, M., Aledhari, M., & Ayyash, M.
(2015). Internet of things: A survey on enabling technologies, protocols, and
applications. IEEE Communications Surveys & Tutorials, 17(4), pp. 2347–2376.
8. Lampropoulos, G., Siakas, K., & Anastasiadis, T. (2019). Internet of things in the
context of industry 4.0: An overview. International Journal of Entrepreneurial
Knowledge, 7(1), pp. 4–19. doi: 10.2478/ijek-2019-000.
64 Asmita Singh Bisen and Himanshu Payal

9. Ngu, A. H., Gutierrez, M., Metsis, V., Nepal, S., & Sheng, Q. Z. (2016). IoT
middleware: A survey on issues and enabling technologies. IEEE Internet of
Things Journal, 4(1), pp. 1–20.
10. Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey.
Computer Networks, 54(15), pp. 2787–2805.
11. Vermesan, O., Friess, P., Guillemin, P., Gusmeroli, S., Sundmaeker, H., Bassi, A.,
… & Doody, P.(2011). Internet of things strategic research roadmap. Internet
of Things-Global Technological and Societal Trends, 1(2011), pp. 9–52.
12. Lee, J., Bagheri, B., & Kao, H. A. (2015). A cyber-physical systems architec-
ture for industry 4.0-based manufacturing systems. Manufacturing Letters, 3,
pp. 18–23.
13. Haddara, M., & Elragal, A. (2015). The Readiness of ERP systems for the fac-
tory of the future. Procedia Computer Science, 64, pp. 721–728.
14. Weyer, S., Schmitt, M., Ohmer, M., & Gorecky, D. (2015). Towards Industry
4.0-Standardization as the crucial challenge for highly modular, multi-vendor
production systems. IFAC-Papers on Line, 48(3), pp. 579–584.
15. Lu, Y. (2017). Industry 4.0: A survey on technologies, applications and open
research issues. Journal of Industrial Information Integration, 6, pp. 1–10.
16. Lampropoulos, G., Siakas, K., & Anastasiadis, T. (2018). Internet of things
(IoT) in industry: Contemporary application domains, innovative technologies
and intelligent manufacturing. International Journal of Advances in Scientific
Research and Engineering, 4(10), pp. 109–118.
17. Gilchrist, A. (2016). Industry 4.0: The industrial internet of things. New York,
NY: Apress. ISBN: 1484220463.
18. Bi, Z., Da Xu, L., & Wang, C. (2014). Internet of things for enterprise systems
of modern manufacturing. IEEE Transactions on Industrial Informatics, 10(2),
pp. 1537–1546.
19. Şen, K. Ö., Durakbasa, M. N., Baysal, M. V., Şen, G., & Baş, G. (2018). Smart
factories: A review of situation, and recommendations to accelerate the evo-
lution process. In The International Symposium for Production Research,
pp. 464–479. Springer, Cham.
20. Schmidt, R., Möhring, M., Härting, R. C., Reichstein, C., Neumaier, P., &
Jozinović, P. (2015, June). Industry 4.0-potentials for creating smart prod-
ucts: empirical research results. In International Conference on Business
Information Systems, pp. 16–27. Springer, Cham.
21. Zhou, K., Liu, T., & Zhou, L. (2015). Industry 4.0: Towards future industrial
opportunities and challenges. In 2015 12th International Conference on Fuzzy
Systems and Knowledge Discovery (FSKD), pp. 2147–2152. IEEE.
22. Lampropoulos, G., Siakas, K., & Anastasiadis, T. (2018). Internet of things
(IoT) in industry: Contemporary application domains, innovative technologies
and intelligent manufacturing. International Journal of Advances in Scientific
Research and Engineering, 4(10), pp. 109–118.
23. Wee, D., Kelly, R., Cattel, J., & Breunig, M. (2015). Industry 4.0-how to navi-
gate digitization of the manufacturing sector. McKinsey & Company, 58.
24. Schmidt, R., Möhring, M., Härting, R. C., Reichstein, C., Neumaier, P., &
Jozinović, P. (2015, June). Industry 4.0-potentials for creating smart prod-
ucts: empirical research results. In International Conference on Business
Information Systems, pp. 16–27. Springer, Cham.
 Internet of Things in modern computing 65

25. Geissbauer, R., Vedso, J., & Schrauf, S. (2016). Industry 4.0: Building the digital
enterprise. Retrieved from: https://www.pwc.com/gx/en/industries/industries-
4.0/landing-page/industry-4.0-building-your-digital-enterprise-april-2016.pdf
26. Hermann, M., Pentek, T., & Otto, B. (2016, January). Design principles for
industrie 4.0 scenarios. In 2016 49th Hawaii international conference on sys-
tem sciences (HICSS), pp. 3928–3937. IEEE.
27. MacDougall, W. (2014). Industrie 4.0: Smart manufacturing for the future.
Germany Trade & Invest.
28. D. Miller, Blockchain and the internet of things in the industrial sector, IT
Professional (2018) 15–18 (2018).
29. Khan, W. Z., Rehman, M. H., Zangoti, H. M., Afzal, M. K., Armi, N., Salah,
K. (2020). Industrial Internet of things: recent advances, enabling technologies
and open challenges. Computers & Electrical Engineering 81, p. 106522.
30. Zhong, R. Y., Xu, X., Klotz, E., & Newman, S. T. (2017). Intelligent manufac-
turing in the context of industry 4.0: A review. Engineering, 3(5), pp. 616–630.
31. Davis, J., Edgar, T., Porter, J., Bernaden, J., & Sarli, M. (2012). Smart manu-
facturing, manufacturing intelligence and demand-dynamic performance.
Computers & Chemical Engineering, 47, pp. 145–156.
32. Francis, D., & Bessant, J. (2005). Targeting innovation and implications for
capability development. Technovation, 25(3), pp. 171–183.
33. Parwez, M. S., Rawat, D. B., & Garuba, M. (2017). Big data analytics for user-
activity analysis and user-anomaly detection in mobile wireless network. IEEE
Transactions on Industrial Informatics, 13(4), pp. 2058–2065.
34. Kumar, S. (2002). Intelligent Manufacturing Systems. B.I.T. Mesra, Ranchi,
India. Retrieved from https://pdfs.semanticscholar.org/fc60/02a47a1a9b312
bda871b47569edeeb8545f9.pdf
35. Küsters, D., Praß, N., & Gloy, Y. S. (2017). Textile learning factory 4.0–prepar-
ing Germany’s textile industry for the digital future. Procedia Manufacturing,
9, pp. 214–221.
36. Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey.
Computer Networks, 54(15), pp. 2787–2805.
37. Khan, R., Khan, S. U., Zaheer, R., & Khan, S. (2012, December). Future inter-
net: the internet of things architecture, possible applications and key challenges.
In 2012 10th international conference on frontiers of information technology,
pp. 257–260. IEEE.
38. Miorandi, D., Sicari, S., De Pellegrini, F., & Chlamtac, I. (2012). Internet of
things: Vision, applications and research challenges. Ad Hoc Networks, 10(7),
pp. 1497–1516.
39. Lee, I., & Lee, K. (2015). The Internet of Things (IoT): Applications, invest-
ments, and challenges for enterprises. Business Horizons, 58(4), pp. 431–440.
40. Sadeghi, A. R., Wachsmann, C., & Waidner, M. (2015, June). Security and pri-
vacy challenges in industrial internet of things. In 2015 52nd ACM/EDAC/
IEEE Design Automation Conference (DAC), pp. 1–6. IEEE.
41. Wang, S., Wan, J., Li, D., & Zhang, C. (2016). Implementing smart factory
of industrie 4.0: an outlook. International Journal of Distributed Sensor
Networks, 12(1), 3159805.
42. Vaidya, S., Ambad, P., & Bhosle, S. (2018). Industry 4.0 – a glimpse. Procedia
Manufacturing, 20, pp. 233–238.
66 Asmita Singh Bisen and Himanshu Payal

43. Lu, Y. (2017). Industry 4.0: A survey on technologies, applications and open
research issues. Journal of Industrial Information Integration, 6, pp. 1–10.
44. Bauer, H., Baur, C., Mohr, D., Tschiesner, A., Weskamp, T., Alicke, K., & Wee,
D. (2016). Industry 4.0 after the initial hype–Where manufacturers are finding
value and how they can best capture it. McKinsey Digital.
45. Küsters, D., Praß, N., & Gloy, Y. S. (2017). Textile learning factory 4.0 – prepar-
ing Germany’s Textile industry for the digital future. Procedia Manufacturing,
9, pp. 214–221.
46. Atzori, L., Iera, A., & Morabito, G. (2010). The internet of things: A survey.
Computer Networks, 54(15), pp. 2787–2805.
47. Akpakwu, G. A., Silva, B. J., Hancke, G. P., & Abu-Mahfouz, A. M. (2017).
A survey on 5G networks for the internet of things: Communication technolo-
gies and challenges. IEEE Access, 6, pp. 3619–3647.
48. Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of
things (IoT): A vision, architectural elements, and future directions. Future
Generation Computer Systems, 29(7), pp. 1645–1660.
49. Hajrizi, E. (2016). Smart solution for smart factory. IFAC-PapersOnLine,
49(29), pp. 1–5.
50. Sundmaeker, H., Guillemin, P., Friess, P., & Woelfflé, S. (2010). Vision and
challenges for realising the internet of things. Cluster of European Research
Projects on the Internet of Things, European Commision, 3(3), pp. 34–36.
51. Miorandi, D., Sicari, S., De Pellegrini, F., & Chlamtac, I. (2012). Internet of
things: Vision, applications and research challenges. Ad Hoc Networks, 10(7),
pp. 1497–1516.
52. Lee, C. P., & Shim, J. P. (2009). Ubiquitous healthcare: Radio frequency iden-
tification (RFID) in hospitals. In Handbook of research on distributed medical
informatics and e-health, pp. 273–281. IGI Global.
53. Zanella, A., Bui, N., Castellani, A., Vangelista, L., & Zorzi, M. (2014). Internet
of things for smart cities. IEEE Internet of Things Journal, 1(1), pp. 22–32.
54. Weiser, M., Gold, R., & Brown, J. S. (1999). The origins of ubiquitous com-
puting research at PARC in the late 1980s. IBM Systems Journal, 38(4),
pp. 693–696.
55. Javed, F., Afzal, M. K., Sharif, M., & Kim, B.-S. (2018). Internet of things (iot)
operating systems support, networking technologies, applications, and chal-
lenges: A comparative review, IEEE Communications Surveys & Tutorials,
20(3), pp. 2062–2100.
56. Zikria, Y. B., Kim, S. W., Hahm, O., Afzal, M. K., & Aalsalem, M. Y. (2019).
Internet of things (iot) operating systems management: Opportunities, chal-
lenges, and solution, Sensors, pp. 1–10.
Chapter 5

Denial of Service Attacks

in the Internet of Things
Aditi Paul
Banasthali Vidyapith, Rajasthan, India

S. Sinha
CHRIST(Deemed to be University), Bangaluru, India

CONTENTS

5.1 Introduction .................................................................................. 67

5.2 IoT architecture ............................................................................. 69
5.3 Categorization of security issues .................................................... 72
5.4 Security issues of the lower layers .................................................. 72
5.5 Security issues of the intermediate levels ........................................ 74
5.6 Security issues of the higher levels ................................................. 74
5.7 Mitigation strategies at various levels of IoT devices ..................... 75
5.8 Distributed Denial of Service (DDoS) attack .................................. 77
5.9 DDoS attack mitigation strategies ................................................. 78
5.10 Research gap ................................................................................. 80
5.11 Fuzzy-Neural Network-based cross-layer DoS attack detection
framework ..................................................................................... 80
5.12 Conclusion .................................................................................... 86
5.13 Future work .................................................................................. 86
References ............................................................................................... 86

5.1 INTRODUCTION

The maiden years of the Internet of Things (IoT) mainly included commu-
nications between machines, also known as machine-to-machine (M2M)
associations [1]. However, over time this concept has developed to incor-
porate the involvement of humans, so much so that our dependence on its
services [2] continues to grow with time. Today our earth is packed with
gazillions of computing devices and sensors that perpetually sense, pick up,
gather, aggregate, and scrutinize the personal information of most of its
population. This information encompasses our address and whereabouts,
contacts list, browsing habits patterns, and sensitive and detailed attributes
about our health [3]. Hearing, collecting, and spreading such intimate per-
sonal data on computer devices is primarily promoted in the aspiration of

DOI: 10.1201/9781003407300-5 67
68 Aditi Paul and S. Sinha

comfort; the idea is that more ingenious devices can better comply with
our needs and desires and even circumstances (e.g., a thermostat senses
its environment and alters the temperatures taking a cue from our loca-
tion, what time of the day or what day of the week or what time of year
it is, etc.). These also tackle crises such as fire or burglary. Unfortunately,
this level of installation increases security costs. Also, it imposes specific
challenges [4] concerning privacy: personal, confidential information, if
available to ill-intended agents, can havoc damage our property, dignity,
and our safety.
The relevance of computing devices having such disastrous inadequacies
is no longer limited to research papers and academic studies. Still, it is, in
fact, a sad reality with a plethora of news snippets about multiple devices
succumbing to daily security breaches. And the fact that these attacks are
not so difficult to coordinate does not help the case. There have been many
demonstrations of such invasions. The attacker can inject [5] malicious code
directly into a wearable device through an editing interface and then capture
sensitive user data. Attacks on critical medical equipment, for instance,
Implantable Cardiac Defibrillators (ICD) and wireless monitors, can be life-
threatening for the patient. The industrial and urban framework has also
seen an increase in recent attacks. As many of today’s premium automobiles
use electronic and embedded devices, the attacker may gain control the vehi-
cle [6] by influencing the Electronic Control Unit (ECU). This would be a
significant driver safety hazard.
The Distributed Denial of Service Attack (or DDoS attack) [7] poses sig-
nificant risks to the performance of IoT applications and is amongst the
leading virtual threats on the internet. They occur because of the simplicity
of their execution, which is further aided by the fact that certain companies
offer such attacks as a service [8]. These attacks are much feared because
they can overpower their target and defeat the accessibility of their victim
within a matter of seconds. Recent reports suggest [9] a growing trend in the
scope of DDoS attacks. This has led to incurring financial losses and also
impacted millions of users worldwide. GitHub’s server attack in 2018 is one
of the most devastating DDoS attacks ever recorded. This attack was facili-
tated by targeting a vulnerable region of the application layer protocol. It
was designed to supply 129 million applications per second and reach a
total capacity of 1.35 Tbps [10], closely following the most notable attack
in 2016 [11], which involved 1.2 Tbps.
This chapter considers various challenges, methods, and practices for IoT
security. Based on these, two taxonomies are designed for security attacks
targeted at IoT systems. The first taxonomy introduces attacks on four-
tiered infrastructures of IoT (Perception Layer, Network Layer, Middleware
Layer, Application Layer). Based on this taxonomy, a systematic analysis of
security and privacy threats throughout the different layers of IoT is done.
The chapter also examines the types of DDoS attacks, how they work, their
impact on the IoT ecosystem, and other available protection mechanisms.
 Denial of Service Attacks in the Internet of Things 69

5.2 IoT ARCHITECTURE

The Internet of Things (IoT) does not have a fixed architecture. However, the
abstraction of layers can be viewed as consisting of either three or four layers.
The three-layer architecture [12] comprises an application layer, a network
layer, and a perception layer. Since IoT regards both objects and devices as
a thing, these are complex. The heterogeneous devices on the internet have
their constraints, and complexities grow when these devices are linked to
providing services to the objects. To address this issue, a middle layer [13]
is introduced to create an abstraction between technology and applications/
services. The middleware layer connects the network layer to the application
layer through cloud storage. This means the features of the network layer
are stored in the cloud and accessed by the application layer. This reduces the
complexity and frees up the Application layer’s load. This layer also provides
APIs [14] to the Application layer. Thus, the middleware layer supports data-
processing capability by simplifying computation. Each layer (Figure 5.1) is
discussed briefly, along with the technology used in each layer:

Application layer: The application layer is the service layer that con-
nects to the objects and provides services. These services include smart
homes, innovative health, automated vehicles, etc. This layer works

Figure 5.1 Four-layer IoT Architecture.

70 Aditi Paul and S. Sinha

on formatting and the presentation of data to the user. HTTP is the

primary service protocol in this case, but it requires substantial over-
heads. The resource-constrained devices cannot cope with these high
overheads, and hence various lightweight protocols are proposed in
Application layers. Constrained Application Protocol (CoAP), Message
Queue Telemetry Transport (MQTT), etc., one crucial aspect of the
application layer is quality of service (QoS). However, the handling
of sensitive information is challenging, and exploitation through the
actions of intruders, such as unauthorized access to data, data modi-
fications, etc. The vulnerabilities in application layers create paths for
attackers to exploit sensitive data.
Middleware layer: The role of the middleware [15] layer is to provide
service management, data storage, and service composition. This layer
receives data from the network layer, stores and processes data on the
cloud, and provides an application interface to the upper layer. The
middleware layer contributes to a more robust data processing and
repository. However, the vulnerabilities of cloud storage are one of the
challenges in this layer. This, in turn, affects the quality of service in
the application layer.
Network layer: This layer is responsible for communication between
IoT devices and also for data exchange. Sensor nodes send data from
the lower layer, routed through this layer. The Routing Protocol for
Low Power and Lossy Networks (RPL) is used, which is suitable for
constrained devices. The other protocols involved are ZigBee, Wi-Fi,
Bluetooth, and 3G. The network layer is prone to attacks like Denial
of Service and eavesdropping [16], which violates the authenticity and
availability of data/communication.
Perception layer: The perception layer [17] is the physical layer that
identifies objects and collects information. This layer is responsible
for converting digital data into signals. This edge layer consists of a
sensor-forming Wireless Sensor Network (WSN). These sensors are
constrained in energy and power consumption. The deployment of
these sensors in a hostile environment also causes vulnerabilities and
destruction. The lifetime of these sensors is dependable on the appli-
cation environment and topology, which causes communication dis-
ruption and a lack of efficiency, hence the attacks on the Snooping,
Sniffing, Compromised Identity, Eavesdropping, etc.

We now turn to an analysis of IoT attacks and security/privacy issues based

on the construction of the four layers described above. Figure 5.2 presents
the division of the attack categories. This section explains the principal risk
factors and mitigation challenges in more detail.
In this chapter, we focus our study on one form of attack, i.e., a Denial-of-
Service (DoS) attack [18, 19] on IoT devices and the need for new and capa-
ble technologies to protect these devices against such attacks. IoT devices
Denial of Service Attacks in the Internet of Things 71
Figure 5.2 Taxonomy of IoT attack at four layers.
72 Aditi Paul and S. Sinha

serving as botnets target vulnerable devices [20]. For example, the Mirai
botnet [21], which affected a significant part of the internet in 2016, affected
several countries and employed a large number of geographically distrib-
uted IoT devices to orchestrate their attack plan. Such a dispersed contin-
gent of IoT devices becomes challenging to spot and counteract, thereby
adding to the seriousness of the attack. A recent report [22] marked that in
the months from July to September of 2019, the highest rate of devices
hosted by botnets to launch DDoS attacks were from the United States and
the Netherlands. This is a problematic situation that demands the attention
of the scientific community to explore new strategies to improve security in
the IoT infrastructure and reduce the vulnerability of devices to dangerous
attacks like DDoS attacks.

5.3 CATEGORIZATION OF SECURITY ISSUES

Table 5.1 represents the mapping of various IoT security threats implied by
denial of service (DoS) with the layers they affect, the level of severity, and
their respective solutions.
Let us understand the IoT taxonomy based on its deployment infrastructure:

• Low level of security issues: The first level of security deals with secu-
rity issues in the physical layer, data link layers of communication, and
hardware level.
• Medium level of security issues: Intermediate security issues are par-
ticularly focused on communication, routing, and session management
network at IoT transport layers.
• High level of security issues: High-level security issues are mostly the
user or system applications running on IoT.

5.4 SECURITY ISSUES OF THE LOWER LAYERS

Jamming attacks: In this type of attack, the attacker causes communica-

tion stoppage by occupying a channel [23], thereby preventing other nodes
from using it. This is a particular type of DOS attack. Such attacks work
by emitting radio frequency signals without following specific protocols or
expected system protocols, hence disrupting networks in wireless devices
in IoT. This radio interference [24] affects network performance and may
act to and from a communication of official information nodes, leading the
system to malfunction.
Sybil and spoofing attacks: Malicious Sybil nodes [25, 26] employ fake
identities to launch attacks on wireless networks, eventually degrading the
system’s performance. Intending to extinguish network resources, Sybil
nodes use randomized MAC values impersonating [27, 28, 29] an authentic
Table 5.1 Layer-wise DoS attacks in IoT and mitigation strategies
Attack type IoT Layer Level of Attack Mitigation Strategies
Jamming Attacks Physical Layer Low-Level Evaluating the strength of the signal, delivery ratio of computer packets,
encrypting packets with error amending codes, as well as changes to
frequencies and locations [21] [22] [23]
Sybil and spoofing Physical Layer Low-Level Measuring signal strength, as well as estimation of the channel [24] [25]
attacks [26]
Insecure initialization Physical Layer Low-Level To set data transfer rates between nodes and introduce artificial noise to
and configuration the channel [27] [28] [29]

Denial of Service Attacks in the Internet of Things

Insecure physical Hardware Low-Level Not granting USB or any other hardware-based module or debugging
interface tools access to system software [30]
Duplication 6LoWPAN Intermediate Level Presentation of Time stamp and nonce protection options against replay
attacks due to adaptation layer attacks and fragment confirmation via hash chains [31] [32]
fragmentation and network layer
Sinkhole and Network layer Intermediate Level Ranking requests via hash chain function, building a level of trust and
wormhole attacks managing it, analyzing the actions of the nodes.
Verification rate through Hash chain function, management of trust
level, incorporating IDS to detect irregularities, cryptographic key
management, signal strength evaluation, traversal of graph [33] [34] [35]
[36] [37] [38]
Session establishment Transport layer Intermediate level Long-lived secret key authentication, and key-based symmetric encryption
and restarting. [39] [40] [41]
CoAP security with Application layer High-level and TLS/DTLS and HTTP/CoAP mapping, Mirror Proxy (MP) and Resource
internet and network layer intermediate level Directory, TLS-DTLS tunnel, and message filtration using 6LBR [42]
Insecure interfaces Application layer High-level Not allowing weak passwords, check the interface against being at risk of
software tools (SQLi and XSS), as well as using HTTPS with firewalls [30]
Insecure software/ Application layer, High-level, General security updates for software/firmware, usage of file signatures,
firmware transport layer, intermediate level, and encryption with confirmation [30]

73
and network layer and low-level
74 Aditi Paul and S. Sinha

device in the physical layer. This results in genuine devices missing network
access.
Insecure initialization and configuration: The configuration and initializa-
tion of IoT devices [30] at the physical layer without obstructing network
services or privacy policies guarantees secure and efficient working of the
entire system. Connections at the physical layer also need protection against
unauthorized recipients trying to gain access.
Insecure physical interface: Many biological factors pose a severe threat
to the efficiency of IoT devices. Imagine the damage that can happen if a
harmful individual gains physical access to a task-critical device. Inefficient
physical security [31], unauthorized access to software through physical
avenues, and testing/debugging tools may compromise the network nodes.

5.5 SECURITY ISSUES OF THE INTERMEDIATE LEVELS

Replay or duplication attacks [32] due to fragmentation: For devices obey-

ing the standards of IEEE 802.15.4, the fragmentation ofIPv6 packets is
necessary. When the packet fragment fields are recreated in the 6LoWPAN
layer, resources may be depleted [33], a buffer may overflow, and the device
may reboot. Duplicate pieces posted by malicious nodes affect packet recon-
struction, thereby preventing the processing other genuine packets.
Sinkhole and wormhole attacks: The attacker employs a malicious node
to reply [34, 35] to routing requests, bringing packets and contact with hos-
tile agents, and thereby exposing it to mishaps. Network attacks may fur-
ther exacerbate the performance of 6LoWPAN due to the wormhole attack
[36], where a tunnel is built between two nodes so that packets reach from
one location to another immediately. This attack has serious consequences,
which include eavesdropping [37], breach of privacy, and Denial of Service.
Session establishment and resumption: If a session is hijacked in the trans-
port layer via fake/forged messages, it leaves the system vulnerable to a DoS
attack. The attacking node can masquerade [38] as a victim to maintain the
session between two nodes. The re-relay of messages by switching sequence
numbers may be required.

5.6 SECURITY ISSUES OF THE HIGHER LEVELS

CoAP security [39] with the internet: CoAP, short for Constrained
Application Protocol, works at the application layer in the high-level layer
and is a web transfer protocol well suited for nodes that run on simple
microcontrollers with a limited amount of RAM and ROM and commu-
nicate over low-power wireless area networks. It employs DTLS ties with
many other ways to protect both ends of the network. Defined in RFC-
7252, they follow a scheme that needs encryption to enhance security.
 Denial of Service Attacks in the Internet of Things 75

Insecure interfaces: The port recruited to access IoT assets via the web,
mobile, or cloud is always susceptible to attacks that threaten data privacy [40].
Insecure software/firmware: IoT devices, when performing updates, go
for insecure network protocols, thus exposing themselves to attacks [41]
where the attacker can span across the network and remodel it to his desire.
He can even infect the other nodes connected via this device’s hub.
Middleware security: The middleware [42] layer in IoT serves the purpose
of establishing correspondence between various bodies in the IoT paradigm,
which, in order to reap the benefit of its service, must be adequately secured.

5.7 MITIGATION STRATEGIES AT VARIOUS LEVELS

OF IoT DEVICES

As we have seen, IoT devices have many such avenues that can be plagued
by various security issues. Some examples of these avenues are insecure web
interfaces, insecure network services, transport lacking encryption, insecure
cloud interface, insecure mobile interface, and insecure software/firmware
to name a few. Since IoT has become such an indivisible part of our society,
threats to its security must be dealt with conscientiously. Here are a few
mitigation strategies at different levels of IoT.
Security solutions for the lower level: In wireless sensor networks, the
interruptions that lead to message conflicts or channel overload are linked
to jamming attacks, as referred to by [23, 24] in the previous section. One
of the methods used (as mentioned in these references) to discern whether a
jamming attack has occurred is by calculating the effective rate in which a
packet is delivered. This design functions by executing statistical tests on the
strength of the signal and node address. An alternative to counter jamming
attacks involves cryptographic functions and error correction codes. The
process progresses by encoding packet codes by breaking it into blocks and
leaving pieces of the coded package. Similarly, strategies involving channel
filtering and site deceleration have also been proposed to combat jamming
attacks. Young et al. [42] proposed a tactic to detect jamming attacks. First
the signal strength is measured, then it is utilized to draw out signals that are
noise-like. Later these numbers are collated with standard threshold values
assigned for attack detection. Channel browsing allows legitimate commu-
nication devices to switch channels; this, in turn, causes these devices to alter
their address in between the process.
As previously referred to in [27, 28, 29] malicious Sybil nodes imperson-
ate a distinct device by using a forged MAC address which leads to resource
exhaustion and denial to grant access to genuine customers. One of the
fundamental approaches here is to track Sybil attacks by determining the
power of the signal. Detector nodes are positioned in the communication
channel to pinpoint the sender’s address. Sybil attacks are characterized by
a similar sender location, but different sender IDs. Signal strength ratings on
76 Aditi Paul and S. Sinha

MAC addresses can also be used to detect fraudulent attacks. Another alter-
native can be the use of Received Signal Strength Indicator (RSSI) for the
detection of a Sybil attack. Trust-based models are also very effective
approaches to detect Sybil attack in MANET.
In general devices with poor physical security owe their failing to insecure
connections to software, firmware and flawed testing/debugging tools.
Hardware ports like USB ports or an SD card slot can be easily compro-
mised by granting access to malicious agents. Open Web Application Security
Project (OWASP) provides recommendations for an improvement in the
physical safety of IoT devices. Testing and debugging tools should be dis-
posed and methods such as Trusted Platform Modules (TPMs) should be
brought in to improve physical safety.
Security solutions for the intermediate level: By adding time stamps and
nonce options to fragmented packet threats from replay packages can be
alleviated. Correlated package sections are connected to the 6LoWPAN
adaptation layer. The timestamp option works for unidirectional packets
whereas the nonce option works for the bidirectional packets. The 64-bit
time stamp on the piece ensures the removal of recurring obsolete headers
and diversions from the network. The fact that the header is made only in
response to a new application is ensured by the nonce option. Also, to make
sure that the order of packets is not disrupted when transferring IPv6 packet
fragments domestically via 6LoWPAN.
Riaz et al. [43] propose a security structure with modules to facilitate
locating safe neighbour, authenticity confirmation, the generation of a key
and the encryption of data. To locate a safe neighbour Elliptic Curve
Cryptography (ECC) can be considered. Weekly et al. [44] propose dealing
with so-called sinkhole attacks on low-power Lossy networks with a pro-
cess that combines failover and verification techniques. In rank verification
regarding Destination Information Object (DIO) message, one way hash
function as well as hash chain function are used.
On network layer Sybil attacks use fake identities to present themselves as
many different individual devices, which are termed Sybil nodes [45].
Participation and the distribution of IoT peers to peers (p2p) systems is
threatened by this. Additionally, reliable streaming on the network is risked
because the defence against Byzantine errors is influenced. For communal
interactions, trust building is encouraged to limit a new Sybil ID from
emerging. Social graphs are used as a preventive measure to enable the iden-
tification of Sybil nodes by either randomly walking through the graph or
using certain detection algorithms in the community [46].
Security solutions for the higher level: Brachmann et al. [47] propose a
method combining Transport Layer Security(TLS) and Datagram Transport
Layer Security (DTLS) to defend CoAP-based low-power, Lossy network
(LLN) in touch with the internet. Instances where this method finds most
uses are in cases where 6LBR is linked to LLN through the internet to
remotely gain access to devices. CoAP and HTTP clients avail the services of
 Denial of Service Attacks in the Internet of Things 77

LLN data nodes. To protect connection security at every end of the system,
TLS and DTLS mapping is suggested. When the design for mapping compu-
tation is effectuated on devices with arrested assets, the system sustains hefty
expenses. Granja et al. [48] offer a way to protect the application messages
sent via the internet, and aided by various CoAP security options. Some new
alternatives to enhance CoAP-related security are SecurityOn, SecurityToken,
and SecurityEncap. SecurityOn defends CoAP messages at the application
level. SecurityToken helps in authentication and verified access to CoAP
services. SecurityEncap [49] employs SecurityOn and principally performs
tasks that enable data channeling necessary for authentication to block relay
attacks.
The OWASP project presents specific remedies to secure IoT devices.
Weak passwords, vulnerable software tools, and not using HTTPS along
with the firewalls should be avoided in order to ensure security at a higher
level. Additionally, the installed software should continuously be updated
only via encrypted channels. Only signed and validated update files must be
downloaded from dependable servers.
When securing distributed applications in IoT, encryption and authentica-
tion, as established in VIRTUS middleware proposed by Conzon et al. [50],
should be considered. While embracing TLS and SASL for data cohesion,
authentication, and the encryption of XML stream, middleware employs an
event-drive communication approach. Validating the source ensures the
secure transmission of data. When web services are enabled with VIRTUS
middleware, it provides dependable and manageable IoT interfaces.
Otsopack [51] is another remedy to help heterogeneous implementations in
a secure manner. A semantic structure acts as a middleware using a semantic
format based on Triple Space Computing (TSC).

5.8 DISTRIBUTED DENIAL OF SERVICE (DDoS) ATTACK

Distributed Denial of Service (DDoS) attacks generally result from drain-

ing network resources or malfunctioning [52, 53] computer hardware. This
means that the resources are unavailable to legitimate users. Consequently,
a DDoS attack involves capturing various devices and further using them
to send multiple requests [54] to the server/host or exploiting some of their
known threats. DDoS attacks can be categorized in three ways [55]:

i. Application layer attack;

ii. Resource exhaustion attack; and
iii. Volume attack

Application layer attack: This includes low-and-slow attacks, GET/POST

floods [56], web server attacks, and so on. These attacks exploit the
vulnerabilities of Apache, Windows or OpenBSD, and other operating
78 Aditi Paul and S. Sinha

systems. The intention of these attacks is to crash the web server. Often,
these attacks are treated as operational errors, as low-volume traffic
requires brutality to copy the behavior of legitimate users. Therefore,
this attack is ignored even after obtaining multiple visual indications.
Slowloris [57, 58] is the most common attack on the applica-
tion layer. In this case, partial requests are sent to the web server at
specified times to keep several connections open for a more extended
period. As a result, the target web server reaches its maximum simul-
taneous connection pool capacity. This makes the server unavailable
to new requests (legitimate) immediately after the start of attacks. To
ensure it works, Slowloris uses the minimum bandwidth, which makes
it remain undetected for a long.
Resource Exhaustion Attack: These attacks take advantage of commonly
used protocols in the network layer to wear out hardware resources
like memory, CPU, and storage. Therefore, such attacks are affected
not just by the amount of traffic but also by the combination of par-
ticular messages.
The typical case of such attacks involves victimizing TCP com-
munication protocol features. TCP SYN Flood, a classic instance of
this attack, initiates a three-way TCP handshake. Then by flooding
SYN request messages to the target from a fake source address, the
attacker indulges the victim in setting up a new link for the hostile
user. After setting up the connection, the target waits for an acknowl-
edgment (ACK) from the client to complete the setup link, which is
never achieved as the attacker does not send ACK. Eventually, it causes
the backlog to exhaust, thereby rendering the option of opening a new
connection impossible.
Volumetric Attack: Volumetric Attack [59, 60] sends considerable data
traffic to a network to exhaust its bandwidth. Since the bulk of data
(100Gpbs) is required, this attack depends more on the volume of
data than application layer attacks and resource exhaustion attacks.
Volumetric attacks are launched from multiple sources’ IP addresses
creating data amplification. This is why this type of attack is hard
to mitigate manually. The high volume of data requests results in the
expansion of response by servers, thus tiring out the target’s band-
width. Domain Name Server (DNS) Amplification and spoofed
Network Time Protocol (NTP) are among these types of attack.

5.9 DDOS ATTACK MITIGATION STRATEGIES

The following sections describe effective strategies to reduce DDoS attacks

in IoT.
Flow filtering: Flow filtering is the most straightforward method amongst
mitigation solutions as it is commonly used on devices that adhere to
 Denial of Service Attacks in the Internet of Things 79

OpenFlow [61, 62]. This strategy factors in the header fields of packages
arriving at OpenFlow devices to block the flow of devices labeled as malig-
nant. Among the major criteria that are effective during filtering are:

i. The address of the source

ii. Address of destination
iii. Origin port
iv. Target port
v. Network layer protocol

Despite being the most efficient option, it risks creating a bottleneck in the
communication interface of the controller owing to its dependence on the
controller for statistics gathering and packet inspection, aggregating to
enhanced delay:

Honey pots: This approach employs systems in a sheltered and secluded

atmosphere to mimic a natural target, thus luring malicious agents. All
the information collected via honey pots [63] helps enable the research
and development in the field of mitigation policies. Despite being a
conventional method for attack reduction, it can be used in alignment
with SDN to gather data alongside the controller about malicious
traffic.
Rate Limiting: In the case of a volumetric attack, a sizeable amount of
malicious traffic results in the unavailability [64] of the network due
to overcrowding of its communication links. In this method, a control-
ler defines a maximum traffic limit that can be entertained without the
system being deemed overcrowded. Once the network tips the edge of
overload, it then declines all the incoming packages. This schema is
generally used in conjunction with Advanced Package Test Processes.
Moving Target Defense: As the name suggests, moving target [65] defense
works by perpetually resetting system characteristics based on a ran-
dom value set, preventing the system from being susceptible to attacks
that would render it unavailable for sizeable periods. Put simply, the
system reinvents certain aspects of its configuration to avoid being tar-
geted. Randomization of IP and MAC addresses is generally followed
to make getting hold of host and server information difficult from the
network while the process is underway. Despite the wide acceptance of
MTD techniques to alleviate DDoS attacks, certain limitations cannot
be ignored. For instance, the effect regards to cost and performance
when installing MDP on a large-scale network.
Traceback: In this method, information from the package title is used to
identify the place of launch/the birthplace of the attacker. In general
networks, this task is tricky as the network switches cannot recognize
the attacker through their falsified source address fields. However, the
advantages that can be enjoyed if we regard the holistic view of the
80 Aditi Paul and S. Sinha

SDN control plane, then mitigation strategies established on traceback

are not a bad option.
Request Prioritization: In this method, the flow of the network processes
are managed by setting a priority value. Source hosts for every arrival
packet at the SDN controller are allocated a default reliability value
based on the hosts’ traffic history. Suspicious activities may alter these
values by reducing them from time to time. The confidentiality value
set by the system administrator, if found low for a source, then its flow
is denied.

5.10 RESEARCH GAP

DoS/DDoS attacks are prevalent in multiple layers of IoT architecture, and

the mitigation strategies that are proposed to date consider a single layer
only. Thus, the research gap lies in the following questions:

• Is it worthwhile to address each layer separately in mitigating DoS

attacks in IoT?
• What happens when an attacker changes its signature in real time?
• Once a detection strategy is implemented, will it be able to respond to
the change in attack signature?
• Is training a system with a fixed value of a data set and testing the
system with a real-time data set feasible?
• Is it appropriate to design a generic system capable of detecting DoS
attacks by analyzing features from multiple layers in real-time?

To answer the above questions, it is essential to approach a cross-layer

attack detection structure in IoT. This is logical because a single-layer detec-
tion mechanism cannot address the heterogeneous nature of IoT devices.
Instead, grouping together closely related attacks at different layers will be a
more efficient approach to understanding the attacks. Thus, designing a sys-
tem capable of analyzing multiple attack signatures in real time is a central
aspect of any future IoT DoS attack detection strategy.

5.11 FUZZY-NEURAL NETWORK-BASED CROSS-LAYER

DoS ATTACK DETECTION FRAMEWORK

To create the cross-layer framework for detecting multiple DoS attacks in

real time, a Fuzzy-Neural Network-based cross-layer attack detection mech-
anism is proposed in this chapter (Figure 5.7). As mentioned in the section
on the research gap, this system works in two stages and can identify real-
time attack signature variations. Since we are explicitly designing a frame-
work for identifying cross-layer attacks, it is evident to start with cross-layer
 Denial of Service Attacks in the Internet of Things 81

attack features to be selected. The following subsections discuss the feature

selection in brief:

a. Cross-layer attack features selection:

As a preliminary proof-of-concept of cross-layer D/DoS attack detec-
tion framework, we consider two attacks, Resource Exhaustion Attack
and Volumetric Attack, as these are very prominent attacks in IoT sce-
narios. As discussed in the previous section, a Resource Exhaustion
Attack is a flooding attack through which the attacker exhausts net-
work resources. TCP SYN Flood Attack (at the Transport Layer) is
one type of such an attack. On the other hand, volumetric attacks
flood the network by sending a considerable amount of data traffic,
thereby exhausting resources. A DIS flooding attack on RPL (Network
Layer) routing protocol is one such attack.
The initial and essential step is to analyze attack features for
which we have used the NetSim simulator to generate trace files for
both attacks (Figures 5.3 and 5.4). One important aspect here is
feature selection, which closely defines the attack signature. This is
done by using trace files in the NetSim simulator. After features are
extracted, each of the attack data sets is created. At this point, data
are pre-­processed for feature selection and classification. After data
pre-processing, these data sets are processed using machine learning
algorithms to select specific features contributing to the attack behav-
ior. In the case of TCP SYN Flood attacks, these features are shown in
Figures 5.5 and 5.6.
Comparing Figures 5.5 and 5.6, we see that number of TCP_SYN
packets sent to destination node 4 is almost double in the attack sce-
nario. In addition, these attacker nodes are 2, 6, 8, 9 and 10, which
indicates that the parameters source id, destination id, TCP_SYN
packet sent, SYN-ACK sent, and Interval contribute to the attack
signature.
Similarly, DIS interval, delay, and DIS packet sent are the parameters
that can be considered attack parameters for a DIS flooding attack.
b. Proposed framework:
Now we proceed to the first stage of the proposed framework. In the
first stage, the cross-layer data set is created by generating traces using
a simulator (NetSim), and a Neural Network is trained with this data
set. Along with the data set, one extra input is added to the NN, which
labels the input data as either normal (0) or attack (1). We call this an
input data type (IDT). This helps to train the NN more precisely with
a specific data category. The output of the NN is the detection of an
attack, i.e., normal (0), SYN flood attack (1), and DIS attack (2).
The second stage starts once the training of NN is completed. Here
we use the NN for accurate time attack detection. At this point, it is
crucial to reconsider the real-time IDT data set. The training data set
 82 Aditi Paul and S. Sinha
Figure 5.3 TCP SYN Flooding attack.
Denial of Service Attacks in the Internet of Things 83
Figure 5.4 DIS Flooding attack.
84 Aditi Paul and S. Sinha

Figure 5.5 T
 CP SYN Flood attack scenario, Total no. of TCP_SYN packet sent to
the destination Node-4 before the attack.

Figure 5.6 T
 CP SYN Flood attack scenario, Total no. of TCP_SYN packet sent to
the destination Node-4 after an attack.

is pre-defined, but the real-time data is unpredictable. Thus, the attack

data set will have variations compared to the training data set. To
reconstruct the IDT, we use the fuzzy rule manager whose output gives
us two IDT, viz. normal (0) and attack (1). The fuzzy rule manager
works in five steps as discussed below:

Step 1: Create a data set for specific time instances (t1, t2, t3 … tn)
Step 2: For each feature, calculate the average value of all time instances
(row average).
 Denial of Service Attacks in the Internet of Things 85

Figure 5.7 Workflow diagram of cross-layer DoS attack detection framework.

Step 3: Calculate the maximum and minimum variation of individual fea-

tures during the time instances (t1, t2, t3 … tn)
Step 4: Calculate upper bound = (average feature value + maximum varia-
tion) and lower bound = (average feature value – minimum varia-
tion) for each feature.
Step 5: If all feature values at any time lie between the upper bound and
the lower bound, the feature set represents a normal IDT (0); other-
wise, if any of the features’ values go above upper bound or below
lower bound, it’s an attack IDT (1).

The fuzzy rule manager’s justification is to analyze each feature’s upper and
lower bound during a time interval. As long as each component lies between
the upper and lower bound, the IDT remains at 0 value, and the fuzzy rule
manager goes on checking the next feature. As soon as any part shows a
value above or below the upper or lower bound, the process stops at that
point and updates IDT as 1. This reduces the processing time of the tech-
nique in the case of an attack data set. The value of n is an integer that helps
to increase the detection accuracy.
Once the IDT is reconstructed, the input testing data set to the NN is sent
in real time, and the output is either expected or any of the two attacks. The
importance of IDT is the minimization of a false positive rate due to unde-
tected variation in the data set. This is solved using a fuzzy rule manager,
which can enhance the detection accuracy once implemented.
86 Aditi Paul and S. Sinha

5.12 CONCLUSION

Cross-layer detection framework for DoS attack detection is complex as

the exact threshold of attack value is never fixed. With the change of attack
signature, the importance of the selected features will change, which may
incorrectly interpret an event as an attack, i.e., generation of false positive
and false negative, which is revealed from an experimental study. In conclu-
sion, it is observed that cross-layer framework accuracy depends on two
significant factors.

• Selection of attack signature as that can vary on an attack basis

• The choice of upper and lower bound for fuzzy rule managers accord-
ing to the variation of features.

To increase the detection accuracy, the variation in attack signature can be

considered dynamically during testing through the NN, which might be
effective according to observation. As proposed in this chapter, the fuzzy
rule manager can calculate this variation by fixing up the upper and lower
bounds of the variation in each feature.

5.13 FUTURE WORK

The heterogeneity of IoT devices drives the implementation of a dynamic

attack detection framework to be the most relevant, and possibly the most
logical, way of implementing cross-layer DDoS attack detection architec-
ture. Thus, the future aspect is to implement the framework with a real-time
data set, increase the detection accuracy by dynamic attack signature selec-
tion, and come up with a uniform set of fuzzy rules with upper and lower
bounds. Overall, the framework is best to be worked with a signature-based
strategy and for a known attack. Still, the framework has to be renewed and
taken as future work for unknown environments and attacks requiring an
anomaly-based approach.

REFERENCES

[1] M. Lombardi, F. Pascale, and D. Santaniello, “Internet of Things: A General

Overview between Architectures”, Information, 21, p. 1, 2021.
[2] S. K. Jain, and N. Kesswani,“Privacy Threat Model for IoT”, In: Nain, N.,Vipparthi,
S. (Eds.) 4th International Conference on Internet of Things and Connected
Technologies (ICIoTCT), Advances in Intelligent Systems and Computing, vol
1122. Springer, Cham, 2020. https://doi.org/10.1007/978-3-030-39875-0_30
[3] F. Meneghello, M. Calore, D. Zucchetto, M. Polese, and A. Zanella, “IoT:
Internet of Threats? A Survey of Practical Security Vulnerabilities in Real IoT
Devices”, IEEE Internet of Things Journal, 6, pp. 1–2, 2019.
 Denial of Service Attacks in the Internet of Things 87

[4] K. Chen, S. Zhang, Z. Li, Y. Zhang, Q. Deng, S. Ray, and Y. Jin, “Internet-of-
Things Security, and Vulnerabilities: Taxonomy, Challenges”, Hardware and
Systems Security, 2, pp. 97–110, 2018.
[5] C.-K. Wu, Internet of Things Security, in Advances in Computer Science and
Technology, Springer, Singapore, pp. 1–12, 2021.
[6] S. Checkoway, D. McCoy, B. Kantor, D. Anderson, H. Shacham, S. Savage, K.
Koscher, A. Czeskis, F. Roesner, and T. Kohno, “Comprehensive Experimental
Analyses of Automotive Attack Surfaces,” in Proceedings of the 20th USENIX
Security Symposium, San Francisco, 2011.
[7] B. B. Gupta and A. Dahiya, “Fundamentals of DDoS Attack: Evolution and
Challenges,” in Distributed Denial of Service (DDoS) Attacks, CRC Press, pp.
1–18, 2021. eBook. ISBN 9781003107354.
[8] https://securelist.com/the-cost-of-launching-a-ddos-attack/77784/
[9] A. Marzano, D. Alexander, O. Fonseca, E. Fazzion, C. Hoepers, K. Steding-
Jessen, M. H. P. C. Chaves, Í. Cunha, D. Guedes, and W. Meira, “The Evolution
of Bashlite and MiraiIoT Botnets”, in 2018 IEEE Symposium on Computers
and Communications (ISCC), Natal, 2018.
[10] https://www.cloudflare.com/learning/ddos/famous-ddos-attacks/.
[11] https://www.theguardian.com/technology/2016/oct/26/ddos-attack-dyn-mirai-
botnet.
[12] M. R. Abdmeziem, D. Tandjaoui, and I. Romdhani, Architecting the Internet of
Things: State of the Art, in Studies in Systems, Decision, and Control, Springer,
Cham, pp. 56–62, 2015.
[13] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, and M. Ayyash, “Internet
of Things: A Survey on Enabling Technologies, Protocols, and Applications”,
IEEE Communications Surveys Tutorials, 17, pp. 2347–2376, 2015.
[14] P. K. Donta, S. N. Srirama, T. Amgoth, and C. S. R. Annavarapu, “Survey on
Recent Advances in IoT Application Layer Protocols and Machine Learning
Scope for Research Directions”, Digital Communications, and Networks, 8, 5,
pp. 2352–8648, 2021.
[15] A. H. Ngu, M. Gutierrez, V. Metsis, S. Nepal, and Q. Z. Sheng, IoT Middleware:
“A Survey on Issues and Enabling Technologies”, IEEE Internet of Things
Journal, 4, pp. 1–20, 2017.
[16] P. Sethi, and S. R. Sarangi, “Internet of Things: Architectures, Protocols, and
Applications”, Journal of Electrical and Computer Engineering, 54, pp. 1–15,
2017.
[17] J. Zhou, Z. Cao, X. Dong, and A. V. Vasilakos, “Security and Privacy for
Cloud-Based IoT: Challenges”, IEEE Communications Magazine, 55, pp. 26–
33, 2017.
[18] S. Mishra, and A. Paul, “A Critical Analysis of Attack Detection Schemes in IoT
and Open Challenges”, 2020 IEEE International Conference on Computing,
Power and Communication Technologies (GUCON), Greater Noida, India,
pp. 57–62, 2020, doi:10.1109/GUCON48875.2020.9231077.
[19] J. Dizdarevic, F. Carpio, A. Lukan, and X. Masip-Bruin, “A Survey of
Communication Protocols for Internet of Things and Related Challenges of
Fog and Cloud Computing Integration”, ACM Computing Surveys, 51, 6,
pp. 1–29, 2019.
[20] C. Wheelus, and X. Zhu, “IoT Network Security: Threats, Risks, and a Data-
Driven Defense Framework”, IoT, vol 1, pp. 259–282, 2020.
88 Aditi Paul and S. Sinha

[21] G. Kambourakis, C. Kolias, and A. Stavrou, “The mirai botnet and the iot zom-
bie armies”, Military Communications Conference (MILCOM), pp. 267–272.
IEEE, 2017.
[22] https://securelist.com/ddos-attacks-in-q3-2021/104796/
[23] W. Xu, W. Trappe, Y. Zhang, and T. Wood, “The feasibility of launching and
detecting jamming attacks in wireless networks”, in Proceedings of the 6th
ACM, New York, 2005.
[24] G. Noubir, and G. Lin, “Low-power DoS Attacks in Data Wireless LANs
and Countermeasures”, ACM SIGMOBILE Mobile Computing and
Communications Review, Vol. 7, No. 3, pp. 29–30, 2003.
[25] Sinha, S., and A. Paul “Neuro-Fuzzy Based Intrusion Detection System for
Wireless Sensor Network”. Wireless Pers Communication, vol 114, pp. 835–851,
https://doi.org/10.1007/s11277-020-07395-y
[26] A. Paul, and S. Sinha, “Performance Analysis of Received Signal Power-Based
Sybil Detection in MANET Using Spline Curve”, International Journal of
Mobile Network Design and Innovation, 7, pp. 222–232, 2018, doi:https://
doi.org/10.1504/IJMNDI.2017.089304.
[27] S. Sinha, A. Paul, and S. Pal, “The Sybil Attack in Mobile Adhoc Network:
Analysis and Detection”, in Proceedings of International Conference on
Recent Trends in Communication and Computer Networks-ComNet 2013,
pp. 95–103, 2013.
[28] M. Demirbas, and Y. Song, “An RSSI-Based Scheme for Sybil Attack Detection
in Wireless Sensor Networks”, International Symposium on a World of
Wireless, Mobile and Multimedia Network, IEEE, Washington, 2006.
[29] A. Paul, S. Sinha, and S. Pal, “An Efficient Method to Detect Sybil Attack Using
Trust Based Model”, in Proceedings of Fourth International Joint Conference on
Advances in Engineering and Technology AET-ACS-2013, pp. 228–237, 2013.
[30] P. Tommaso, L. Brilli, and L. Mucchi, “The Role of Physical Layer Security in
IoT: A Novel Perspective”, Information, 7, 49, 2016.
[31] https://owasp.org/index.php/Top_IoT_Vulnerabilities.
[32] H. Kim, “Protection Against Packet Fragmentation Attacks at 6LoWPAN
Adaptation Layer”, in 2008 International Conference on Convergence and
Hybrid Information Technology, IEEE Xplore, 2008.
[33] R. Hummer, J. Hiller, H. Wirtz, M. Henze, H. Shafagh, and K. Wehrle,
“6LoWPAN Fragmentation Attacks”, in Sixth ACM Conference on Security
and Privacy in Wireless and Mobile Networks, New York, 2013.
[34] K. Weekly, and K. Pister, “Evaluating Sinkhole Defence Techniques in RPL net-
works”, in 20th IEEE International Conference on Network, Washington, pp.
1–6, 2012.
[35] F. Ahmed, and Y.-B. Ko, “Mmitigation of Black Hole Attacks in Routing
Protocol for Low Power and Lossy Network”, Security and Communication
Networks, 9, 5143–5154, 2016.
[36] A. A. Pirzada, and C. McDonald, “Circumventing sinkholes and wormholes”,
in International Workshop on Wireless Ad-Hoc Networks, 2005.
[37] W. Wang, J. Kong, B. Bhargava, and M. Gerla, “Visualisation of Wormholes in
Underwater Sensor Networks: A Distributed Approach”, International Journal
of Security and Networks (IJSN), 3, pp. 10–23, 2008.
[38] N. Park, and N. Kang, “Mutual Authentication Scheme in Secure Internet of
Things Technology for Comfortable Lifestyle”, Sensors, 16, p. 20, 2016.
 Denial of Service Attacks in the Internet of Things 89

[39] M. H. Ibrahim, “Octopus: An Edge-Fog Mutual Authentication”, International

Journal of Network Security, 18, pp. 1089–1101, 2015.
[40] S. Raza, D. Trabalza, and T. Voigt, “6LoWPAN Compressed DTLS for CoAP”,
in 8th International Conference on Distributed Computing in Sensor Systems,
Hangzhou, pp. 287–289, 2012.
[41] H. G. C. Ferreira, R. T. de Sousa, F. E. G. de Deus, and E. D. Canedo, “Proposal
of a secure, deployable and transparent middleware for the Internet of Things”,
2014 9th Iberian Conference on Information Systems and Technologies
(CISTI), pp. 1–4, 2014.
[42] M. Young, and R. Boutaba, “Overcoming Adversaries in Sensor Networks:
A Survey of Theoretical Models and Algorithmic Approaches for Tolerating
Malicious Interference”, IEEE Communication Surveys & Tutorials, 13,
pp. 617–641, 2011.
[43] R. Harkanson, and Y. Kim, “Applications of elliptic curve cryptography: A
light introduction to elliptic curves and a survey of their applications”, in
Proceedings of the 12th Annual Conference on Cyber and Information Security
Research, Tennessee, 2017.
[44] K. Weekly, and K. Pister, “Evaluating Sinkhole defense techniques in RPL net-
works”, in 20th IEEE International Conference on Network, Washington,
DC, 2012.
[45] H. Yu, M. Kaminsky, P. B. Gibbons, and A. Flaxman, “Sybil Guard: Defending
Against Sybil Attacks via Social Networks”, ACM SIGCOMM Computer
Communication Review, 36, pp. 267–278, 2006.
[46] D. Quercia, and S. Hailes, “Sybil attacks against mobile users: friends and foes
to the rescue”, in The 29th Conference on Computer Communications (IEEE
INFOCOM 2010), Piscataway, NJ, USA, pp. 1–5, 2010.
[47] M. Brachmann, S. L. Keoh, O. G. Morton, and S. S. Kumar, “End-to-End
Transport Security in the IP-based”, in 21st International Conference on
Computer Communications and Networks, Munich, Germany, pp. 1–5, 2012.
[48] J. Granjal, E. Monteiro, and J. S. Silva, “Application-layer security for the WoT:
Extending CoAP to Support end-to-end message security for internet-inte-
grated sensing applications”, in International Conference on Wired/Wireless
Internet Communication, St. Petersburg, pp. 140–153, 2013.
[49] M. Brachmann, O. Garcia-Morchon, S.-L. Keoh, and S. S. Kumar, “Security con-
siderations around end-to-end security in the IP-based Internet of Things”, In
Workshop on Smart Object Security, in conjunction with IETF83, Paris, 2012.
[50] D. Conzon, T. Bolognesi, P. Brizzi, A. Lotito, R. Tomasi, and M. A. Spirito, “The
VIRTUS Middleware: an XMPP based architecture for secure IoT communica-
tions”, in 21st International Conference on Computer Communications and
Networks, IEEE, pp. 1–6, 2012.
[51] A. Gómez-Goiri, P. Orduña, J. Diego, and D. López-de-Ipiña, “Otsopack:
Lightweight Semantic Framework for Interoperable Ambient”, Computers in
Human Behavior, 30, pp. 460–467, 2013.
[52] C. H. Liu, B. Yang, and T. Liu, “Efficient Naming, Addressing and Profile
Services in Internet-of-Things Sensory Environments”, Ad Hoc Networks, 18,
pp. 85–101, 2014.
[53] A. Lohachab, and B. Karambir, “Critical Analysis of DDoS—An Emerging
Security Threat over IoT Networks,” Journal of Communications and
Information Networks, 3, 3, pp. 57–78, 2018.
90 Aditi Paul and S. Sinha

[54] S. T. Zargar, J. Joshi, and D. Tipper, “A Survey of Defence Mechanisms Against

Distributed Denial of Service (DDoS) Flooding Attacks”, EEE Communications
Surveys & Tutorials, 15, pp. 2046–2069, 2013.
[55] N. Dayal, P. Maity, S. Srivastava, and R. Khondoker, “Research Trends in
Security and DDoS in SDN”, Security and Communication Networks, 9,
pp. 6386–6411, 2016.
[56] R. Vishwakarma, and A. K. Jain, “A Survey of DDoS Attacking Techniques and
Defense Mechanisms in the IoT Network”, Telecommunication Systems, 73,
pp. 3–25, 2020.
[57] S. McGregory, “Preparing for the Next DDoS Attack”, Network Security, 5,
pp. 5–6, 2013.
[58] Y. G. Dantas, V. Nigam, and I. E. Fonseca, “A selective defense for applica-
tion layer DDoS attacks”, in IEEE Joint Intelligence and Security Informatics
Conference, The Hague, The Netherlands, pp. 75–82, 2014.
[59] S. Mansfield-Devine, “The Evolution of DDoS”, Computer Fraud & Security,
10, pp. 15–20, 2014.
[60] B. Sieklik, R. Macfarlane, and W. J. Buchanan, Evaluation of TFTP DDoS
Amplification Attack, Computers & Security, 57, pp. 67–92, 2016.
[61] T. Lukaseder, K. Stolzle, S. Kleber, B. Erb, and F. Kargl, “An SDN-based
Approach For Defending Against Reflective DDoS Attacks”, in IEEE 43rd
Conference on Local Computer Networks, Chicago, pp. 299–302, 2018.
[62] G. Sun, W. Jiang, Y. Gu, D. Ren, and H. Li, “DDoS Attacks and Flash Event
Detection Based on Flow Characteristics in SDN”, in Proceedings of the 2018
15th IEEE International Conference on Advanced Video and Signal Based
Surveillance (AVSS), Auckland, pp. 1–6, 2018.
[63] X. Luo, Q. Yan, and M. Wang, Huang W, “Using MTD and SDN-based
Honeypots to Defend DDoS Attacks in IoT”, in Proceedings of the 2019
Computing, Communications and IoT Applications (ComComAp), Shenzhen,
China, pp. 392–395, 2019.
[64] S. S. Bhunia, and M. Gurusamy, “Dynamic attack detection and mitigation in IoT
using SDN”, in Proceedings of the 2017 27th International Telecommunication
Networks and Applications Conference (NAC), Melbourne, Australia, pp. 1–6,
2017.
[65] D. Ma, Z. Xu, and D. Lin, “Defending Blind DDoS Attack on SDN Based
on Moving Target Defense”, In: Tian, J., Srivatsa, M. (Eds.), International
Conference on Security and Privacy in Communication Networks, Springer
International Publishing: Cham, Switzerland, pp. 463–480, 2015.
Chapter 6

Extending the Unified Theory

of Acceptance and use of
technology model to understand
the trainees’ acceptance and
usage of Internet of Things (IoT)
by skill development course
Ananta Narayana
DHSS, MNNIT, Allahabad, India

P. S. Birla
FOE, IGNTU, Amarkantak, India

R. K. Shastri
DHSS, MNNIT, Allahabad, India

CONTENTS

6.1 Introduction .................................................................................. 92

6.2 Theoretical background and hypothesis ........................................ 95
6.3 Performance expectancy ................................................................ 96
6.4 Effort expectancy .......................................................................... 97
6.5 Social influence ............................................................................. 97
6.6 Computer self-efficacy ................................................................... 98
6.7 Relative advantage ........................................................................ 99
6.8 Facilitating conditions ................................................................... 99
6.9 Behavior intention and actual usage .............................................. 99
6.10 Research methodology ................................................................ 100
6.11 Measurement .............................................................................. 100
6.12 Data analysis and research findings ............................................. 102
6.12.1 Descriptive analysis ......................................................... 102
6.12.2 Measurement model ........................................................ 102
6.12.3 Structural model .............................................................. 104
6.13 Discussion and conclusion .......................................................... 106
References �������������������������������������������������������������������������������������������� 108

DOI: 10.1201/9781003407300-6 91
92 Ananta Narayana et al.

6.1 INTRODUCTION

Technologies related to the Internet of Things system are expanding rapidly.

In the world, there are billions of physical IoT devices, all of which are
connected to the internet. We can also describe the Internet of Things as a
digital world which is embedded with technologies and connects everything
through its system and communication networks. The Internet of Things
consists of electronic devices and is no longer limited to the original three
connected devices: computers, laptops, and smartphones. The Internet of
Things is beyond sharing data with further internet-enabled devices. It is
important from a technical, social, and economic standpoint to pay atten-
tion to the Internet of Things. Consumer goods, durable goods, transpor-
tation equipment, industrial and utility components, sensors, and other
related items are being integrated with Internet connectivity and potent
data analytics capabilities that promise to transform the way we work, live,
and play. Impressive predictions have been made as to how the Internet of
Things will affect the economy and the number of linked devices, with some
predicting that by 2025 there willbe more than 100 billion connected IoT
devices and the Internet of Things will contribute some $11 trillion to the
global economy. However, there are also significant obstacles raised by these
developments that could prevent its potential advantages from being real-
ized. The public’s attention has already been piqued by attention-grabbing
news concerning the hacking of Internet-connected devices, concerns about
surveillance, and privacy anxieties. New issues with regard to policy, law,
and development are also emerging, in addition to the ongoing technical
difficulties.
The Internet of Things (IoT) is a new paradigm that makes it possible for
electrical devices and sensors to communicate with one another over the
internet in order to make our lives easier. IoT uses the internet and smart
devices to offer creative answers to problems faced by businesses, govern-
ments, and both public and private sectors around the world. It is steadily
gaining in importance and is now pervasive in our daily lives. IoT, as a
whole, is a technological advancement that combines a wide range of smart
systems, frameworks, intelligent devices, and sensors. Additionally, it makes
use of quantum and nanotechnology to achieve previously unthinkable lev-
els of data storage, sensing, cloud computing, processing speed, and artificial
intelligence. To demonstrate the potential effectiveness and applicability of
IoT changes, extensive research studies have been conducted and are avail-
able in the form of scholarly articles, and press reports, both on the internet
and in print form. It could be used as a preliminary backgroundto inform
the development of original, inventive concepts while incorporating security,
assurance, and interoperability into consideration.
The Internet of Things incorporates a wide range of concepts that are intri-
cate and intertwined from several aspects so as to take the advantage of the
Internet of Things and make the technology accessible to the public acrossthe
 Extending acceptance, usage of IoT by skill development course 93

whole of India, the Central Electronics Engineering Research Institute orga-

nizes a six-week Internet of Things skill development course to Industry
Professionals, Faculties, and B.E./B. Tech students from both the 6th semester
onwards and Diploma holders.
For the benefit of society, CSIR-CEERI (CSIR-Central Electronics
Engineering Research Institute) is committed to research and development,
technology development, and academic assistance (including skill develop-
ment training/human resources) in the fields of electronics, ICT, and related
areas. A prestigious institute has excelled in cutting-edge information and
communication technology research and IoT technology is essential for the
future of the industry since it keeps the services linked and operating
effectively.
The goal of the skill development course is to teach participants how to
design and program an Internet of Things (IoT)-based system utilizing real
hardware (a Raspberry Pi-3 and an MSP430 board) and the Linux operat-
ing system with server connectivity. Connecting IoT devices, such as edge
and gateway devices, to the server/cloud, as well as troubleshooting using
the Raspberry Pi-3 and MSP430 board and expected job roles include engi-
neers of embedded systems, technicians of embedded systems, and engineers
of field applications who can troubleshoot IoT-based electronic systems and
products and the creation of small electrical devices based on Internet of
Things (IoT) applications, the creation of Internet of Things (IoT)-based
solutions for smart cities, intelligent transportation and building systems,
environment monitoring and control systems through entrepreneurship.
Course highlights were:

• Expert instruction that is conceptually focused and in-depth, including

pertinent updates from associated research teams.
• Training that maximized competency development was made possible
by scientific study.
• Expert faculty, lab instructors, and highly skilled trainers with relevant
research & development and academic experience.
• Training outcome and competency development is proposed to fulfil
criteria specific to industry and academia.

Learning results include:

• Distinguishing between IoT hype and reality

• Raspberry Pi-3 with MSP430 plus CC2520 and Arduino Mega 2560
programming
• Peripheral Raspberry Pi-3 plus MSP430 interface
• Embedded systems based on the Raspberry Pi-3 and MSP430
• Programming methods using LAMP
• Examine networking technologies to be used in Internet of Things
applications.
94 Ananta Narayana et al.

• Analyze the interconnections between IoT, cloud services, and soft-

ware agents.
• Use practical methods to develop IoT-based projects.

Despite the fact that the idea of using computers, sensors, and networks to
monitor and manage items has been around for some time, the recent con-
vergence of important technological advancements and commercial trends
is ushering in a new era for the “Internet of Things.” IoT promises to intro-
duce a revolutionary, globally networked “smart” future, where connections
between people and their environment and between products and their envi-
ronment will become increasingly intricate. People’s perceptions of what
it means to be “online” may change significantly if the Internet of Things
becomes a common array of connected devices.
A variety of obstacles, including those relating to security, privacy, interop-
erability and standards, legal, regulatory, and rights issues, as well as the
inclusion of emerging economies, could stand in the way of this goal, despite
the fact that the potential repercussions are considerable. There are many
different stakeholders involved in the Internet of Things, and there are many
complicated and changing technological, social, and policy factors. The
Internet of Things is now present in our everyday lives; therefore, it’s impor-
tant to manage its problems, reap its benefits, and lower its hazards.
IoT is an important aspect in the development of the so-called “Internet
Society” because it represents a developing component of how individuals
and organizations are likely to engage with and integrate network connec-
tivity into their personal, social, and professional lives. If we see the develop-
ment of a divisive discussion that contrasts the potential advantages of IoT
with its potential problems, solutions to maximizing the benefits and avoid-
ing the risks of IoT are unlikely to be identified. Instead, to choose the most
efficient course of action, educated involvement, debate, and collaboration
among a variety of stakeholders are required.
In this study we make a critical assessment and incorporation of eight
models of technology from the course related to social psychology to the
same degree as social cognitive theory, theory of reasoned action, innovation
diffusion theory, theory of planned behavior, technology acceptance models,
the model of perceived credibility utilization, a hybrid model combining
constructs from technology acceptance model, the theory of planned behav-
ior and the motivational model have resulted in the advanced technology
model, Unified theory of acceptance and use of technology by (Venkatesh et
al., 2003) plus these multiplicities of illustrious models (Yi et al., 2006)
along with theories are used to explicate this relationship involving trainees’
attitude, perception and behavior intention (Yousafzai, 2012) regarding the
usage of technology (Venkatesh et al., 2012).
The Unified Theory of Acceptance and Use of Technology model is both a
predictive tool of adoption and behavior and a valid research instrument
(Al-Qeisi, 2009). The UTAUT model turns out to be amongst the most
 Extending acceptance, usage of IoT by skill development course 95

extensively used models (Abu-Shanab et al., 2010) because of its parsimony

(Foon & Fah, 2011), its usefulness (Venkatesh & Zhang, 2010) and the
cluster of antecedents (Yuen et al., 2010; Tarhini et al., 2015). Furthermore,
this model is verified to be advanced from the existing technology accep-
tance model (Venkatesh et al., 2003).
The following are explicit research objectives that can provide insights
into the most significant aspects either hindering or facilitating the accep-
tance and usage of the Internet of Things by trainees from the skill develop-
ment course.

• To examine which aspect to be precise performance expectancy, effort

expectancy, social influence, facilitating conditions, computer self-­
efficacy and relative advantage affect trainees’ behavior intention in
the usage of the Internet of Things.
• To identify the characteristics and effectiveness of the relationship
between the constructs and elucidate whichever construct influences
the decision to a greater extent to use the Internet of Things.
• To analyze the incorporation of two constructs, computer self-efficacy
and relative advantage, with a unified theory of acceptance and use of
technology model in providing a robust hypothetical basis in favor of
investigating the acceptance of the Internet of Things.
• To drive relevant suggestions and recommendations to research insti-
tutes in order to ensure a successful training of Internet of Things skill
development course.

In general, this chapter will prove to be useful for the government to per-
suade for adopting and accepting the Internet of Things and for research
institutes to provide more robust internet of things skill development
courses that adhere better to the market demands for professionals in India.
The next section gives atheoretical background which helps in developing
a hypothetical research model that influence the adoption of the Internet of
Things as the implementation of effective and efficient skill is required for
the adoption of innovative technologies (Finis Welch, 1970).

6.2 THEORETICAL BACKGROUND AND HYPOTHESIS

The Internet of Things (IoT) was first introduced by Kevin Ashton in 1999.
At the time he defined the IoT as a network of radio-frequency identifica-
tion (RFID)-enabled technology that is uniquely identifiable and intercon-
nected (Pretz, 2013). The Internet of Things (IoT) was generally described as
a “dynamic global network infrastructure with self-configuring capabilities
based on standards and interoperable communication protocols; physical
and virtual ‘things’ in an IoT have identities and attributes and are capa-
ble of using intelligent interfaces and being integrated as an information
96 Ananta Narayana et al.

Performance Expectancy

Effort Expectancy

Social Influence

Behavior Intention Actual Usage

Computer Self Efficacy

RelativeAdvantage

Facilitating Conditions

Figure 6.1 Hypothetical research model (adapted from Venkatesh et al., 2003).

network” (Kirtsis, 2011). In essence, the Internet of Things (IoT) can be

thought of as a superset of interconnecting objects that can be individu-
ally identified using current near-field communication (NFC) technologies.
The terms “Internet” and “Things” refer to a global network of intercon-
nected devices that uses networking, information processing, and sensory
technologies. This network may represent the future of Information and
Communications Technology (ICT) (Kranenburg & Anzelmo, 2011).
Despite disagreements over its definition, the Internet of Things has been
widely debated and relevant technologies have been created rapidly by a
number of institutions. In particular, intelligent sensing and wireless com-
munication techniques have entered the IoT, opening up new problems
and research frontiers. The International Telecommunication Union (ITU)
explored IoT implications, growing difficulties, emerging markets, and
enabling technologies (ITU, 2013).
This section shows the hypothetical research model in Figure 6.1, which
is proposed on the basis of the theoretical background and provides a com-
prehensive discussion of presented constructs

6.3 PERFORMANCE EXPECTANCY

Yu (2012) states that performance expectancy shows a significant effect on

the trainees’ behavior intention in terms of their actual usage of the Internet
of Things. Performance Expectancy measures the degree to which a trainee
accepts the Internet of Things as the key to keeping the services intercon-
nected and working effectively (Al-Somali et al., 2009). Correspondingly,
Lee (2009) determined performance expectancy as an important factor in
 Extending acceptance, usage of IoT by skill development course 97

understanding the usage of the Internet of Things. With regard to the per-
spective of the current research study, especially predictable to facilitate the
trainees by providing an Internet of Things skill development course to be
valuable, so that they can possibly use and accept the Internet of Things.
Therefore, the researcher suggests the below hypothesis:
H1. 
Performance Expectancy significantly affects trainees’ Behavior
Intention on the usage of the Internet of Things

6.4 EFFORT EXPECTANCY

Venkatesh et al. (2003) define effort expectancy as the degree of effort in

which trainees are affiliated with the actual usage of technology. In accor-
dance with Venkatesh’s model, effort expectancy positively modifies the
behavior intention regarding technology usage. Trainees are more likely to
adopt if it does not require much effort and finds the Internet of Things easy
to use (Yu, 2012). Yoon and Steege (2013) found that effort expectancy has
an effect and proves the strongest predictor of behavior intention. As per,
Martins et al. (2014) effort expectancy shows a positive significant effect on
behavior intention. Consequently, Zhou et al. (2010) fail to maintain the
positive association between effort expectancy and behavior intention. The
present research study predicts that if the trainees find the Internet of Things
skill development course relevant and easy to use, then they possibly use and
adopt it. Contradictorily, if the trainees find the Internet of Things irrelevant
and not easy to use, then the trainees will show less interest in adopting it.
Hence, the researcher hypothesizes the subsequent hypothesis:
H2. Effort Expectancy will significantly affect trainees’ Behavior Intention
in the usage of the Internet of Things

6.5 SOCIAL INFLUENCE

Ajzen (1991) determined Social Influence as a particular belief of an indi-

vidual who is important to the other individual and does not assume the
behavior in their perception like social norms mentioned in renowned theo-
ries theory of reasoned action, technology acceptance model 2, innovation
diffusion theory and theory of planned behavior (Amin, 2009). To put it
another way, social influence indicates the societal pressure that comes from
the surroundings and uncontrollable factors and can influence the behavior
and perception of individuals in certain actions, such as the judgment of
superiors, friends, or relatives (Tarhini et al., 2013b; Tarhini & Liu, 2014).
While Davis (1989) excluded the social influence factor due to theoreti-
cal and measurement tribulations from the original technology acceptance
98 Ananta Narayana et al.

model and was added further in the technology acceptance model 2 due
to its significance in elucidating the external influence of other people on
the behavior of self (Venkatesh & Zhang, 2010). Social influence persuades
technology acceptance (Venkatesh & Zhang, 2010). Social influences refers
to the degree to which the use of the Internet of Things boosts a trainee’s
social status (Moore & Benbasat, 1991). As per Galan et al. (2013), trainees
believe that adopting the Internet of Things will improve their performance
for a particular social group (Im & Kang, 2011). Most of the experimental
research in information and communication technology (Yousafzai, 2012)
found social influence to be a significant precursor of behavior intention
(Venkatesh et al., 2003) with regard to the Internet of Things (Kesharwani
& Bisht, 2012). Mentioned researches indicate a positive significant rela-
tionship which is empirically allying linking social influence and behavior
intention of trainees’ in adopting the Internet of Things.
H3. Social Influence indicates significant influence on trainees’ behavior
intention in using the Internet of Things

6.6 COMPUTER SELF-EFFICACY

The word self-efficacy is introduced by Bandura (1986). In his social cog-

nitive theory he explained self-efficacy as individuals’ belief in their capa-
bilities to accomplish a task confidently within a specified period. Inspired
by the idea of the conceptualization of self-efficacy, Compeau and Higgins
(1995a) derived the term computer self-efficacy and further explained it as
an individual’s belief in their capabilities to complete a computer-related
assignment confidently within a specified period (Compeau& Higgins,
1995b). In the general computing realm, Marakas, Yi, & Johnson (1998)
also illustrated computer self-efficacy as individuals’ ability in performing
computer-specific tasks. The person who possesses higher computer self-
efficacy to own higher confidence in attaining computer-associated tasks
successfully. Fagan et al. (2004) specified how computer self-efficacy is posi-
tively inclined by motivating teams, colleagues, and management. Computer
self-efficacy also controls feelings of anxiety (Tung & Chang, 2008) as indi-
viduals who possess a high degree of computer self-efficacy to feel less anxi-
ety when compared with a lower level of self-efficacy. Previous literature
found that computer self-efficacy influences individual behavior in adapting
to information technology, information, and communication technology;
internet banking, etc. so considering these studies researcher inculcates that
computer self-efficacy can also be taken as a construct in influencing the
behavior intention of individuals in adopting the Internet of Things. Hence,
we can formulate the hypothesis as below:
H4. Computer Self-Efficacy will positively affect trainees’ behavior inten-
tion towards usage of the Internet of Things.
 Extending acceptance, usage of IoT by skill development course 99

6.7 RELATIVE ADVANTAGE

Relative advantage is a degree that demonstrates the advantage of new

technology over existing one and leads to an increase in the adoption rate
(Ekong et al., 2012). Rogers (1995) defined relative advantage as “the
degree to which an innovation is perceived as being better than the idea it
supersedes.” As per Tornatzky and Klein (2012), relative advantage shows
a significant positive relation with the adoption of innovative technology.
Relative advantage shows considerable advantage in the adoption of the
Internet of Things (Weber, 2010). Tu (2018) further stated that, as compared
to existing offerings, new innovation and technology give considerable ben-
efits. We have come to know from existing literature that relative advantage
can prove to be a positive antecedent in influencing behavior intention for
technology adoption. Therefore, we articulate the hypothesis as
H5. Relative advantage indicates a positive significant influence on train-
ees’ behavior intention toward usage of the Internet of Things

6.8 FACILITATING CONDITIONS

Venkatesh et al. (2003) elucidate facilitating condition as the degree to

which people think about the departmental and technological infrastruc-
ture that plays a vital role in technology usage. Undoubtedly, Zhou et al.
(2010) indicated that the actual usage of the Internet of Things needed req-
uisite resources and technological infrastructure. Further, certain amenities
are never provided generally at any cost with regard to trainees. There are
many possibilities for organizations to promote the Internet of Things by
eliminating barriers to usage and persuading adoption. Trainees perceive
the Internet of Things to be a crucial service and required the use of the
most recent technology. Facilitating conditions can be measured on the basis
of the trainees’ perception that will they can be able to use the requisite
resources and services related to the Internet of Things. Consequently, it is
predictable that the necessary requisite resources will effectuate the trainees
to use and accept the Internet of Things. Hence, based on the above theoreti-
cal background our next hypothesis would be as stated below
H6. 
Facilitating conditions indicate a positive influence on trainees’
actual usage of the Internet of Things

6.9 BEHAVIOR INTENTION AND ACTUAL USAGE

Venkatesh et al. (2003, 2012) elucidated behavior intention to be a direct

precursor of usage behavior and provide an indicator (Tarhini et al., 2013a)
in relation to measuring individuals’ ability in adopting a particular behavior
100 Ananta Narayana et al.

(Venkatesh & Zhang, 2010). Actual behavior is the clear, obvious reaction
in a particular situation regarding a particular target (Ajzen, 1991). There is
a substantial indication of the significant effect of behavior intention (Im &
Kang, 2011) on actual usage in studies related to technology adoption
(Ma et al., 2010). In recent times, it has extended to the perspective of the
Internet of Things (Yu, 2012). As a consequence, the researcher postulates
the below hypothesis:
H7. Trainees’ behavior intention indicates a positive significant influence
on their actual usage of the Internet of Things.

6.10 RESEARCH METHODOLOGY

As per previous research studies on the UTAUT Model and the Internet of
Things, a quantitative research approach is immersed by the researchers to
test the proposed hypothetical model. A structured questionnaire consisting
of 38 questions is employed for data collection through convenient sampling
from the trainees of the Internet of Things skill development course orga-
nized by CSIR-Central Electronics Engineering Research Institute. Trainees
were requested to fill out the presented questionnaire on the basis of his/
her judgment and attitude about the acceptance and use of the Internet of
Things. Each trainee took 15 minutes to fill out the complete questionnaire.
The researcher distributed a total of 700 questionnaires to the trainees;of
these some 522 were returned to the researcher,meaning a response rate
of some 70%. After the researcher review 14 questionnaires were found
to be invalid, meaning that 508 responses were selected for final analysis.
Table 6.1 shows the beneficiaries’ demographic summary; with regard to
gender, 63% of respondents are male and 37% are female, their age ranges
from 18 to 60 years. With regard to education, 8% of them had secondary
education, 34% did a diploma and 58% of them are bachelor’s and above,
of which 52% of total respondents have experience in using the Internet of
Things, 36% have experience of using the Internet of Things to some extent,
and 12% have no experience of usage of the Internet of Things.

6.11 MEASUREMENT

The scales were adopted from existing literature associated with the uni-
fied theory of acceptance and use of technology model pertaining to the
current investigation (Foon& Fah, 2011) and prior empirical studies were
taken related to computer self-efficacy and relative advantage (Im & Kang,
2011) to protect the validity and reliability of presented items. Purposely,
performance expectancy, and effort expectancy were measured by means
 Extending acceptance, usage of IoT by skill development course 101

Table 6.1 Beneficiaries demographic summary

Categorization Freque ncy Percent
Gender
Male 321 63.10
Female 187 36.81
Age
18 to 25 217 42.71
26 to 35 134 26.37
36 to 45 105 20.66
Above 46 52 10.26
Educational level
Bachelors and above 296 58.26
Diploma 170 33.46
Secondary education 42 8.28
Experience in usage of Internet of Things
Experienced 264 51.96
Some experience 185 36.43
No Experience 59 11.61

of 5 items, while behavior intention, social influence, and facilitating condi-

tions were measured by 4 items. 2 items were adapted from the work of
Venkatesh et al. (2003, 2012). Additionally, computer self-efficacy and rela-
tive advantage were measured using 5 items and the scales were adopted
from the works of Murphy et al. (1989) and Moore& Benbasat (1991)
respectively. In the direction of measuring particular factors of this pro-
posed hypothetical research model, the researcher used a seven-point Likert
scale, which ranges from 1 to 7; that is, from strongly disagree to strongly
agree, and a nominal scale is used to measure the demographic figures con-
cerning the beneficiaries likewise gender, age, experience, and education.
The researcher is unable to take the actual use of the Internet of Things
through the trainee’s log file as it was not feasible so they measured actual
usage using a self-structured questionnaire. The foremost question mea-
sured how frequently the trainees are using the Internet of Things while the
second question measured the average usage of the Internet of Things by
the trainees. The questionnaire is well structured in English, and the items
are adapted from the perceptive UTAUT model.Although before proceeding
study further, the researchers executed a pilot test of 50 respondents who
were randomly chosen for revising and modifying the questionnaire items
and establishing content validity and reliability. Several items are examined
and adjusted through the results of pilot testing.
102 Ananta Narayana et al.

6.12 DATA ANALYSIS AND RESEARCH FINDINGS

6.12.1 Descriptive analysis
Table 6.2 presents descriptive statistics which affect all constructs in the
hypothetical research model and each mean was > 4.49, indicating that the
majority of the respondents respond positively to the structures measured
in the study. Furthermore, Cronbach’s α score pointed toward the strong
internal reliability of all constructs.

6.12.2 Measurement model
The present study is based on a multiple steps approach which examines the
relationship between factors of a structural model from the study (Anderson
& Gerbing, 1988). The researcher analyzed the measurement model to check
whether the instrument is reliable and valid for testing research hypotheses
in the proposed hypothetical model (Arbuckle, 2009). Hence, to contem-
plate the measurement model fit, researchers have to initially analyze the
confirmatory factor on the basis of AMOS 20.0 and subsequently calcu-
late the validity appertaining to the measurement model (Schumacker &
Lomax, 2010). To offer an approximate calculationof the model’s factors,
the present examination accepts the maximum-likelihood method where all
of the investigation was carried out on variance and covariance matrices.
Conversely, before the analysis it is assumed that it is crucial to test for mul-
ticollinearity, and multicollinearity depends on a higher correlation between
the variables. Tabachnick and Fidell (2007) illustrated that correlation values
in the approximate range of 0.8 or 0.9 are supposed to be highly problem-
atic, whereas a correlation value of 0.7 or higher is supposed to be a cause of
study. As per Pallant (2010), correlation value tolerance and variance infla-
tion aspect determined the presence of multicollinearity. There would be no
chance of multicollinearity on the assumption that the tolerance value is
in excess of 0.10 and the value of the variance inflation factor is below 3.0.

Table 6.2 Measures of central tendency and variability

Constructs Mean Standard deviation Cronbach’s α
Performance Expectancy 4.61 1.24 0.912
Effort Expectancy 5.41 1.12 0.909
Social Influence 4.26 1.29 0.836
Computer Self Efficacy 4.49 1.38 0.918
Relative Advantage 4.92 0.99 0.862
Facilitating Conditions 5.13 1.27 0.793
Behavior Intention 4.97 1.39 0.857
Actual Usage 5.24 1.37 0.769
 Extending acceptance, usage of IoT by skill development course 103

The sample in our study shows the nonexistence of multicollinearity as the

variance inflation value cause is below 3.0 and the tolerance value is more
than 0.10. Hairetal and Kline (2010) suggested various fit indices in support
of evaluating the model’s goodness-of-fit. This was determined by χ2, which
is termed as the minimum fit function. The fraction of χ2 is static to its degree
of freedom, which is indicative of an acceptable fit (χ2/df) (Hu & Bentler,
1999; Carmines & McIver, 1981). As per the recommendation of Hair et al.
(2010), there is a particular scale of additionally fit indices those indices
consist of “goodness-of-fit indices; parsimony normal fit indices; normal fit
indices; root mean square residuals; the root mean square error of approxi-
mation; adjusted goodness-of-fit indices; and comparative fit i­ ndices”. From
the initial measurement model, researchers removed a few indicators, Social
Influence 3, Facilitating Conditions 4, Computer Self Efficacy 5 as well as
relative Advantage 6 to sustain a good fit between the model and data.
Table 6.3 clearly lists the actual estimate of the model fit index in the sug-
gested range below, hence the researcher can continue to evaluate and vali-
date the discriminant validity and convergent validity along with reliability
for checking the adequacy related to psychometric values pertaining to the
measurement model. Convergent validity validates whether all the factors
got rejected through its indicators conducive to ensuring unidirectional mul-
tiple-item constructs by eliminating unpredictable indicators (Gefen et al.,
2000; Bollen, 1989). Discriminant validity evaluates a degree to measure the
concepts of distinct statistics while composite reliability, maximum shared
squared variance, average variance extracted and average shared squared
variance can all assess reliability, convergent validity, and discriminant
validity. To establish reliability, composite reliability should be at least 0.6
and preferably above 0.7 and to establish convergent validity the average

Table 6.3 Outline of absolute measurement and structural model fit index
Measurement Structural
Fit index Recommended valuea model model

Minimum it Function χ2 NS at p is less than 0.05 582 597

Degrees of Freedom n/a 347 356
The ratio of the χ2 static to its less than 5 preferable 2.667 2.669
degree of freedom (χ2/df) less than 3
Goodness-of-Fit Is greater than .9 0.936 0.938
Adjusted Goodness-of-Fit Is greater than .8 0.885 0.887
Comparative Fit Index Is greater than .9 0.945 0.944
Root Mean Square Residuals Is greater than .10 0.086 0.088
Root Mean Square Error of Is greater than .09 0.065 0.065
Approximation
Normed it Index Is greater than .9 0.950 0.946
a Hu and Bentler (1999) Kline (2010) Hair et al. (2010).
104 Ananta Narayana et al.

variance extracted should be at least 0.5 (50% of the variance of indicators

has to be accounted for by the latent variables) and composite reliability is
greater than the average variance exceeds. Whereas discriminant validity is
supported if maximum shared square variance is less than average variance
extracted (Hair et al., 2010). If the extraction of average variance exceeds
the shared squared variance and the maximum shared square variance,the
resulting discriminant validity is supported. Table 6.4 shows constructs of
composite reliability, ranging from 0.76 to 0.91 and exceeding the thresh-
old value of 0.7; at the same time, average variance extracted ranging from
0.53 to 0.77, all of which is above 0.5, resulted in adequate internal consis-
tencies and also supported convergent validity. Furthermore, the maximum
shared square variance for all constructs is less than the average variance
extracted, resulting in sufficient discriminant value which is illustrated on
the basis of a higher square root value of extraction of the average variance
than its correlation value. Based on the examination of the measurement
model, it is concluded that each variable represents a reliable and valid fac-
tor. Therefore the next step is to assesses the structural model in order to
test the research model and examine the hypotheses. Consequently, the final
measure is headed toward a structural model for testing the hypothetical
research model.

6.12.3 Structural model
Owing to the measurement model’s criteria, researchers evaluate the good-
ness-of-fit for measuring the proposed hypothetical research model, and,
according to Table 6.3, the results of both the structural and the measure-
ment models were found to be the same, which again showed good fit data.
Hence, we continue to study the hypothesized relations of the factors of
the proposed hypothetical model. As we see in Table 6.5, all hypotheses
except H2 were supported consequently due to the path coefficient. Notably,
performance expectancy, effort expectancy, social influence, and computer
self-efficacy all indicated a significant positive effect toward behavior inten-
tion about the usage of the Internet of Things and performance expectancy
shows the maximum influence on the correlation with behavior intention.
The presented results supported Hypothesis-1, Hypothesis-3, Hypothesis-4
and Hypothesis-5 unpredictably; the path coefficient from effort expectancy
to behavior intention was not significant. As a result, this study failed to
find support for Hypothesis-2. Performance expectancy, social influence,
computer self-efficacy, and relative advantage to account for 63% of the
variance of behavior intention than other added factors. Contrary to the
effect of the statistical analysis of the presented proposed model explained
in the context of facilitating condition and behavior intention indicated sig-
nificantly influence the actual usage of the system and jointly accounted for
66% of the variance in actual usage, with behavior intention contributing
the most compared to facilitating conditions, consequently, Hypothesis-6
Table 6.4 Construct reliability, convergent and discriminant validity

Extending acceptance, usage of IoT by skill development course

Maximum Average
Average Shared Shared
Composite Variance Square Square Computer Performance Effort Social Relative Facilitating Actual Behavior
Constructs Reliability Extracted Variance Variance Self-Efficacy Expectancy Expectancy Influence Advantage Conditions Usage Intention

Computer 0.924 0.783 0.47 0.276 0.889

Self-
Efficacy
Performance 0.916 0.674 0.327 0.229 0.373 0.87
Expectancy
Effort 0.924 0.759 0.339 0.277 0.579 0.574 0.879
Expectancy
Social 0.846 0.542 0.273 0.189 0.497 0.463 0.386 0.724
Influence
Relative 0.873 0.624 0.487 0.286 0.481 0.526 0.532 0.436 0.788
Advantage
Facilitating 0.893 0.753 0.433 0.292 0.657 0.364 0.586 0.468 0.522 0.877
Conditions
Actual Usage 0.774 0.542 0.446 0.242 0.523 0.394 0.389 0.324 0.438 0.594 0.723
Behavior 0.874 0.717 0.487 0.347 0.527 0.576 0.584 0.538 0.584 0.537 0.574 0.854
Intention

105
106 Ananta Narayana et al.

Table 6.5 Consequence of path coefficients

Path
Path co-efficient Results

Hypothesis-1 Performance Expectancy→ Behavior 0.271** Supported

Intention
Hypothesis-2 Effort Expectancy → Behavior Intention 0.084 Not
supported
Hypothesis-3 Social Influence→ Behavior Intention 0.223** Supported
Hypothesis-4 Computer Self-efficacy → Behavior 0.143* Supported
Intention
Hypothesis-5 Relative Advantage → Behavior Intention 0.228** Supported
Hypothesis-6 Facilitating Conditions → Actual Usage 0.186** Supported
Hypothesis-7 Behavior Intention → Actual Usage 0.464*** Supported
* p is less than 0.05;
** p is less than 0.01;
*** p is less than 0.001; NS p is greater than 0.01

and Hypothesis-7 were supported. The proposed hypothetical research

model of the present study elucidates more variances of behavior intention
and actual usage of the Internet of Things.

6.13 DISCUSSION AND CONCLUSION

A new paradigm called the Internet of Things (IoT) has transformed tradi-
tional living into a high-tech lifestyle. The changes brought about by IoT
include smart cities, smart homes, pollution control, energy conservation,
smart transportation, and smart industries. Many important research proj-
ects and investigations have been carried out in an effort to advance technol-
ogy through IoT. Globally, academics and developers are interested in recent
IoT breakthroughs. Researchers and IoT developers are collaborating to
expand the technology to the fullest possible extent and to produce improve-
ments in the system as a whole. The present research study is determined
by extending the unified theory of acceptance and usage of technology by
integrating constructs such as computer self-efficacy and relative advantage
which study the indicators affecting trainees’ behavior intention in terms
of usingthe Internet of Things. The findings of the study support the effi-
ciency and aptitude of Venkatesh’s technology model,both theoretically and
empirically, to be a functional theoretical model for giving insight into train-
ees’ acceptance of the Internet of Things. Path coefficients of the proposed
hypothetical model indicated statistical significance between effort expec-
tancy and behavior intention. In particular, the above outcomes indicated
that the trainees’ intention in terms of the acceptance and usage of the sys-
tem of the Internet of Things could be significantly affected by performance
 Extending acceptance, usage of IoT by skill development course 107

expectancy, social influence, computer self-efficacy, and relative advantage.

By contrast, effort expectancy was not found to be particularly influential
with regard to behavior intention in using the Internet of Things. The results
of this study also indicated that facilitating conditions and behavior inten-
tion, respectively, proved a significant antecedent of the actual use of the
Internet of Things. Taken as a whole, the proposed research model attained
acceptable fit and account for 63% and 66% of the variances, respectively,
for behavior intention and actual use i.e., figures normally higher than the
novel model unified theory of acceptance and use of technology. While
taking performance expectancy into consideration it proved the strongest
antecedent that reflects the trainees’ perception for using the Internet of
Things in a better way. For that reason, skill development course instruc-
tors need to enhance the effectiveness of the Internet of Things system. The
results of the study specify that social influence has a positive significant
effect on behavior intention, suggesting that trainees become influenced by
peer pressure.Thus, skill development course instructors should understand
that social influence has an importanteffecton trainees’ intention to accept
the Internet of Things. The outcome of the present research confirms that
facilitating conditions played a significant role in the actual usage of the
Internet of Things system, so for that reason there is a need to invest more
on information and communication technology infrastructure.For instance,
after the training period skill development course instructors should also
provide facilities, such as full infrastructure training centers to increase the
skills in using the Internet of Things so that the interest of trainees in adopt-
ing skills might increase. Whereas effort expectancy unexpectedly shows an
insignificant predictor of trainees’ intention to usethe Internet of Things as
the trainees give priority to the perceived usefulness of the system rather
than its ease of use. There is thus a need to design a more user-friendly sys-
tem which encourages trainees to adopt the system. Consequently, this study
proved the importance of indicating a significant positive influence between
computer self-efficacy and behavior intention to derive interest in adopting
the Internet of Things since computer self-efficacy is already a proven factor
in enhancing performance when dealing with computers (Cocorada, 2014).
Prior command of computer self-efficacy is thus needed before any such
training coursesince it leads to a lower level of anxiety during training and
perceived high levels of acknowledged usefulness towardthe adoption of
technology (Downey & Kher, 2015). Accordingly, empirical verification of
the current research demonstrates that relative advantage has proven a sig-
nificant factor in deriving behavior intention in using the Internet of Things
because it leads to an improved use of resources, efficient operation manage-
ment, cost-effective operation, and improved performance. Therefore, gov-
ernment and research institutes need to understand that the most important
factor concerns trainees’ perception to adopt the Internet of Things as the
rate of acceptance of the technology is directly proportional to its usage. The
greater the usage, the greater will be its acceptance. Expertise in the Internet
108 Ananta Narayana et al.

of things is required for both upskilling and career advancement. Hence,

research institutes should emphasize on providing more robust Internet
of Things skill development courses that will better addressthe market
demands for professionals in India.

REFERENCES

E. Abu-Shanab, J. M. Pearson, and A. J. Setterstrom, “Internet banking and custom-

ers’ acceptance in Jordan: The unified model’s perspective,” Communications of
the Association for Information Systems, vol. 26(1), pp. 23, 2010.
I. Ajzen, “The theory of planned behavior,” Organizational Behavior and Human
Decision Processes, vol. 50(2), pp. 179–211, 1991.
K. I. Al-Qeisi, “Analysing the use of UTAUT model in explaining an online behavior:
internet banking adoption,” unpublished doctoral dissertation, Brunel Business
School, Brunel University, London, 2009.
S. A. Al-Somali, R. Gholami, and B. Clegg, “An investigation into the acceptance of
online banking in Saudi Arabia,” Technovation, vol. 29(2), pp. 130–141, 2009.
H. Amin, “An analysis of online banking usage intentions: An extension of the tech-
nology acceptance model,” International Journal of Business and Society, vol.
10(1), pp. 27–40, 2009.
C. Anderson, and D. W. Gerbing, “Structural equation modeling in practice: Areview
and recommended two-step approach,” Psychological Bulletin, vol. 103(3),
pp. 411–423, 1988.
J. Arbuckle, AMOS 18 User’s Guide, SPSS Incorporated, Armonk, NY, 2009.
A. Bandura, Social foundations of thought and action: A Social Cognitive
Theory,Prentice Hall, Upper Sadle River, NJ, 1986.
K. A. Bollen, Structural equations with latent variables,John Wiley and Sons, Oxford,
1989.
G. Carmines, and J. P. McIver, Analyzing models with unobserved variables: Analysis
of covariance structures,SAGE, Thousand Oaks, CA, 1981.
S. Cocorada, “Computer anxiety, computer self-efficacy and demographic variables,”
the International Scientific Conference eLearning and Software for Education,
vol. 1, Carol I, National Defence University, 2014.
D. R. Compeau, and C. A. Higgins, “Computer self-efficacy: Development of a mea-
sure and initial test,” MIS Quarterly, vol. 19(2), pp. 189–211, 1995a.
D. R. Compeau, and C. A. Higgins, “Application of social cognitive theory to train-
ing for computer skills,” Information Systems Research, vol. 6(2), pp. 118–143,
1995b.
F. D. Davis, “Perceived usefulness, perceived ease of use, and user acceptance of
information technology,” MIS Quarterly, vol. 13(3), pp. 319–340, 1989.
J. Downey, and H. V. Kher, “A longitudinal examination of the effects of computer
self-efficacy growth on performance during technology training,” Journal of
Information Technology Education, vol. 14. pp. 91–111, 2015.
U. O. Ekong, P. Ifenedo, C. K. Ayo, and A. Ifenedo, “E-commerce adoption in
Nigerian businesses: An analysis using the technology-organization-environmental
framework,” In Leveraging Developing Economies with the Use of Information
Technology: Trends and Tools, IGI Global, Hershey, PA, pp. 156–178, 2012.
 Extending acceptance, usage of IoT by skill development course 109

M. H. Fagan, S. Neill, and B. R. Wooldridge, “An empirical investigation into the

relationship between computer self-efficacy, anxiety, experience, support and
usage,” Journal of Computer Information Systems, vol. 44(2), pp. 95–105, 2004.
Y. S. Foon, and B. C. Y. Fah, “Internet banking adoption in Kuala Lumpur: An appli-
cation of UTAUT model,” International Journal of Business and Management,
vol. 6(4), pp. 161–167, 2011.
P. Galan, M. Giraud, and L. Meyer-Waarden, “A theoretical extension of the
Technology Acceptance Model (TAM) to explain the adoption and the usage of
new digital services,” Annual conference of the European Marketing Academy,
42, Istanbul, 2013.
D. Gefen, W. Straub, and M. C. Boudreau, “Structural equation modeling and regres-
sion: Guidelines for research practice,” Communication of the Association for
Information Systems, vol. 4(1), pp. 1–77, 2000.
J. F. J. Hair, W. C. Black, B. J. Babin, R. E. Anderson, and R. L. Tatham, Multivariate
data analysis, Prentice Hall, Upper Saddle River, NJ, 2010.
L. T. Hu, and P. M. Bentler, “Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives,” Structural Equation
Modeling: A Multidisciplinary Journal, vol. 6(1), pp. 1–55, 1999.
S. H. Im, and M. S. Kang, “An international comparison of technology adoption:
Testing the UTAUT model,” Information & Management, vol. 48(1), pp. 1–8,
2011.
ITU, “The internet of Things, International Telecommunication Union (ITU),
Internet Report,” available from http://www.itu.int/dms_pub/itu-s/opb/pol/
S-POL-IR.IT-2005-SUM-PDF-E.pdf, 2013.
A. Kesharwani, and S. S. Bisht, “The impact of trust and perceived risk on inter-
net banking adoption in India: an extension of technology acceptance model,”
International Journal of Bank Marketing, vol. 30(4), pp. 303–322, 2012.
D. Kirtsis, “Closed-loop PLM for intelligent products in the era of the internet of
things,” Computer-Aided Design, vol. 43(5), pp. 479–501, 2011.
R. B. Kline, Principles and practice of structural equation modeling, The Guilford
Press, New York, NY, 2010.
R. Kranenburg, and E. Anzelmo, “The internet of things, 1st Berlin symposium on
internet and society,” Oct25–27, 2011.
M. C. Lee, “Factors influencing the adoption of internet banking: An integration of
TAM and TPB with perceived risk and perceived benefit,” Electronic Commerce
Research and Applications, vol. 8(3), pp. 130–141, 2009.
Z. Ma, L. Ma, and J. Zhao, “Evidence on E-banking quality in the China commercial
bank sector,” Global Journal of Business Research, vol. 5(2), pp. 73–83, 2010.
G. M. Marakas, M. Y. Yi, and R. D. Johnson, “The multilevel and multifaceted
character of computer self-efficacy: Toward clarification of the construct and an
integrative framework for research,” Information Systems Research, vol. 9(2),
pp. 126–163, 1998.
C. Martins, T. Oliveira, and A. Popovič, “Understanding the Internet banking adop-
tion: Aunified theory of acceptance and use of technology and perceived risk appli-
cation,” International Journal of Information Management, vol. 34(1), pp. 1–13,
2014.
G. C. Moore, and I. Benbasat, “Development of an instrument to measure the per-
ceptions of adopting an information technology innovation,” Information Systems
Research, vol. 2(3), pp. 173–191, 1991.
110 Ananta Narayana et al.

A. Murphy, D. Coover, and S. V. Owen, “Development and validation of the com-

puter self-efficacy scale,” Educational and Psychological Measurement, vol. 49(4),
pp. 893–899, 1989.
J. Pallant, SPSS survival manual: A step-by-step guide to data analysis using SPSS
(4th ed.), Open University Press, Berkshire, 2010.
K. Pretz, “The Next Evolution of the Internet,” available from http://theinstitute.ieee.
org/technology-focus/technology-topic/the-next-evolution-of-the-internet, 2013.
E. M. Rogers, Diffusion of innovations (4th ed.), Free Press, New York, 1995.
R. E. Schumacker, and R. G. Lomax, A beginner’s guide to structural equation mod-
eling, Routledge, Lawrence Erlbaum, York, 2010.
G. Tabachnick, and L. S. Fidell, Using multivariate statistics (5th ed.), Allyn & Bacon,
Needham Height, MA, 2007.
A. Tarhini, K. Hone, and X. Liu, “Factors affecting students’ acceptance of e-learning
environments in developing countries: A structural equation modeling approach,”
International Journal of Information and Education Technology, vol. 3(1), pp. 54–59,
2013a.
A. Tarhini, K. Hone, and X. Liu, “User acceptance towards web-based learning
systems: Investigating the role of social, organizational and individual factors in
European higher education,” Procedia Computer Science, vol. 17(1), pp. 189–197,
2013b.
A. Tarhini, T. Teo, and T. Tarhini, “A cross-cultural validity of the E-learning accep-
tance measure (ElAM) in Lebanon and England: Aconfirmatory factor analysis,”
Education and Information Technologies, vol. 21(5), pp. 1269–1282, 2015.
K. H. Tarhini, and X. Liu, “The effects of individual differences on e-learning users’
behavior in developing countries: Astructural equation model,” Computers in
Human Behavior, vol. 41(1), pp. 153–163, 2014.
G. Tornatzky, and K. J. Klein, “Innovation characteristics and innovation adoption
implementation: A meta-analysis of findings. IEEE Transactions on Engineering
Management,” vol. 29(1), pp. 28–45, 2012.
M. Tu, “An exploratory study of internet of things (IoT) adoption intention in
logistics and supply chain management,” The International Journal of Logistics
Management, vol. 29(1), pp. 131–151, 2018.
F. C. Tung, and F. C. Chang, “Nursing student's behavioral intention to use online
courses: Aquestionnaire-based survey,” International Journal of Nursing Studies,
vol. 45(9), pp. 129–309, 2008.
V. Venkatesh, M. G. Morris, G. B. Davis, and F. D. Davis, “User acceptance of informa-
tion technology: Toward a unified view,” MIS Quarterly, vol. 27(3), pp. 425–478,
2003.
V. Venkatesh, J. Thong, and X. Xu, “Consumer acceptance and use of information
technology: Extending the unified theory of acceptance and use of technology,”
MIS Quarterly, vol. 36(1), pp. 157–178, 2012.
V. Venkatesh, and X. Zhang, “Unified theory of acceptance and use of technology: US
vs. China,” Journal of Global Information Technology Management, vol. 13(1),
pp. 5–27, 2010.
R. H. Weber, “Internet of things–new security and privacy challenges,” Computer
Law &Security Review, vol. 26(1), pp. 23–30, 2010.
F. Welch, “Education in production,” Journal of Political Economy, vol. 78(1),
pp. 35–59, 1970.
 Extending acceptance, usage of IoT by skill development course 111

M. Y. Yi, J. D. Jackson, J. S. Park, and J. C. Probst, “Understanding information

technology acceptance by individual professionals: Toward an integrative view,”
Information & Management, vol. 43(3), pp. 350–363, 2006.
H. S. Yoon, and L. M. B. Steege, “Development of a quantitative model of the impact
of customers’ personality and perceptions on internet banking use,” Computers in
Human Behavior, vol. 29(3), pp. 1133–1141, 2013.
S. Y. Yousafzai, “A literature review of theoretical models of internet banking adop-
tion at the individual level,” Journal of Financial Services Marketing, vol. 17(3),
pp. 215–226, 2012.
C. S. Yu, “Factors affecting individuals to adopt mobile banking: Empirical evidence
from the UTAUT model,” Journal of Electronic Commerce Research, vol. 13(2),
pp. 104–121, 2012.
Y. Y. Yuen, P. Yeow, N. Lim, and N. Saylani, “Internet banking adoption: Comparing
developed and developing countries,” Journal of Computer Information Systems,
vol. 51(1), pp. 52–61, 2010.
T. Zhou, Y. Lu, and B. Wang, “Integrating TTF and UTAUT to explain mobile bank-
ing user adoption,” Computers in Human Behavior, vol. 26(4), pp. 760–767, 2010.
Chapter 7

Intelligent approaches for disease

detection and prevention
Saumya Yadav
Indraprastha Institute of Information Technology (IIIT), Delhi, India

Deepak Chandra Joshi

Shiv Nadar University, Ghaziabad, India

Abhishek Joshi
Sungkyunkwan University, Seoul, South Korea

Sanjay Mathur and Manoj Bhatt

College of Technology, G.B.P.U.A&T., Pantnagar, India

CONTENTS

7.1 Introduction .................................................................................. 113

7.2 Artificial intelligence, Internet of Things, and blockchain
technology .................................................................................... 115
7.2.1 Artificial intelligence .......................................................... 115
7.2.2 Machine learning ............................................................... 116
7.2.3 Internet of Things ............................................................... 122
7.2.4 Blockchain technology ....................................................... 122
7.3 Detection, prevention and treatment of disease ............................. 123
7.3.1 Diabetes mellitus ................................................................ 124
7.3.2 Cardiovascular Disease (CVD) ........................................... 124
7.3.3 Chronic Kidney Disease (CKD) .......................................... 125
7.3.4 Coronavirus disease (COVID-19) ....................................... 126
7.4 Conclusion .................................................................................... 127
References ............................................................................................. 128

7.1 INTRODUCTION

The healthcare sector is presently experiencing the implementation of abun-

dant innovations. Human health being the most important aspect to consider,
researchers have put a lot of efforts in developing various new medicines,
devices and systems to detect, prevent and treat different diseases. Currently,

DOI: 10.1201/9781003407300-7 113

114 Saumya Yadav et al.

there are two main concerns in the medical sector: increased health main-
tenance expenditures and a shortage of medical experts. According to the
World Health Organizations (WHO), in 2013 the worldwide demand and
actual number of health staff were 60.4 million and 43 million, respec-
tively [1]. By 2030, it is estimated that these numbers will rise to a demand
for 81.8 million with an actual number of staff of 67.3 million. The short-
age of medical amenities and medical experts is still a serious problem.
Worldwide, by 2050 there will be a considerable increase in the population
of adults above 60 years of age [2] due to the continuing decrease in the
birth rate, a trend which will persist in the coming years. With the increase
of age, an individual is more likely to get infected with different viruses
and diseases which require long-term care for them to return to health.
Ultimately, the increased human resources, medicines, medical devices, and
expenditure will be required by the population. The constant evolution of
new viruses and diseases makes it challenging to stay ahead of the curve, but
with the help of different approaches to the treatment of information, it is
possible to conquer these challenges.
Across the globe countries devote a considerable proportion of their gross
national product (GNP) is spent on public healthcare. A WHO report exhib-
ited that the related figures in India, Australia, Russia, Canada, the U.S.A.,
and China were 4.7%, 9.4%, 7.1%, 10.5%, and 5.6% respectively [3] and
it is likely that these figures will increase in the coming years. In order to
optimize efficiency, the world’s healthcare systems requires an intelligent
solution which gives accurate results in real time. These intelligent solutions
can be achieved with the help of AI, IoT, and blockchain. Integrating all
these technologies with medical data will undoubtedly help in resolving
healthcare problems. Taken together, these systems can help medical special-
ists to identify the disease by collecting information from the patient using
different sensors, comparing the symptoms of the disease with previous
patients using blockchain and detect the disease through the use of AI tech-
niques. Such techniques are supportive to medical experts in detecting the
disease at the early stages of development, reaching the outcome of complex
disease in an easy and faster way. In addition, these techniques support the
patients at reduced charge, take lesser time and make an accurate analysis
about the probable disease. As a large amount of healthcare data is available
today and new advanced data analytics techniques made it possible to
develop accurate medical diagnostic tools, primarily to help a lot of people
in remote locations which lack medical professionals. AI is merged with the
IoT to monitor the health conditions of the patient using different sensors
and giving the output in real time. It aims to optimize the time and financial
status of a person without compromising the health of a person. Additionally,
it provides support to medical staff and reduces government spending in the
health sector.
Bearing in mind the importance of this important evolution in healthcare
systems, this work presents a detailed study of different technologies
 Intelligent approaches for disease detection and prevention 115

employed in the healthcare sector. The chapter includes various works per-
formed by the researchers in the medical field using machine learning, IoT,
and blockchain technology. The chapter further aims to look at different
researches performed for detecting, preventing, and treating diseases such as
diabetes mellitus (DM), cardiovascular disease, chronic kidney disease
(CKD), and coronavirus using machine learning, the IoT, and blockchain
technology. The structure of the chapter is as follows: section 7.2 discusses
AI, the IoT, and blockchain technology; and section 7.3 presents the research
work done in the healthcare sector to prevent, detect and treat different
diseases using these technologies. Finally, section 7.4 concludes the chapter
and discuss the future possibility of these technologies in healthcare.

7.2 ARTIFICIAL INTELLIGENCE, INTERNET OF THINGS,

AND BLOCKCHAIN TECHNOLOGY

7.2.1 Artificial intelligence
AI is a subsection of computer science that tries to develop intelligent
machine and technologies. It makes efforts to program machines and imple-
ment human intelligence in them to impersonate their actions. The prob-
lem-solving and learning approach is being evolved in machines to make
them more comprehensive. John McCarthy coined the term “artificial intel-
ligence” in 1956 during his workshop. The goal of this was “to proceed on
the basis of the conjecture that every aspect of learning or any other feature
of intelligence can in principle be so precisely described that a machine can
be made to simulate it”. Machine learning (ML) is a subsection of AI devel-
oped by Arthur Samuel in 1959 to develop various algorithms for extraction
of the information from the input data and the exploration of generalised
characteristics in the data points. These characteristics are general math-
ematical models established with certain rules and principles. The models
and decision process is generated from the input data and human guidance.
Once developed, ML automates the process using mathematical models and
programming that eliminate the human expert in the process.
Among the existing technologies, AI is the most promising and powerful
technology in the medical field [4–6]. AI can be described as a simulation of
human intelligence in a machine. AI is implemented when the system starts
making decisions like a human being. It is a large field that includes different
techniques such as computer vision, machine learning, and deep learning,
which are widely used in the medical field for performing segmentation,
classification, detection, prevention and treatment [7]. It aims to develop an
automatic system that learns from the input data and solves complex prob-
lems. AI uses the mathematical tool, i.e., ML, to iteratively learn from the
input data and tries to find the pattern from these data. On getting the
­pattern, it generates a ML model which is utilized by the user to perform
116 Saumya Yadav et al.

different tasks. AI is hungry for data. The bigger data used for training the
model, the more accurate the result it will provide. In order to make use of
AI in the healthcare sector, medical data is collected from clinical activities,
such as diagnosis, screening, scanning, treatment etc. Various modalities,
such as magnetic resonance imaging (MRI), positron emission tomography
(PET), ultrasound, microscopy, X-Ray, and computed tomography (CT), are
widely used to obtain a fast and accurate medical diagnosis. In addition,
they help healthcare experts to make a decision for surgeries and navigating
tools while performing surgeries [8].

7.2.2 Machine learning
Researchers are working in the field of AI with systems with abilities such as
perception, reasoning and learning. This field is evolved further and applied
in different domains and industries for a range of applications, e.g. com-
puter science, self-driving cars, drug development, mathematics, finance,
production, automobile, and many more. ML algorithms can learn from the
data of a similar group of subjects, a connection between subject features
and from the outcome. Computational systems are trained through differ-
ent algorithms and statistical models for the analysis of the sample data, so
as to develop an efficient learning process. Domain knowledge is required
for the selection of proper features in the ML algorithm for improving the
predictions. Its features include the variable or descriptive attributes that are
recorded and quantified from raw data to train the ML model. In a broad
manner, ML is developing an algorithm or computer program which pro-
gressively improve its performance on its own. ML is principally divided into
four types: Supervised Learning, Unsupervised Learning, Semi-supervised
Learning, and Reinforcement Learning (Figure 7.1).

Figure 7.1 Types of ML.

 Intelligent approaches for disease detection and prevention 117

Supervised Learning occurs when the algorithm is trained using the

labelled data set. These data sets have both input and output parameters.
For supervised learning, both training and validation data sets are labelled.
In supervised learning algorithms, the input is a set of N samples, repre-
sented as X in Equation 7.1,

X   xi , xi 1, , xN 1  (7.1)

and the value of i ranges from 0 to N-1 for all used samples. The term xi
represents the feature vector for the ith sample in the input data. The output
class or label is represented by Y corresponding to each input data given in
Equation 7.2,

Y   yi , yi 1, , yN 1  (7.2)

The whole training set is represented by T, which is the set of input data xi
and output class or label yi. T is represented as in Equation 7.3,

T   xi , yi  ,  xi 1, yi 1  , ,  xN 1, yN 1   (7.3)

where i  0, 1, 2, .,  N  1  .

The learning algorithm tries to build a relationship or function f which
maps X to Y, as shown in Equation 7.4.

f :X →Y (7.4)

Steps for solving a problem using a supervised learning are given as in Figure 7.2.

Data analysis: There should be an analysis of the data before training to

know what kind of data is to be given, e.g. there may be a different
possibility of the training set for optical character recognition. It can
be a word or perhaps a single character.

Figure 7.2 Steps for supervised learning.

118 Saumya Yadav et al.

Data collection: The data is to be collected according to the requirement

of the targeted task which should have good representation to the real-
world use. The input data is collected with their corresponding labels.
The labels can be generated with the expert’s opinion or through the
conventional method’s measurement results.
Feature representation: The training model’s performance is significantly
dependent on the selection of the features from the input data. The
number of the features represents the characteristics of the input
object. The number of the features should be optimally selected; it
should not be too large or not very low. It should contain several fea-
tures that could be required for an accurate prediction of the output.
Learning algorithm: The next step is to choose or design the learning
methodology for training with input data. The learning should be
done depending on the mathematical computation which is related to
the performance of the system. An ideal system should be optimised
such that it is accurate enough and in the same way computationally
inexpensive.
Parameter optimization: Once the algorithm is selected to train with a
training set of the input data, the parameters used in the algorithm
should be optimized subsequently to enhance the performance of the
system. The optimization is done concerning the test set’s obtained
performance or validation set, which is a kind of unseen data to the
training algorithm. Cross-validation is also used to exhaustively evalu-
ate the performance of the system.
Performance evaluation: The final step is to evaluate the performance of
the system in a certain set of parameters. The chosen parameter can be
different according to the selected problem. The most common param-
eters are accuracy, specificity, sensitivity, F1-score, kappa, area under
the curve, confusion matrices etc.

Various machine-learning algorithms, such as Support Vector Machines

(SVM), Random Forest, Decision tree, Naïve Bayes and k-Nearest Neighbour
classifiers, are employed to perform different tasks such as population growth
prediction, weather forecasting, speech recognition, digit recognition etc. It
has been attracted in research for solving complex problems in medical [9–14].
Unsupervised Learning allows the ML model to learn without a labelled
data set and with no learning guidance provided. In unsupervised learning,
the role of the algorithm is to cluster unprepared and unsorted data accord-
ing to some similarities and variations with minimum human supervision. It
tries to find the variation in the input data and classifies the output based on
some differences. The most common unsupervised learning method is clus-
tering and principal component analysis. Steps for unsupervised learning are
shown in Figure 7.3.
Principal component analysis and clustering are the two most commonly
used probabilistic methods in the domain of unsupervised learning. Clustering
 Intelligent approaches for disease detection and prevention 119

Figure 7.3 Steps for unsupervised learning.

is used to group the unlabelled, unclassified, uncategorised data. It discov-

ers patterns of interest in the natural feature space of the data, such as
behavioural analysis of a group of customers. A cluster in a feature space
represents the similar data points of the domain which are closer to other
groups of the data points or clusters. Generally, clusters have their centroid
point or sample in that feature space with similar traits. The cluster may
have a centre (the centroid) that is a sample or a point feature space, and
may also have a boundary or extent. Clustering is broadly classified into
two groups, namely soft clustering and hard clustering. In the case of soft
clustering, the data points are assigned with a probability of belongingness
to a certain cluster. But the hard clustering gives the binary results, i.e. the
given data point may or may not be associated with the cluster. The data
point can belong to only one cluster out of all generated clusters in the case
of soft clustering. The other two major clustering techniques with wide
applications are K-means clustering and hierarchical clustering. The
K-means clustering is an iterative method of clustering that works on
the principle to find local maxima in each iteration. The term ‘K’ in K-means
clustering represents a number of the clusters generated with this method.
Hierarchical clustering works on building the hierarchy of the clusters in
which different clusters are generated from the given data points. Then, the
two nearest clusters are merged into the same cluster and the algorithms
end in a single cluster. It is implemented using the bottom-up approach.
The merging of the clusters is done based on different mathematical crite-
ria such as Mahalanobis distance, Euclidean Distance, Manhattan dis-
tance, squared Euclidean distance etc. K-means algorithms can handle a
large amount of data as compared to hierarchical clustering. The time
complexity of hierarchical clustering is quadratic, whereas the time com-
plexity of the K-means algorithm is linear. K-means algorithm works well
with the data in situations where the clusters are spherical or circular in
shape. Other examples of the clustering methods include Density-Based
Spatial Clustering of Applications with Noise (DBSCAN), mean-shift algo-
rithm, agglomerative hierarchical algorithm, affinity propagation, spectral
clustering, etc.
120 Saumya Yadav et al.

The principal component analysis is basically a dimensionality reduction

method and is employed in dealing with large-sized data sets. It transforms
large size data sets into smaller sets such that most of the information is still
preserved. With the aid of the principal component analysis, the number of
variables is reduced so that their analysis becomes easier. Although it comes
at the cost of compromised accuracy it can make the methodology much
simpler and easier to implement. The first step is the standardization of the
data so that each variable has an equal contribution to the analysis. This
step helps to avoid those situations where variables with larger ranges domi-
nate over data with small range variables. The next step is the computation
of the covariance matrix to analyse the variation of the input data from the
mean value. In some situations where two or more variables are highly
related to each other, then those features are redundant to process and used
for dimensionality reduction. Then, in the next step, Eigenvalues and
Eigenvectors of the covariance matrix are to be computed and those features
will be chosen which are depicting high variance among them. Unsupervised
algorithms are widely used to perform several disease detection and local-
ization tasks [15–19].
Semi-supervised learning is a fusion of supervised and unsupervised learn-
ing. The training data for semi-supervised learning employs both labelled
and unlabelled data sets. In semi-supervised learning, models are first trained
with the labelled data set to set some learning rules; then the rules are modi-
fied to train with the unlabelled data set. The robustness of such ML models
is dependent on the consistency of the labelled data set. The better the label-
ling of the data set, the more robust the model. One method of semi-super-
vised learning is to combine the classification and the clustering method to
generate a model. Using the clustering method to assemble the input data
allows us to secure the most relevant samples and label of the data. Clustering
is unsupervised learning, so it does not require any instruction for grouping
the data. Later, you can use the classification method on the selected rele-
vant data to generate the prediction model. The healthcare sector uses semi-
supervised learning to assist the experts using different technologies
[20–26].
Reinforcement Learning is a type of ML whose agents learn from the
consequences of its action rather than learning from external resources. It
selects its action from past experiences and also by exploring new actions. In
this method, an agent tries to achieve a goal in a potentially complex and
uncertain environment. Here, the computer tries to solve problems in a man-
ner similar to that of a game. The machine works on the trial-and-error
phenomenon to solve a problem and generate the output. The AI-based
method gets rewards or penalties according to the performed action. The
main target to enhance the amount of the reward which depicts its good
chances to take the correct decision. In this type of learning, the model learns
on its own and only rewards and penalties are provided. The machine is
performing multiple trials to maximize its performance. The main criticality
 Intelligent approaches for disease detection and prevention 121

in this kind of system is to prepare the simulation environment in which

these tasks will be performed. The system which works in real-world situa-
tions become more challenging, e.g. automatic self-driving vehicles, other
simulators etc. As there is no communication to the network apart from the
reward and penalties system, scaling and controlling the agent is another
criticality. In such a situation, the computer goes into a situation called cata-
strophic forgetting in which the computer’s previous knowledge is erased
from the network while it trains for new knowledge [27].
Positive and negative reinforcement are two techniques based on the type
of reinforcement. If some event occurs due to some behaviour, that will
increase the frequency and strength of the behaviour. Thus, this gives a posi-
tive trigger to the system, meaning that the system can be sustained longer.
Yet these systems can be overloaded with the states due to greater reinforce-
ment which can decrease the performance of the system. It has multiple
options to explore before opting for the correct option. It is different from
supervised learning as it does not have a labelled training data set to per-
form correct training [28–32]. Steps for reinforcement learning are shown in
Figure 7.4.
In the recent years, a subfield of ML, “deep learning”, has provided sub-
stantial growth in accuracy, robustness and provides real-time output using
new learning algorithms. It comprises a deep graph with different layers
performing different tasks. Deep learning layers consist of convolution,
pooling, activation, and fully connected layers. These deep layers easily find
the complex features from the input medical data to generate a robust model
for performing a different task. Various deep models are proposed by the
researchers to perform different tasks on the input medical data [33–36].
Computer vision (CV) is another field of AI. It supports computers in iden-
tifying and interpreting the visual world. It assists the medical experts by
becoming an eye of the medical devices to identify, classify, and detect differ-
ent diseases from the input modalities of a patient [37]. CV automatically
extracts information from the image and with the help of a ML algorithm,
it classifies the image content. CV in the healthcare sector enables the experts

Figure 7.4 Steps for reinforcement learning.

122 Saumya Yadav et al.

to diagnose their patient, perform surgery, monitor the evolution of different

diseases, and detect disease in clinical images, the early identification of dis-
eases and the detected part of the body.

7.2.3 Internet of Things
IoT can be characterized as an interconnection of physical objects or
“things”, consisting of sensors, processing network and software to connect
and exchange the data with other devices and systems over the Internet.
Sensors and other connected devices collect the data from the given environ-
ment and information are extracted from the raw data. Then the informa-
tion is transferred to cloud servers and other devices via the Internet. Its
applications are able to function with the different domains and enable high
analyses and management of complex interactions in the devices [38]. IoT
devices, such as wearable sensors, monitors, and smart pills, help to collect
the data and ML models can use this data for disease prediction. Different
sensors are embedded into the medical system via automated signal process-
ing, automation, and computer technologies. These sensors collect informa-
tion that allows a clinician to recognize the situation of the patient in real
time and to help them by providing treatment. Different medical signals can
be found in the form of blood pressure, body temperature, heartbeat rate,
electromyograms (EMGs), electrocardiograms (ECGs), electroglottographs
(EGGs), electroencephalograms (EEGs), and electrooculograms (EOGs). An
IoT-based healthcare monitoring system can use these signals to help the
clinician to monitor the patient proactively. A smart gateway assembles data
from different smart devices. It can perform processing, compress data, use
noise filtering on the medical devices, and analyse the data to detect and
predict the risky pattern in a patient’s heath. The Internet of Medical Things
(IoMT) is a further tweaking of the system. It is a combination of medical
applications and systems that connect healthcare data using network tech-
nologies. IoMT plays a vital role in developing a smart standalone health-
care system. IoT with the help of AI supports the medical experts to take the
medical data of the patient remotely, leading to effective diagnosis. It trans-
fers the data from one device to another to enhance the performance of the
system in an automatic manner without any human intervention [39–48].

7.2.4 Blockchain technology
Blockchain technology was introduced in 2008 with the introduction of
the Bitcoin cryptocurrency technology. It is a public ledger, tracking assets,
distributed, recording transaction, and assure immutability in between a
p2p network of computers. Blockchain technology is widely used in vari-
ous fields, including healthcare, finance, industries, and business. It is an
innovative data structure that contains the list of records called a block [46].
Every data block consists of time-stamped batches of the latest transaction,
 Intelligent approaches for disease detection and prevention 123

a hash (unique identifier or digital fingerprint), and a hash of the previ-

ous data block. Modifying any block of blockchain technology is therefore
difficult as changing one block leads to changes in all the blocks behind
the modified block. Several features of this technology, such as the immu-
tability of stored data, decentralization, and traceability are attracting the
healthcare sector for merging blockchain with medical devices [49–54]. It
is expected to upgrade the information management of patients, improve
clinical research, insurance claim process, and the management of data
records [55]. Blockchain technology enables institute-driven interoperabil-
ity to transform to patient-centred interoperability. Researchers can access
some part of the patient data for a limited time period after taking consent
from the patients using blockchain technology. It also allows patients to
connect to other hospitals and automatically collect their medical data from
the previous hospitals. In addition, blockchain in healthcare is capable of
protecting the patient’s information [15]. It is constantly upgrading, and it
has several challenges which must be resolved to merge it with healthcare
and medical applications. The foremost challenge concerns confidentiality
and the second challenge scalability and speed.
AI with IoT and blockchain technology is a boon for the healthcare sec-
tor. These technologies are helping to boost up the clinical data to create
efficient and accurate healthcare treatments for patients and increase pro-
ductivity in the workforce. These computational tools also help in optimiz-
ing the parameters by which the root cause of a disease can be identified.
Using IoT sensors, clinicians are collecting, analysing, and reporting the data
promptly and sending it to doctors for deeper insights into the health of the
patients. Doctors are performing real-time monitoring of patients who are
seriously ill. Blockchain is bringing accessibility and transparency in critical
patient’s information from all the relevant doctors. Also, it is easier for the
patient to switch to another hospital using a blockchain network that
includes the patient data. Using AI algorithms, medical experts can easily
detect, prevent and diagnose the disease in real time. The perfect integration
of these three technologies and digital innovations is leading to improve-
ments in the healthcare field. This enhancement brings new options of treat-
ment by medical experts. Technologies also assist the healthcare members to
track the patients in the hospital with the help of CV [37] and maintain their
routine. Figure 7.5 represents the integration of AI, IoT and blockchain with
healthcare.

7.3 DETECTION, PREVENTION AND TREATMENT

OF DISEASE

With the advancement in medical treatment, the patient population is

increasing with more complex and multiple diseases being detected. This is
making medical decisions harder as medication may help one patient and
124 Saumya Yadav et al.

Figure 7.5 AI, IoT and blockchain technology merging with healthcare.

harm another. The latest emerging technologies in healthcare sectors assist

medical experts. The researcher used AI algorithms with IoT-based sensors
and blockchain technology to perform different tasks such as the detection,
prevention and treatment of different diseases.

7.3.1 Diabetes mellitus
Diabetes mellitus (DM) is a complex metabolic disorder that can lead to the
damage of multiple body organs. It is also one of the world’s primary meta-
bolic diseases [56]. There is an emergent need to develop models/tools for
the risk assessment of disease, along with detecting susceptibility and prog-
nosis. The disease is linked to abnormal blood glucose levels in the human
body. Type 1 diabetes is the result of inadequate insulin production and
Type 2 diabetes is an output of oxidative stress, which results from defective
redox reactions and increased reactive metabolites (RMs) [57]. Looking at
the high prevalence of DM in the population, it would seem that most of the
individuals are undiagnosed. Hence, there is an emergent need to develop a
tool to assess the risk assessment of disease, susceptibility and prognosis. In
addition, there is a requirement to prevent diabetes by periodically taking
measurement and keep track of health data. Various risk assessment and
information security models are developed using AI, IoT and blockchain to
detect, prevent and cure DM in the patients as shown in Table 7.1.

7.3.2 Cardiovascular Disease (CVD)

According to a WHO report, an annual number of 12 million people die due
to heart disease at the global level. Heart disease is the reason for the death
of around 31% of the world’s population. WHO assumes that by 2030, the
 Intelligent approaches for disease detection and prevention 125

Table 7.1 Comparative analysis of AI+IoT+blockchain technology for DM

Reference Objective Method Accuracy (%)
[58] Security and privacy Blockchain and IoT NA
[59] Preventing by predicting the disease ML algorithm 83.30
will occur within 5 years
[60] Prediction of diabetes Hadoop cluster 94
[61] Glucose monitoring system for IoT NA
diabetes prevention
[62] Detecting the disease AI algorithm 94.1417
[63] Prediction of DM ML algorithm 81

Table 7.2 Comparative analysis of AI+IoT+blockchain technology for CVD

Reference Objective Method Accuracy (%)
[66] Prediction of heart disease Autoencoder-based 90
artificial neural network
[67] Heart disease prediction Supervised learning 83
algorithms
[68] Cuff-Less Blood Pressure IoT NA
Measurement
[69] Heart disease prediction Cluster-based DT learning 85.90
[70] Heart Rate Monitoring IoT NA
System
[71] Heart disease prediction Hybrid ML Techniques 88.7

death rate due to heart disease will increase up to 23.6 million [64]. The
increased death rate and numerous reasons for heart disease make predic-
tion complicated using a conventional method. Researchers are therefore
focused on developing new methods for diagnosing heart disease detection,
prediction and treatment with the latest technologies. AI predicts heart dis-
ease using the attachment of different IoT sensors. It evaluates patient data
and predicts the chances of heart disease [65]. Input data required for the
heart disease prediction using AI involves time series, images, tabular data,
and text data. The input data to be fed into the AI model are the risk fac-
tor features for predicting the disease. Data include the weight, heart rate,
blood pressure, and physical activities of the respective patients. Some heart
disease prediction using AI, and IoT are given in Table 7.2.

7.3.3 Chronic Kidney Disease (CKD)

CKD is a category of kidney disease in which there is a steady loss of glo-
merular filtration rate (GFR) over a period of more than three months and
the intensifying problem of health [72]. It is a highly complex, secret, and
126 Saumya Yadav et al.

Table 7.3 Comparative analysis of AI+IoT+blockchain technology for kidney disease

Reference Objective Method Accuracy (%)
[74] Detection and diagnosis of IoMT with ML 99.7
CKD
[75] Disease classification ML using multi-kernel 98.5
[76] Diagnosis of CKD ML 99.83
[77] Rule Induction and Hybrid ML Techniques 99.75
Prediction of CKD
[78] Cuff-Less Blood Pressure ML, Internet of Things and 97.8
Measurement Cloud Computing
[79] Predicting the Risk of CKD ML algorithms NA

progressive condition that focuses on the physical functioning of some

organs. According to WHO estimates, the estimated past costs of dialysis
treatment stand at $1.1 trillion [73]. It is difficult to predict the timing of
failure of the kidney because of its complicated nature, the heterogeneity in
different patient’s condition, and the absence of early symptoms. If it is diag-
nosed at an earlier stage, treatment can be started earlier and may reduce
the chances of death. Using ML for disease detection can be more accurate,
robust, and make a prediction in real time when the model is trained with a
clean and suitable data set. The data set for generating a detection or predic-
tion model can be of different modalities. Researchers performed image pro-
cessing to segment out the kidney image, illustrating all defects. Table 7.3
includes CKD detection and prevention using AI and IoT.

7.3.4 Coronavirus disease (COVID-19)

COVID-19 originated in China. It is a disease which is caused by severe
acute respiratory syndrome coronavirus-2 (SARS-CoV-2). It has been found
to be a disease that can be severe in patients suffering from other conditions
and has a current fatality rate of 2% [80]. This life-threatening COVID-19
pandemic is having a very strong impact on the health of the global popu-
lation and the number of patients is increasing for individual testing with
regard to the disease. Conventionally, medical experts have used a chest
x-ray image or CT scan of an individual in order to identify the disease.
This leads to a heavy workload and pressure on radiologists to perform
the test and give accurate results in real time. The fusion of the conven-
tional method with technologies will reduce the workload and also for the
simultaneous performance of multiple tests. An infected patient’s modalities
may include numerous opacities when compared with healthy modalities.
Researchers took advantage of this feature and with the help of AI proposed
various models which can detect COVID-19 from the input data. With the
help of IoT, a clinician can ensure that the virus-infected patient remains
 Intelligent approaches for disease detection and prevention 127

Table 7.4 Comparative analysis of AI+IoT+blockchain technology for COVID-19

Reference Objective Method Accuracy (%)
[81] Detection and diagnosis of DL framework 99.81
CKD
[82] Monitoring the patient IoT applications NA
[83] Social distance measure IoT applications NA
[84] Uniquely tracking donation Blockchain-based network NA
[30] Prediction of COVID-19 Supervised machine 94.99
learning
[85] Diagnostic framework for Deep Learning algorithms 99.81
COVID-19

quarantined for the appropriate time period. The physician can measure
body temperature, oxygen rate, and blood pressure using different IoT sen-
sors. In addition, it helps to monitor the patients’ health. Successfully imple-
menting these technologies can improve the productivity of healthcare staff,
reduce the number of mistakes, cut down on the costs of scans, and reduce
the workload. Several such methods are included in Table 7.4.

7.4 CONCLUSION

People around the globe are affected by different types of deadly diseases.
Among the most common diseases are DM, cardiovascular disease, CKD,
and, most recently, coronavirus disease, all of which are having an adverse
effect on the health and economy of the world. Disease detection is the ini-
tial step taken by the medical experts on finding any symptoms. With the
increasing population, the demand for medical experts is increasing, and so
ultimately is brings the workload and pressure. Merging AI, IoT, and block-
chain with the medical field reduces the experts’ workload and give precise
output in real time. AI and ML techniques are used extensively in clinical
decision-making. With the help of AI, researchers generated different models
for detecting, segmenting, and predicting diseases. Different sensors of IoT
are used to collect data from the patients and also to remotely monitor the
health of the patients. Blockchain technology is used to secure the patients’
information, to permit the researcher to access the limited data, in order to
consulting multiple doctors. The proposed work represents different meth-
ods employed in the health sector using AI, IoT, and blockchain technologies.
It also outlines the methods used for performing different types of work.
In future, the researchers can merge more relevant techniques in health-
care, which will improve disease detection, prevention, and treatment. It will
improve performance, ensure that medical services are financially feasible,
safeguard future health, and reduce the workload of medical experts.
128 Saumya Yadav et al.

REFERENCES

[1] World Health Organization, Health workforce requirements for universal

health coverage and the sustainable development goals. (human resources for
health observer, 17), 2016.
[2] J. R. Beard, et al., “The World report on ageing and health: a policy framework
for healthy ageing.” The Lancet, vol. 387, pp. 2145–2154, 2016.
[3] World Health Organization, “Total expenditure on health as a percentage of
gross domestic product (US $). Global Health Observatory”, 2017.
[4] S. E. Dilsizian, E. L. Siegel, “Artificial intelligence in medicine and cardiac
imaging: harnessing big data and advanced computing to provide personalized
medical diagnosis and treatment.” Current Cardiology Reports, vol. 16(1), pp.
1–8, 2014.
[5] V. L. Patel, N. A. Yoskowitz, J. F. Arocha, “Towards effective evaluation and
reform in medical education: a cognitive and learning sciences perspective.”
Advances in Health Sciences Education, vol. 14(5), pp. 791–812, 2009.
[6] S. Jha, E. J. Topol, “Adapting to artificial intelligence: Radiologists and pathol-
ogists as information specialists.” JAMA, vol. 316(22), pp. 2353–2354, 2016.
[7] G. Litjens, et al., “A survey on deep learning in medical image analysis.”
Medical Image Analysis vol. 42, pp. 60–88, 2017.
[8] Y. Shelke, C. Chakraborty Shelke, “Augmented reality and virtual reality
transforming spinal imaging landscape: a feasibility study.” IEEE Computer
Graphics and Applications, vol. 41(3), pp. 124–138, 2020.
[9] M. Zhang, et al., “Prediction of virus-host infectious association by supervised
learning methods.” BMC Bioinformatics, vol. 18(3), pp. 143–154, 2017.
[10] Y. Cai, X. Tan, Y. Cai, X. Tan, X. Tan, “Selective weakly supervised human detec-
tion under arbitrary poses.” Pattern Recognition, vol. 65, pp. 223–237, 2017.
[11] A. Singh, R. Kumar, A. Singh, R. Kumar, “Heart disease prediction using
machine learning algorithms.” International conference on electrical and elec-
tronics engineering (ICE3). IEEE, 2020.
[12] E. Choi, A. Schuetz, W. F. Stewart, et al., “Using recurrent neural network mod-
els for early detection of heart failure onset.” Journal of the American Medical
Informatics Association, vol. 24(2), pp.361–370, 2017.
[13] D. Ichikawa, T. Saito, W. Ujita, H. Oyama, “How can machine-learning
methods assist in virtual screening for hyperuricemia? A healthcare machine-­
learning approach.” Journal of Biomedical Informatics, vol. 64, pp. 20–24,
2016.
[14] S. P. Patro, N. Padhy, D. Chiranjevi, “Ambient assisted living predictive model
for cardiovascular disease prediction using supervised learning.” Evolutionary
Intelligence vol. (2), pp. 941–969, 2021.
[15] G. Muhammad, “Automatic speech recognition using interlaced derivative pat-
tern for cloud based healthcare system.” Cluster Computing, vol. 18(2), pp.
795–802, 2015.
[16] F. A. Khan, et al., “A continuous change detection mechanism to identify
anomalies in ECG signals for WBAN-based healthcare environments.” IEEE
Access, vol. 5, pp. 13531–13544, 2017.
[17] S. Lim, C. S. Tucker, S. Kumara, “An unsupervised machine learning model
for discovering latent infectious diseases using social media data.” Journal of
Biomedical Informatics, vol. 66, pp. 82–94, 2017.
 Intelligent approaches for disease detection and prevention 129

[18] A. Fong, et al., “Identifying influential individuals on intensive care units: using
cluster analysis to explore culture.” Journal of Nursing Management, vol. 25(5),
pp. 384–391, 2017.
[19] K. Bhatia, R. Syal, “Predictive analysis using hybrid clustering in diabetes diag-
nosis.” Recent Developments in Control, Automation & Power Engineering
(RDCAPE). IEEE, 2017.
[20] S. R. Jonnalagadda, “A semi-supervised learning approach to enhance health
care community–based question answering: A case study in alcoholism.” JMIR
Medical Informatics, vol. 4(3), pp. e5490, 2016.
[21] A. Albalate, W. Minker, Semi-supervised and unsupervised machine learning:
novel strategies. John Wiley & Sons, 2013.
[22] H. Wu, et al., “Automated comprehensive adolescent idiopathic scoliosis
assessment using MVC-Net.” Medical Image Analysis, vol. 48 pp. 1–11, 2018.
[23] Y. Wang, et al., “Semi-supervised classification learning by discrimination-aware
manifold regularization.” Neurocomputing, vol. 147, pp. 299–306, 2015.
[24] L. Nie, et al., “Bridging the vocabulary gap between health seekers and health-
care knowledge.” IEEE Transactions on Knowledge and Data Engineering,
vol. 27(2), pp. 396–409, 2014.
[25] L. Jin, et al., “Integrating human mobility and social media for adolescent psy-
chological stress detection.” International conference on database systems for
advanced applications. Springer, Cham, 2016.
[26] R. A. R. Ashfaq, et al., “Fuzziness based semi-supervised learning approach for
intrusion detection system.” Information Sciences, vol. 378, pp. 484–497, 2017.
[27] C. Kaplanis, M. Shanahan, C. Clopath, “Continual reinforcement learning
with complex synapses.” International Conference on Machine Learning.
PMLR, 2018.
[28] M. F. Zohora, et al., “Forecasting the risk of type ii diabetes using reinforcement
learning.” 2020 Joint 9th International Conference on Informatics, Electronics
& Vision (ICIEV) and 2020 4th International Conference on Imaging, Vision
& Pattern Recognition (icIVPR). IEEE, 2020.
[29] D. Zhang, B. Chen, S. Li, “Sequential conditional reinforcement learning for
simultaneous vertebral body detection and segmentation with modeling the
spine anatomy.” Medical Image Analysis, vol. 67, pp. 101861, 2021.
[30] L. J. Muhammad, E. A. Algehyne, S. S. Usman, “Supervised machine learning
models for prediction of COVID-19 infection using epidemiology dataset.” SN
Computer Science, vol. 2, p. 11, 2021.
[31] J. Wang, et al., “Deep reinforcement active learning for medical image clas-
sification.” International Conference on Medical Image Computing and
Computer-Assisted Intervention. Springer, Cham, 2020.
[32] Z. Liu, et al., “Deep reinforcement learning with its application for lung cancer
detection in medical Internet of Things.” Future Generation Computer Systems,
vol. 97, pp. 1–9, 2019.
[33] Y. Cao, et al., “Improving tuberculosis diagnostics using deep learning and
mobile health technologies among resource-poor and marginalized communi-
ties.” 2016 IEEE first international conference on connected health: applica-
tions, systems and engineering technologies (CHASE). IEEE, 2016.
[34] S. Mishra, A. Dash, L. Jena, “Use of deep learning for disease detection and
diagnosis.” Bio-inspired neurocomputing. Springer, Singapore. pp. 181–201,
2021.
130 Saumya Yadav et al.

[35] J. Venugopalan, et al., “Multimodal deep learning models for early detection of
Alzheimer’s disease stage.” Scientific Reports. vol. 11(1), pp. 1–13, 2021.
[36] J. Wang, et al., “Detecting cardiovascular disease from mammograms with
deep learning.” IEEE Transactions on Medical Imaging, vol. 36(5), pp. 1172–
1181, 2017.
[37] R. C. Joshi, et al., “Object detection, classification and tracking methods for
video surveillance: a review.” 2018 4th International Conference on Computing
Communication and Automation (ICCCA). IEEE, 2018.
[38] B. D. Martino, et al., “Internet of things reference architectures, security and
interoperability: a survey.” Internet of Things, vol. 1, pp. 99–112, 2018.
[39] H. Sattar, et al., “An IoT-based intelligent wound monitoring system.” IEEE
Access, vol. 7, pp. 144500–144515, 2019.
[40] S. Muthukumar, et al., “Smart Humidity Monitoring System for Infectious
Disease Control.” 2019 International Conference on Computer Communication
and Informatics (ICCCI). IEEE, 2019.
[41] F. Miao, et al., “Multi-sensor fusion approach for cuff-less blood pressure mea-
surement.” IEEE Journal of Biomedical and Health Informatics, vol. 24(1),
pp. 79–91, 2019.
[42] F. Yang, et al., “Multi-method fusion of cross-subject emotion recognition based
on high-dimensional EEG features.” Frontiers in Computational Neuroscience,
vol. 13, pp. 53, 2019.
[43] Q. Gu, et al., “Health and safety situation awareness model and emergency
management based on multi-sensor signal fusion.” IEEE Access, vol. 7, pp.
958–968, 2018.
[44] M. Muzammal, et al., “A multi-sensor data fusion enabled ensemble approach
for medical data from body sensor networks.” Information Fusion, vol. 53, pp.
155–164, 2020.
[45] K. Lin, et al., “Multi-sensor fusion for body sensor network in medical human–
robot interaction scenario.” Information Fusion, vol. 57, pp. 15–26., 2020.
[46] A. Shahnaz, U. Qamar, A. Khalid, “Using blockchain for electronic health
records.” IEEE Access, vol. 7, pp. 147782–147795, 2019.
[47] J. Chen, et al., “A novel medical image fusion method based on Rolling
Guidance Filtering.” Internet of Things, vol. 14, pp. 100172, 2020.
[48] D. Fabiano, S. Canavan. “Emotion recognition using fused physiological
signals.” 2019 8th International Conference on Affective Computing and
Intelligent Interaction (ACII). IEEE, 2019.
[49] S. Tanwar, K. Parekh, R. Evans. “Blockchain-based electronic healthcare record
system for healthcare 4.0 applications.” Journal of Information Security and
Applications, vol. 50, pp. 102407, 2020.
[50] K. Shuaib, et al., “Blockchains for secure digitized medicine.” Journal of
Personalized Medicine, vol. 9(3), pp. 35, 2019.
[51] T. Zhou, X. Li, H. Zhao, “Med-PPPHIS: blockchain-based personal healthcare
information system for national physique monitoring and scientific exercise
guiding.” Journal of Medical Systems, vol. 43(9), pp. 1–23, 2019.
[52] H. Kaur, et al., “A proposed solution and future direction for blockchain-
based heterogeneous medicare data in cloud environment.” Journal of Medical
Systems, vol. 42(8), pp. 1–11, 2018.
[53] E. A. Breeden, C. Davidson, T. K. Mackey. “Leveraging blockchain technol-
ogy to enhance supply chain management in healthcare: an exploration of
 Intelligent approaches for disease detection and prevention 131

challenges and opportunities in the health supply chain.” Blockchain Healthc

Today, 2018.
[54] V. Patel, “A framework for secure and decentralized sharing of medical imag-
ing data via blockchain consensus.” Health Informatics Journal, vol. 25(4), pp.
1398–1411, 2019.
[55] T. T. Kuo, H. E. Kim, L. O. Machado, “Blockchain distributed ledger tech-
nologies for biomedical and health care applications.” Journal of the American
Medical Informatics Association, vol. 24(6), pp. 1211–1220, 2017.
[56] S. Kang et al., “An efficient and effective ensemble of support vector machines
for anti-diabetic drug failure prediction.” Expert Systems with Applications,
vol. 42(9), pp. 4265–4273, 2015.
[57] M. Bhattacharyya, et al., “Oxidative stress-related genes in type 2 diabetes:
association analysis and their clinical impact.” Biochem Genet, vol. 53(4–6),
pp. 93–119, 2015.
[58] K. Azbeg, et al., “Blockchain and IoT for security and privacy: A platform for
diabetes self-management.” 2018 4th international conference on cloud com-
puting technologies and applications (Cloudtech). IEEE, 2018.
[59] N. Singh, P. Singh, “Stacking-based multi-objective evolutionary ensemble
framework for prediction of diabetes mellitus.” Biocybernetics and Biomedical
Engineering, vol. 40, pp. 1–22, 2020.
[60] N. Yuvaraj, K. R. SriPreethaa, “Diabetes prediction in healthcare systems using
machine learning algorithms on Hadoop cluster.” Cluster Computing, vol. 22(1),
pp. 1–9, 2019.
[61] F. Valenzuela, et al., “An IoT-based glucose monitoring algorithm to prevent
diabetes complications.” Applied Sciences, vol. 10(3), pp. 921, 2020.
[62] M. Shuja, S. Mittal, M. Zaman. “Effective prediction of type ii diabetes mel-
litus using data mining classifiers and SMOTE.” Advances in computing and
intelligent systems. Springer, Singapore, pp. 195–211, 2020.
[63] H. M. Deberneh, I. Kim, “Prediction of Type 2 diabetes based on machine
learning algorithm.” International Journal of Environmental Research and
Public Health, vol. 18(6), pp. 3317, 2021.
[64] C. Dangare, S. Apte, “A data mining approach for prediction of heart disease
using neural networks.” International Journal of Computer Engineering and
Technology (IJCET), vol. 3(3), pp. 30–40, 2012.
[65] Y. Yan, et al., “The primary use of artificial intelligence in cardiovascular dis-
eases: what kind of potential role does artificial intelligence play in future med-
icine?” Journal of Geriatric Cardiology: JGC, vol. 16(8), p. 585, 2019.
[66] I. D. Mienye, Y. Sun, Z. Wang, “Improved sparse autoencoder based artifi-
cial neural network approach for prediction of heart disease.” Informatics in
Medicine Unlocked, vol. 18, p. 100307, 2020.
[67] M. Chakarverti, S. Yadav, R. Rajan, “Classification technique for heart disease
prediction in data mining.” 2019 2nd International Conference on Intelligent
Computing, Instrumentation and Control Technologies (ICICICT).
[68] M. Simjanoska, et al., “Multi-level information fusion for learning a blood
pressure predictive model using sensor data.” Information Fusion, vol. 58, pp.
24–39, 2020.
[69] G. Magesh, P. Swarnalatha. “Optimal feature selection through a cluster-based
DT learning (CDTL) in heart disease prediction.” Evolutionary Intelligence,
vol. 14(2), pp. 583–593, 2021.
132 Saumya Yadav et al.

[70] J. N. Kalshetty, P.M. Varghese, K. Karthik, R. Raj, N. Yadav, “IoT-Based Heart

Rate Monitoring System”, In Advances in Artificial Intelligence and Data
Engineering: Select Proceedings of AIDE 2019, pp. 1465–1475, Singapore:
Springer Singapore, 2020.
[71] S. Mohan, C. Thirumalai, G. Srivastava. “Effective heart disease prediction using
hybrid machine learning techniques.” IEEE Access, vol. 7, pp. 81542–81554, 2019.
[72] P. E. Stevens, A. Levin, Kidney Disease: Improving Global Outcomes Chronic
Kidney Disease Guideline Development Work Group Members*, “Evaluation
and management of chronic kidney disease: synopsis of the kidney disease:
improving global outcomes 2012 clinical practice guideline.” Annals of
Internal Medicine, vol. 158(11), pp. 825–830, 2013.
[73] M. J. Lysaght, “Maintenance dialysis population dynamics: current trends
and long-term implications.” Journal of the American Society of Nephrology,
pp. S37–S40, 2002.
[74] F. Ma, et al., “Detection and diagnosis of chronic kidney disease using deep
learning-based heterogeneous modified artificial neural network.” Future
Generation Computer Systems, vol. 111, pp. 17–26, 2020.
[75] H. Fang, et al., “An improved arithmetic optimization algorithm and its appli-
cation to determine the parameters of support vector machine.” Mathematics,
vol. 10(16), pp. 2875, 2022.
[76] J. Qin, et al., “A machine learning methodology for diagnosing chronic kidney
disease.” IEEE Access, vol. 8, pp. 20991–21002, 2019.
[77] S. H. Ripon, “Rule induction and prediction of chronic kidney disease using
boosting classifiers, Ant-Miner and J48 Decision Tree.” 2019 international
conference on electrical, computer and communication engineering (ECCE).
IEEE, 2019.
[78] A. Abdelaziz, et al., “A machine learning model for predicting of chronic kid-
ney disease based internet of things and cloud computing in smart cities.”
Security in smart cities: models, applications, and challenges. Springer, Cham,
pp. 93–114, 2019.
[79] W. Wang, G. Chakraborty, B. Chakraborty, “Predicting the risk of chronic kidney
disease (CKD) using machine learning algorithm.” Applied Sciences, vol. 11(1),
p. 202, 2021.
[80] F. Wu, et al., “A new coronavirus associated with human respiratory disease in
China.” Nature, vol. 579(7798), pp. 265–269, 2020.
[81] A. I. Khan, J. L. Shah, M. M. Bhat, “CoroNet: a deep neural network for detec-
tion and diagnosis of COVID-19 from chest x-ray images.” Computer Methods
and Programs in Biomedicine, vol. 196, p. 105581, 2020.
[82] R. P. Singh, et al., “Internet of things (IoT) applications to fight against COVID-
19 pandemic.” Diabetes & Metabolic Syndrome: Clinical Research & Reviews,
vol. 14(4), pp. 521–524, 2020.
[83] M. Gupta, M. Abdelsalam, S. Mittal, “Enabling and enforcing social distancing
measures using smart city and its infrastructures: a COVID-19 Use case.” arXiv
preprint arXiv, vol. 2004(09246). 2020.
[84] N. Kamalakshi, “Role of blockchain in tackling and boosting the supply chain
management economy post COVID-19.” Convergence of internet of things and
blockchain technologies. Springer, Cham, pp. 193–205, 2022.
[85] S. T. Ahmed, S. M. Kadhem, “Using machine learning via deep learning algorithms
to diagnose the lung disease based on chest imaging: a survey.” International
Journal of Interactive Mobile Technologies, vol. 15(16), pp. 95–112, 2021.
Chapter 8

Interoperability in IoT-driven
smart buildings
Employing Rule-based decision support
systems
Mohan Krishna S.
Alliance University, Bangalore, India

Thinagaran Perumal
University of Putra, Selangor, Malaysia

Sumukh Surya
Bosch Global Software Technologies, Bangalore, India

Chandrashekar
L & T Technological Services, Bangalore, India

CONTENTS

8.1 Introduction: background and driving forces ................................ 133

8.2 The necessity of decision support inbuilding energy
subsystems .................................................................................... 135
8.3 Decision support model for BEMS ................................................ 136
8.4 Opportunities and challenges for BEMS interoperability .............. 140
8.5 Conclusion and future scope ......................................................... 141
References ............................................................................................. 142

8.1 INTRODUCTION: BACKGROUND AND DRIVING

FORCES

Sustainable energy development and climate change are closely linked, and
it is increasingly felt across the world that energy consumption must be opti-
mized, and that energy efficiency must be increased. This is applicable to all
sectors, but there is greater thrust on the demand side. Commercial buildings

DOI: 10.1201/9781003407300-8 133

134 Mohan Krishna S. et al.

Figure 8.1 Components of smart (intelligent) buildings.

are potentially large energy guzzlers. The power of technology and the inter-
net was harnessed to ensure the building sector optimizes energy use. The
rapid spread of the internet and automation across the world, combined
with increasingly affordable and high-performance sensors, has led to the
concept of intelligent buildings [1–3]. The quest for making commercial
buildings and homes intelligent has opened a whole new world of possi-
bilities and challenges with respect to the technology, economy, and envi-
ronment. The different components of a smart or intelligent building are
succinctly represented in Figure 8.1. From the technological point of view,
the presence of a multitude of heterogeneous subsystems in a building, along
with associated sensor and communication technologies, have also thrown
up issues related to coordination and interoperability. From the economic
point of view, this requires heavy capital investment, however, the return
on investment (ROI) is encouraging with payback periods of less than five
years. The environmental degradation is contained as the carbon footprint
of the buildings is reduced drastically (with some of them even achieving net
zero emissions).
Therefore, it can be safely interpreted that the concept of intelligent build-
ings is here to stay and, if implemented and executed properly, it will have a
cascading effect on the energy patterns of the population and ensure the
achievement of the United Nations Sustainable Development Goals (Goal 7)
related to energy security and sustainable energy development. This chapter
throws light on the technological aspects of intelligent buildings with an
emphasis on coordination.
 Interoperability in IoT-driven smart buildings 135

8.2 THE NECESSITY OF DECISION SUPPORT

INBUILDING ENERGY SUBSYSTEMS

A commercial intelligent building is made up of several heterogeneous sub-

systems, namely:

1. Heating, ventilation and air conditioning (HVAC) and energy man-

agement subsystems (comprising controllers and sensors)
2. Audio and video subsystems (consoles and speakers)
3. Access control subsystems (door access)
4. Fire management subsystems (alarms)
5. Digital surveillance subsystems (Internet protocol camera, closed cir-
cuit television)
6. Water subsystems (valves and pipelines)
7. Other external subsystems and gateways

A designer for intelligent buildings must be able to integrate all these sub-
systems in a seamless way such that they can easily talk to (coordinate with)
each other [4–6].
This is where the issue of interoperability of all these subsystems emerges.
The proliferation of internet and automation, low cost of sensors and
embedded system technologies and the availability of highly versatile com-
munication infrastructure have revolutionized subsystems inside buildings
and made them intelligent. This resulted in increased data intensity of the
buildings as huge amounts of data from these subsystems need to be pro-
cessed, monitored, and analyzed. Figure 8.2 summarizes the different data

Figure 8.2 Data types for heterogeneous subsystems in a building.

136 Mohan Krishna S. et al.

Table 8.1 Pitfalls of Heterogeneous Building Subsystems

Pitfalls Issues
Integration There are different resources, platforms and protocols which
complexity contributed to the diversity in the BEMS.
Gateways Although gateways (at the building server infrastructure) were
employed for integrating different subsystems, their inherent
disadvantage was limited information exchange and not providing
full functionality. Besides this, they also created issues related to
trouble shooting and support.
Coordination Coordination problems arise and there is a necessity for the
problems various subsystems to perform interoperationally to realize
a particular building operation. The data exchange must be
seamless and timely.

types employed for some heterogeneous subsystems in a building. The inte-

gration complexity also increased and posed several challenges for the man-
agement of subsystems in an intelligent building. Table 8.1 provides an
overview of the pitfalls involved.
To address these challenges, there is a necessity to have a mechanism
which monitors and manages all the subsystems in the building. The BEMS
automatically controls the different subsystems by means of real-time data
monitoring. The DSS supplements the BEMS and is instrumental in enabling
building owners to take decisions related to energy usage and costs. A DSS
enables an intelligent building to perceive, learn and adapt. Keeping this in
mind, considerable research focused on the development of DSS along with
building management.

8.3 DECISION SUPPORT MODEL FOR BEMS

A DSS is a particular class of computerized information systems which is

employed for the support of business and decision-making activities of an
organization. It is instrumental in the collection, organization and analysis
of data and facilitates optimal decision-making and aids the management/
facility in operations and planning. DSS is employed in many capital inten-
sive and energy intensive applications like:

Energy management in industries (all sectors)

Financial decision-making in banking and engineering projects
Regulation and Policy making (all sectors)

The basic components of a DSS for BEMS are illustrated in Figure 8.3.
The data can be in the form of raw data, manuals and documents, knowl-
edge inputs from employees and other stakeholders and business models.
 Interoperability in IoT-driven smart buildings 137

Figure 8.3 Basic components of a DSS for BEMS.

For a building, the data can pertain to different subsystems of the building,
the sensor and communication data, building inventory data, data sheets,
energy bills, energy consumption data from the star labels of equipment’s
and buildings and the like. Table 8.2 details the different types of data types
employed for heterogeneous subsystems. Before analyzing the impact of a
DSS on different building subsystems and the different rule bases which are
employed, there are certain ambiguities associated with a DSS used in BEMS
which need to be clarified. Table 8.3 summarizes the characteristics of a DSS
and addresses certain anomalies pertaining to the use of them.

Table 8.2 Data types employed for heterogeneous subsystems

Building subsystems Different datatypes employed
Audio and video subsystems Sensor data, video and audio channels, audio and
video flags, control flags for panning, tilting, and
zooming etc.
Heating, ventilation, and air- Temperature and thermal management data,
conditioning subsystems energy consumption data etc.
Energy management Energy analytics data, energy storage data, energy
subsystems use data etc.
Surveillance subsystems Occupational sensor data, time stamp data etc.
Fire management subsystems Hazard sensor data, equipment data etc.
Lighting subsystems Occupational sensor data, photo sensor data,
luminous efficacy data etc.
138 Mohan Krishna S. et al.

Table 8.3 Characteristics of a DSS for BEMS

Characteristics and anomalies associated
Facilitates The DSS must only support and not automate the
decision-making decision-making process
Interaction and ancillary Supports interaction between user and the different
support layers of IoT-based BEMS. Also supports repeated
usage or for ad hoc decision-making
Task-oriented Have capabilities which provide support for a single
or multiple task which may include data intelligence
and analysis, identification, design and selection of
alternatives and implementation
Identifiable with It can either be an independent system or a subset of a
information systems large integrated information system
Impact on Designed to incorporate more accuracy, timeliness,
decision-making quality, and effectiveness in decision-making

The main purpose therefore is to provide relevant information by the

appropriate classification and categorization of data. The management and
monitoring of subsystems in a smart building has always been a complex
feature due to the heterogeneity and lack of interoperability. A DSS is the
need of the hour to manage everyday energy use and ensure optimal quality
of life. Rule-based mechanisms are also employed to foresee the logic of the
energy management model. Web services and other associated middleware
components also significantly integrate the different heterogeneous subsys-
tems (from different manufacturers and operating platforms). There is a
distinct difference here in the role played by the web services and the DSS.
The web services could take care of integration requirements but for ensur-
ing complete interoperability, a decision model is necessary. In essence, both
middleware as well as DSS is required for intelligent buildings. Additionally,
a DSS can be of different types which are detailed in Table 8.4.
Intelligent building subsystems produce events which correspond exclu-
sively to their functions. Each event triggered must be conveyed by means
of a message which is encoded in a common format in the network. Also,
there is a necessity to ensure the best structure for storing the data in
databases. This is where a standard message format is the need of the hour

Table 8.4 Types of DSS

Data-based Model-based Knowledge-based Document-based
Data analytics Optimization Rules and Managing
and information models, empirical regulations, webpages and
systems, storage models, statistical facts and unfiltered
systems and data models procedures etc. documents
bases
 Interoperability in IoT-driven smart buildings 139

for event sensing, perceiving, and triggering action for interoperability.

SOAP (Simple Object Access Protocol), in recent times, has emerged as a
popular messaging mechanism between building subsystems. SOAP-based
Industrial Messaging Specification (SIMS) entails services where the energy
management system and power generation data can be exchanged in real-
time. It is highly flexible and can be easily implemented. It has also been
used in SCADA (Supervisory control and data acquisition) for tele control-
ling a very high-capacity power plant comprising of hundreds of hydraulic
and tele-controlling units. SOAP makes available an open vendor indepen-
dent standard for end-to-end communication and high flexibility. This
should also be supported by a database module architecture shown in
Figure 8.4, for management of the queries of SOAP messages received from
the building heterogeneous systems. A SQL-based building management
server acts as a mediator for exchange of messages between different sub-
systems. The module is also responsible for data storage from all the build-
ing subsystems along with their related information like status, response,
and action codes. Rules also form an integral part of middleware in an
intelligent building. Rules enforce subsystems to coordinate and work
together. Event–Condition–Action (ECA)-based interoperability frame-
work has been popularly used for intelligent building energy management
along with the ability to provide decision support for heterogeneous build-
ing subsystems [7–8].

Figure 8.4 Basic architecture of database module.

140 Mohan Krishna S. et al.

It is based on Event–Condition–Action (ECA) rules and in the format,

“on” some event

“if” some condition
“then” some actions

The below example elucidates a simple ECA-based rule for interoperation

of building subsystems:

On <FA_alarm_triggered>
If <status_zone001_enable>
Do <trigger_PA001>

The ECA rule-based approach revolutionizes the control and management

of heterogeneous systems in an intelligent building. Their utility is multifold
for the following reasons:

1. Subsystems are always event-driven and therefore need to be reactive

2. Subsystem reaction to an event is also subject to some conditions
3. The application logic must be isolated from the execution of procedures

The ECA interoperation schema defines the rule by employing three-tiered

tables, namely the Status, Response and Action tables. The Status table is
responsible for ensuring data on subsystems status, internal and external
events. The Response table ensures data on subsystems variable and reference
codes. Finally, the Action table takes care of Zone ID, action codes, filename,
and messages. The subsystems will have their own dedicated action tables con-
figured in the middleware architecture. Also, there should be an element of
flexibility in the rules to ensure the facility owners can modify the same [9–10].

8.4 OPPORTUNITIES AND CHALLENGES FOR BEMS

INTEROPERABILITY

Interoperable BEMS enhances the building performance while simultane-

ously reducing the costs. The concept of interoperability in BEMS is pres-
ently at the application level after having matured at the development stage.
One inherent disadvantage is the cost, which is comparatively more than
the existing legacy systems (which are non-interoperable). The installation
and commissioning of new interoperable BEMS or retrofitting the exist-
ing legacy systems is cumbersome, but the benefits incurred are worthy
enough and compensates for any shortcomings. Table 8.5 portrays the ben-
efits accrued from an interoperable BEMS. However, interoperability is also
a double-edged sword. There are inherent disadvantages and limitations
 Interoperability in IoT-driven smart buildings 141

Table 8.5 Benefits of interoperable BEMS

Impact of building performance
Maintenance Building managers will have more ways to operate and manage
facilities. Operations could be consolidated.
Flexibility and There is flexibility for the manager to approach any vendor
operating costs who supports the standards for the features to be
implemented. Savings due to vendor independence
Scheduling A predictive/preventive maintenance program reduces
maintenance costs and equipment.
Data access Huge amounts of information are provided to the managers by
the BEMS’ interoperable systems, with their seamless access
to information across all functions.
Deregulation Interoperable systems enable managers to control the
electrical energy use in real time

present which need to be investigated and rectified for more scalability. The
following points need to be addressed:

1. Innumerable internet-enabled devices (create issues for service provid-

ers to manage the fault, performance, and security of the devices)
2. Since there is no common standard for IoT security, a high level of
privacy and security for users cannot be guaranteed
3. Massive amounts of data need to be stored and processed, leading to
higher power consumption.
4. Cloud services, although may serve the purpose of data saving and
processing, suffer from standardization and synchronization issues
5. Power requirements for sensing, computing and communication must
be optimized as more and more smart devices are integrated
6. Cost and size of the smart devices (with sensing, computing, and com-
munication blocks) must also be optimized.

8.5 CONCLUSION AND FUTURE SCOPE

The chapter presented an overview of the coordination issues faced in

BEMS subsystems owing to their heterogeneous nature and how interop-
erability tries to address the same. It also tries to emphasize the need for
a decision support model in the measurement and management of energy
usage in buildings. The overall framework of DSS for buildings was also elu-
cidated. Also, the popularly used event driven mechanisms like the Event–
Condition–Action (ECA)-based interoperability framework as well as the
middleware employed for the same in smart and intelligent buildings is also
dealt upon. Since it is an area of continuing research, prominent questions
142 Mohan Krishna S. et al.

also are raised which would enable researchers in the domain to pander to
and find solutions for the same.

REFERENCES

[1] Khajenasiri, I., Estebsari, A., Verhelst, M., Gielen, G. A review on internet of
things solutions for intelligent energy control in buildings for smart city appli-
cations. Energy Procedia. 2017, 1(111): 770–779.
[2] Perumal, T. Making buildings smarter and energy-efficient—using the inter-
net of things platform. IEEE Consumer Electronics Magazine. 2021, 10(3):
34–41.
[3] Almusaylim, Z.A., Zaman, N. A review on smart home present state and chal-
lenges: linked to context-awareness internet of things (IoT). Wireless Networks.
2019, 25(6): 3193–3204.
[4] Mohamed, R., Perumal, T., Sulaiman, M.N., Mustapha, N. Multi resident
complex activity recognition in smart home: a literature review. International
Journal of Smart Home. 2017, 11(6): 21–32.
[5] Power, D.J., Power Enterprises. (n.d.). Ask Dan! about DSS, from http://dss
resources.com/faq/index.php?action=artikel&id=13
[6] Le, D.N., Le Tuan, L., Tuan, M.N. Smart-building management system: an
Internet-of-Things (IoT) application business model in Vietnam. Technological
Forecasting and Social Change. 2019, 1(141): 22–35.
[7] Lillo, P., Mainetti, L., Mighali, V., Patrono, L., Rametta, P. An ECA-based
semantic architecture for IoT building automation systems. Journal of
Communications Software and Systems. 2016, 12(1): 24–33.
[8] Park, J.H., Salim, M.M., Jo, J.H., Sicato, J.C., Rathore, S., Park, J.H. CIoT-Net:
a scalable cognitive IoT based smart city network architecture. Human-centric
Computing and Information Sciences. 2019, 9(1): 1–20.
[9] 9 Ways to Create an Intelligent Office. (n.d.). Retrieved January 28, 2021, from
https://www.opensourcedworkplace.com/news/9-ways-to-create-an-intelligent-
office
[10] Interoperability: 5 Ways It Pays. (2002, August 1). Retrieved January 28, 2021, from
https://www.facilitiesnet.com/buildingautomation/article/Interoperability-5-
Ways-It-Pays--1710
Chapter 9

IoT-based parking system

using Web-App
Samyak Jain, Ankit Gupta and Abhinav Sharma
University of Petroleum & Energy Studies, Dehradun, India

Vinay Chowdary
uGDX Institute of Technology, Hyderabad, India

CONTENTS

9.1 Introduction .................................................................................. 143

9.2 Material and methods ................................................................... 144
9.3 Result and discussion .................................................................... 148
9.4 Conclusion .................................................................................... 154
References ............................................................................................. 154

9.1 INTRODUCTION

In today’s world, there are around one billion automobiles, owning a car is
more of a need than a luxury, which highlights the importance of parking
places. It is critical to have an adequate number of parking places in cities,
malls, offices, schools, hospitals, and other locations in order to accommo-
date tourists and inhabitants while also avoiding traffic congestion. There is
a high density of vehicles, specially in metropolitan areas and in hill stations,
that have typically irritated drivers while they waste their time and effort
looking for a parking spot and who typically end up parking their vehicles
on the street. Further, the rapid growth in the number of vehicles on the road
is exacerbating the problem of parking space scarcity. According to industry
estimates, around 30% of traffic is caused by tha motorists’ failure to find a
parking spot. As a result, an effective and smart parking system will save a
lot of time, energy, and fuel.
In recent years, the parking management problem has been studied from
a variety of different perspectives. With the incorporation of advanced tech-
nologies such as the Internet of Things (IoT) [1], artificial intelligence and
machine learning [2], smart parking systems have been implemented in
developed nations such as the United States, Japan, and so on. IoT [3] is
defined as a collection of smart objects (sensors) which can interact with
each other through a communication network. In [4] the authors developed
a low-cost IoT-based parking system for smart cities. In [5, 6] the authors

DOI: 10.1201/9781003407300-9 143

144 Samyak Jain et al.

developed a cloud-based smart parking solution for large parking lots. In [7]
the authors utilized the IoT and a genetic algorithm to develop navigation
and reservation-based parking system. In [8] authors placed TTGO-ESP32-
LoRa boards with ultrasonic sensors to identify free parking slots and then
shared this information through the WiFi module present at the end user. In
[9] the authors proposed the introduction of a graph-based smart parking
system using IoT with its analysis being conducted through a VDM-SL
toolbox.
In this contribution the authors have developed an IoT-based smart park-
ing system using low-cost Light Dependent Resistor (LDR) sensors. Today
most smart parking systems are based on ultra-sonic sensors or IR sensors
[10], but they are costlier and therefore LDR sensors have been used in this
work as it is cost-effective and can be deployed easily. The LDR sensor
works by the shadow detection method where, with the presence of light or
luminous source such as parking lights, luminous intensity is being calcu-
lated and with the absence of luminous intensity shadow is being created.
Therefore, in this work the LDR sensor has been deployed beneath a vehicle
to identify its presence and to indicate the unavailability of a parking slot.
XAMPP is used as our local server in the proposed work to test clients or
websites before transferring them to a remote web server. The chapter is
organized as follows; section 9.2 presents the hardware used and the meth-
odology of this research work; section 9.3 discusses the results; and section
9.4 summarizes the chapter.

9.2 MATERIAL AND METHODS

In this section all of the hardware used to build the IoT-based smart parking
system, architecture, step-wise procedure and algorithm is discussed briefly:

a. Hardware components utilized to build the system

i. LDR Sensor
The LDR, also referred to as the photo resistor, photocell is a
light-dependent resistor whose resistance changes with the inten-
sity of light impinging on its surface, as shown in Figure 9.1. This
resistance finds application in electronic circuits where the detec-
tion of light is required. These resistors have range of shapes and
sizes with the resistance ranging from kilo-ohm to mega-ohm.
ii. Servo Motor
A servo motor is a close-packed component that comprises of a
DC motor, an integrated circuit (IC), a gear train and a potentiom-
eter, with a bearing for the output shaft, as shown in Figure 9.2.
Servos are commonly used in radio-controlled models, such as
vehicles, robots, puppets, etc. The servo motor uses a pulse width
modulation (PWM) signal to control the DC motor; unlike the
 IoT-based parking system using Web-App 145

Figure 9.1 LDR sensor.

Figure 9.2 Servo motor.

typical PWM used in conventional DC motors, this PWM signal

is utilized to regulate the motor direction or position rather than
the rotation speed. Most servo motors function effectively with
a PWM frequency of 50Hz, which change the angular rotation
from 0 to 180 degrees.
iii. PIR
PIRs is a round metal container built from a pyroelectric sensor
that consists of a rectangular crystal in the middle which detects
infrared radiations, as shown in Figure 9.3.
iv. NodeMCU ESP8266
The NodeMCU ESP8266 is a Lua-based open-source firm-
ware development board, as shown in Figure 9.4. It comprises
146 Samyak Jain et al.

Figure 9.3 PIR sensor.

Figure 9.4 NodeMCU ESP8266.

firmware that operates on Espressif Systems’ ESP8266 Wi-Fi SoC,

as well as hardware based on the ESP-12 module which contains
an ESP8266 chip with a Tensilica Xtensa 32-bit LX106 RISC
microprocessor.
v. Arduino UNO
The Arduino/Genuino Uno microcontroller board is based on the
ATmega328P microcontroller, as shown in Figure 9.5. It com-
prises of fourteen digital input/output pins, six analogue inputs, a
16 MHz quartz crystal oscillator, a USB connection, a power jack,
an ICSP header, and a reset button on the board.
vi. Ultrasonic (HR-SR04) Sensor
The ultrasonic sensor, as shown in Figure 9.6, produces a sig-
nal with a high frequency. These signals travel with the speed of
sound and bounce back to the module if it encounters an obstruc-
tion within its path. The sensor has a multi vibrator on its base,
which combines a resonator and a vibrator. An ultrasonic wave is
produced by the vibrator and transmitted through the resonator.
b. Architecture
The architecture of the proposed IoT-based parking system is shown
in Figure 9.7. LDR, passive infrared (PIR) and ultrasonic sensors are
 IoT-based parking system using Web-App 147

Figure 9.5 Arduino UNO.

Figure 9.6 Ultrasonic sensor.

Figure 9.7 Architecture of proposed system for IOT-based smart parking system.
148 Samyak Jain et al.

used in this research work. These sensors detect approaching vehicles

and find the availability of the parking slots. PIR sensors are used to
identify the presence of a car, while an ultrasonic sensor is used at the
exit gate to tally how many cars have left the parking lot and LDR is
installed on the floor of each parking slot. When a vehicle is parked, it
blocks the light. LDR sensor detects light and signals it to Node MCU.
Thereafter, based on the received data, NodeMCU communicates with
the web server by sending desired sensor values. The web server uses
PHP scripts to store and retrieve data from the database server. The
web server is accessed using any web app which enables the user to
find the free parking slots.
c. Implementation of the proposed IoT-based parking system

Step 1: Setting up the webapp

Devolve webapp using open-source tools XAMPP to display
all the calibrations and calculations as desired outputs on the
web application.
Step 2: Setting up the SQL Tables
This module will use XAMPP to establish a local SQL server
and tables with the needed variables and data types to give us
the desired output from the LDR sensor. The real-time sensor
values will be stored and viewed using this table.
Step 3: Setting up the hardware
This module will integrate LDR, ultrasonic sensor, servo
motor, PIR sensor with NodeMCU. Table 9.1 shows the spec-
ifications of the components used in the proposed system.
Step 4: Joining Circuit with Server
This module will calibrate the sensor values and will provide
the Wi-Fi connection to NodeMCU, as well as to the webapp
running on localhost.
Step 5: Testing
The circuit is tested on the webapp and the required values are
stored in the constructed tables by the local server, which will
then be graphically displayed on the webapp. Based on this, the
empty parking spaces are identified and provided to the user.

The flow chart of the admin system and the user system is shown in Figures
9.8 and 9.9.

9.3 RESULT AND DISCUSSION

The proposed parking system is simple and cost-effective and will prove
helpful in developing countries like India. It is simple to deploy and oper-
ate, with lower maintenance and operational costs. Therefore, with little
 IoT-based parking system using Web-App 149

Table 9.1 Specifications of supportive components and use in the proposed system
Component Specification Role

Node MCU Sixteen digital I/O pins Act as a MCU server,

(ESP8266) One analog input pin Providing hardware connectivity and
4MB flash memory data transmission to a web server.
64 KB SRAM
Small sized Module
In-built Wi-Fi/Bluetooth
module chip
LDR Sensor Range – 100 ohm to 10M This sensor will assist the user in
(Light ohm determining which slots are available
Dependent Diameter – 3–20 mm and which are reserved.
Sensor) Max Power – 200 W
HC-SR04 Power Supply: +5V DC The ultrasonic sensor helps us
Ultrasonic Working current: 15mA determine whether or not a vehicle
Sensor Effectual Angle: <15° is approaching parking, and if it is,
Ranging Distance: 2400 cm the LDR value is refreshed and
Resolution: 0.3 cm displayed on the web.
Servo Motor Voltage: 4.8V–6V The ultrasonic sensor will
Speed: 0.14sec/60 degrees communicate a value to the
Connector type: JR type NodeMCU, which will then
trigger the servo motor to open
the gate.
PIR Sensor Input Voltage: DC 4.5V~20V This sensor will assist in detecting
Static Current: <50uA the motion of an approaching car,
Output Voltage: 0V/3V and the servo motor will react
Sensing Range: 7 Meter in accordance with the sensor's
(120 degree once) response.
Arduino Operating Voltage: 5 V Acts as an MCU that interfaces with
UNO Input Voltage: 6V–20 V servo motors, PIR sensors, servo
Digital I/O pins: 14 motor, and ultrasonic sensors, as
Analog input pins: 8 well as providing 5v power to these
sensors.

expenses the proposed model can easily be deployed to manage parking

problems in metropolitan cities. This model can cut down vehicle time,
fuel consumption, and pollutants and can save the time of common people
across the country.
To reduce the amount of manual labour, a servo motor is added managed
by NodeMCU which controls the automatic opening and closing of parking
gates. As shown in Figures 9.10 and 9.11, a webapp is created for both the
user and the administrator. The user side application will assist users in
determining which slots are booked and which slots are available, while
the administrator will be able to monitor the value of the sensor, calibrate
ON/OFF, and can access the web application to start and stop displaying the
sensor values.
150 Samyak Jain et al.

Figure 9.8 Flow chart of admin system.

An XAMPP server shown in Figure 9.12 assists in the setup of the local
host and MySQL query in myAdminPHP, which we used to sort the sensor
values directly driven by NodeMCU, and the application then queries those
values and displays to the user which slot is free and which slot is booked.
A basic prototype with eight slots and LDR sensors installed at the bot-
tom of each slot, which are connected to the NodeMCU as shown in
Figure 9.13. The NodeMCU takes the value and sends it to the server data-
base using Wi-Fi, and the web page shows which slot is occupied and which
slot is free based on the sensor values.
At the entry gate, a PIR sensor and servo motor, and at the exit gate, an
ultrasonic sensor is installed as shown in Figure 9.14. PIR will detect the
 IoT-based parking system using Web-App 151

Figure 9.9 Flow chart of user system.

Figure 9.10 User side of web-application.

152 Samyak Jain et al.

Figure 9.11 Admin side of web-application.

Figure 9.12 XAMPP server.

motion of the car and send data to Arduino Uno, which will control the
servo motor to open and close the gate. As we progress, a limit has to be
based on the number of parking slots available. For example, if four parking
slots are available, the gate will only open four times. At the exit gate, as the
car leaves, the ultrasonic sensor will inform the Arduino that another slot is
available, and Arduino will send data to the servo motor.
 IoT-based parking system using Web-App 153

Figure 9.13 LDR connected with NodeMCU.

Figure 9.14 Complete model.

154 Samyak Jain et al.

9.4 CONCLUSION

The objective of the proposed research work is to create an internet-based

system that allows users to check the availability of parking spaces. With the
use of simple and cost-effective components this objective has been achieved.
Low-cost sensors, real-time data collecting, and a cloud-based automatic
parking system are all part of the proposed system. There are numerous
technological alternatives available to ease urban traffic congestion and to
improve the efficiency and management of on-street parking. As a scope of
future work, deep learning algorithms can be integrated with an IoT-based
system to build a smart parking system.

REFERENCES

[1] Xia, F., Yang, L. T., Wang, L., & Vinel, A. (2012). Internet of things. International
Journal of Communication Systems, 25(9), 1101.
[2] Sharma, A., Jain, A., Gupta, P., & Chowdary, V. (2020). Machine learning
applications for precision agriculture: A comprehensive review. IEEE Access,
9, 4843–4873.
[3] Chowdary, V., Bera, T., & Jain, A. (2022). IoT solution reference architectures.
In Internet of things (pp. 39–52). CRC Press.
[4] Vakula, D., & Kolli, Y. K. (2017). Low cost smart parking system for smart
cities. In 2017 International Conference on Intelligent Sustainable Systems
(ICISS) (pp. 280–284). IEEE.
[5] Ramasamy, M., Solanki, S. G., Natarajan, E., & Keat, T. M. (2018). IoT based
smart parking system for large parking lot. In 2018 IEEE 4th International
Symposium in Robotics and Manufacturing Automation (ROMA) (pp. 1–4).
IEEE.
[6] Pham, T. N., Tsai, M. F., Nguyen, D. B., Dow, C. R., & Deng, D. J. (2015). A
cloud-based smart-parking system based on Internet-of-Things technologies.
IEEE Access, 3, 1581–1591.
[7] Aydin, I., Karakose, M., & Karakose, E. (2017). A navigation and reserva-
tion based smart parking platform using genetic optimization for smart cities.
In 2017 5th International Istanbul Smart Grid and Cities Congress and Fair
(ICSG) (pp. 120–124). IEEE.
[8] Kodali, R. K., Borra, K. Y., GN, S. S., & Domma, H. J. (2018). An IoT based
smart parking system using LoRa. In 2018 International Conference on
Cyber-enabled Distributed Computing and Knowledge Discovery (CyberC)
(pp. 151–1513). IEEE.
[9] Latif, S., Afzaal, H., & Zafar, N. A. (2018, December). Modelling of graph-
based smart parking system using internet of things. In 2018 International
Conference on Frontiers of Information Technology (FIT) (pp. 7–12). IEEE.
[10] Park, Wan-Joo, et al. Parking space detection using ultrasonic sensor in park-
ing assistance system. In 2008 IEEE Intelligent Vehicles Symposium IEEE,
2008.
Chapter 10

A next-gen IoT-based
semi-automatic mobile manipulator
Mukul Kumar Gupta
University of Petroleum and Energy Studies, Dehradun, India

Shival Dubey
Institute of Design, Robotics & Optimization, University of Leeds,
United Kingdom

CONTENTS

10.1 Introduction ................................................................................ 155

10.1.1 Classification of mobile robots ....................................... 156
10.1.2 Mobile base robotic manipulator ................................... 157
10.1.3 IoT-based robotic manipulator ....................................... 158
10.2 Design of the proposed semi-automatic robot manipulator ......... 159
10.3 Proposed model .......................................................................... 160
10.3.1 Denavit-Hartenberg parameters ..................................... 161
10.3.2 Frame Assignment for D-H Parameter Calculation ......... 161
10.4 Robot control ............................................................................. 162
10.4.1 Joint space control .......................................................... 163
10.4.2 Task space control .......................................................... 163
10.4.3 Robotic arm simulation using MATLAB GUI ................. 164
10.5 Hardware Implementation .......................................................... 166
10.6 Conclusion and future scope ....................................................... 166
References �������������������������������������������������������������������������������������������� 167

10.1 INTRODUCTION

The Internet of Things (IoT) has been one of the most familiar features of
recent years, scaling new heights and creating a benchmark in the world.
It has transformed the real world into an intelligent communication world.
In addition, existing challenges are highlighted in these areas.
The value of IoT is increasing, in both commercial and daily settings. In
many ways, it is currently making our lives easier and it will certainly con-
tinue to do so. It is addressing challenges which we didn’t even understand
were a challenge, along with the problems we have, before the solution has
magically emerged. The chances are that you already understand how useful

DOI: 10.1201/9781003407300-10 155

156 Mukul Kumar Gupta and Shival Dubey

IoT is in your daily life through the use of smart thermostats, remote door
locks, wired home hubs and all of the different app-controlled appliances.
The application considered in this chapter is a semi-automatic mobile
manipulator. A mobile robotic manipulator system is a robotic articulated
arm mounted over a mobile base to improve mobility and achieve greater
degrees of freedom [1–3]. In such mechanical structures, the degrees of free-
dom of the robotic arm is enhanced with the degrees of freedom of the
mobile base. There is also an improvement in the workspace of the robotic
manipulator when it is mounted on a mobile base rather than a fixed base.
Any object in free space is represented with a degree of 6 DOF, that is, 3 to
represent the object in the cartesian plane and 3 degrees to denote the object
orientation. Therefore, to fully manipulate an object in a free space 6 joints
are needed to achieve complete movement [4, 5].
When a robotic manipulator has additional degrees of freedom, this is
known as kinematic redundancy. When the number of degrees of freedom or
robotic system joints exceeds the number of controllable variables, redun-
dancy arises. A kinematically redundant mobile manipulator has more
DOFs than is necessary to undertake its task. In this scenario, the inverse
kinematics issue yields an unlimited number of solutions. Secondary objec-
tives, such as avoiding mutual limits, singularities, and obstructions, can be
better met by selecting mobile manipulator configurations and a motion
trajectory from these redundant options.

10.1.1 Classification of mobile robots

Mobile robots can be classified into various categories depending on their
locomotion systems, design aspects, and medium for the robot movement
and other technical aspects, such as terrain conditions, stability, controlla-
bility, manoeuvrability etc. They can be classified into major categories, as
illustrated in Figure 10.1, which is based on their locomotion, including sta-
tionary robots (arm/manipulators) and land robots. These are then further
classified into Wheeled Mobile Robots (WMRs), Legged Robots, tracked
locomotion, Hybrid, Aerial robots, water-based robots, space robots, and
Bomb Disposal robots [6–9]. WMRs are more energy-efficient than legged
robots.
With regard to stability and controllability, both wheel design and wheel
geometry play an important role. The stability of a mobile robot is based on
the number of wheels used for its motion. The controllability of a mobile
robot is the inverse of its manoeuvrability. Coordination between the robotic
arm and the platform, which is a speciality of mobile manipulation, is
required in various cases for systems combining manipulation and mobility
capabilities.
Classification of Manipulators: Manipulators are kinematic chains with
special tools as end effectors to handle objects and perform specific tasks,
such as welding, assembling and machining operations depending upon
 A next-gen IoT-based semi-automatic mobile manipulator 157

Figure 10.1 Types of mobile robots.

their application. Manipulators can be classified into various criteria based

on the following classification:

By Motion Characteristics: Planar manipulator, Spherical manipulator,

Spatial manipulator
By Kinematic Structure: Serial robot, parallel robot, Hybrid Robot

10.1.2 Mobile base robotic manipulator

Mobile manipulators are either autonomously operated or teleoperated
remotely. A manipulator and a mobile platform are the two subsystems
that make up the mobile manipulator system, as shown in Figure 10.2.
End effector trajectory and mobile platform trajectory are to be managed
simultaneously for the performance of complex tasks such as picking up an
object while moving and avoiding an obstacle. Both trajectories should be
followed concurrently by the end effector to assure the gripping of the item
while avoiding the barrier to align along with the object orientation and the

Figure 10.2 Mobile base robotic manipulator.

158 Mukul Kumar Gupta and Shival Dubey

mobile base has to follow a trajectory to avoid the obstacle in the case of
autonomous systems.

10.1.3 IoT-based robotic manipulator

The number of gadgets connecting with one another has expanded dramati-
cally over recent years. The Internet of Things (IoT) allows items to commu-
nicate with one another and to perform tasks such as sensing, identifying,
and networking. IoT applications have grown rapidly across all study topics
as a result of the limitless number of possible interactions, and they are now
leading new research fields.
The Internet of Things is used for remote monitoring and the remote con-
trol of devices. Robotic manipulator parameters, such as position, accelera-
tion, tilt, etc., can be monitored to ascertain the manipulator’s position and
can also be controlled to decide its position. In Figure 10.3, a sample archi-
tecture of an IoT-based manipulator is depicted in which a microcontroller
acts as the heart of the system. Sensors for measuring important parameters
will be placed on the body of the manipulator and the data measured from
these sensors will be given to the microcontroller. A predefined algorithm/
code will be running on the microcontroller which takes the sensor param-
eters as an input to decide the position of the manipulator. In order to initi-
ate a control action to the manipulator, a threshold will be set in the
algorithm; if the measured parameters breach this threshold then the micro-
controller initiates the control action to the manipulator [9–11].
Edge computing is an advanced IoT technique in which information pro-
cessing is performed on the end device rather than in the cloud. This reduces
the computation load on the cloud to a great extent and thus is helpful in
fast and parallel processing on the robot side. To implement edge computing
on the robot an array of sensors are required, such as:

i. A proximity sensor to provide information on the presence of an

obstacle and also to calculate the safe distance between the robot and
the obstacle. There are different types of proximity sensors, such as
Infrared sensors, Ultrasonic sensors, Passive Infrared sensors (PIR), etc.

Figure 10.3 IoT based robot manipulator.

 A next-gen IoT-based semi-automatic mobile manipulator 159

ii. A tactile sensor to provide the contact information of a robot and

object. As proximity sensors are used to detect objects, tactile sen-
sors are used to avoid objects. Examples of tactile sensors are touch
or contact sensors and force sensors, which are used in the grippers
of the robot to calculate the force to be applied while gripping any
object.
iii. Temperature and humidity sensors to record the operating conditions
of the environment where the robot will be operating. Temperature
sensors are of contact and non-contact types. Contact-type tempera-
ture sensors measure the temperature of liquid materials whereas non-
contact temperature sensors operate in the open environment.
iv. Navigation sensors to provide path planning information, along with
the position sensor which gives the exact position of the sensor.
v. Acceleration and tilt sensors to provide the rate of robot acceleration
and its tilt, if any.

The robot can automatically acquire and send its information and the infor-
mation about the environment using IoT and sensor technology. With the
help of a feedback mechanism on the robot, it is also possible to process
this information. One of the key applications based on information from
the sensor network is the inspection and control of physical objects with
the most up-to-date status, which includes both static properties, viz., the
inheritance features of objects such as shape, size, and colour, and dynamic
properties, such as the real-time position, movement, gesture, and motion of
objects that will change over time, etc. These benefits are critical in allowing
large sensor nodes to capture multimodal sensory data in the future ubiqui-
tous IoT framework.
IoT activities are centred on managing, monitoring, and optimizing sys-
tems and processes using linked devices with simple, onboard passive sen-
sors. Despite its impact, the IoT solution’s full potential could be realized by
delving deeper into the more complex and transformative features of perva-
sive connectivity to and communication among smart objects. Because of
their innate ability to detect, think (compute), act (manipulate), and move
about, robotic systems can assist in this shift (mobility).

10.2 DESIGN OF THE PROPOSED SEMI-AUTOMATIC

ROBOT MANIPULATOR

The proposed mobile manipulator is a 5DOF Manipulator arm with a two-

finger gripper as the end effector mounted over a 3 DOF Mobile Robot with
a tracked wheel chassis. Altogether this makes it a 9 DOF Mobile manipu-
lator system. The proposed structure is redundant, and it can manipulate
in any environment. Redundancy in the manipulator ensures that it can
manipulate tasks in an unstructured environment also.
160 Mukul Kumar Gupta and Shival Dubey

Figure 10.4 Design of the proposed model.

The manipulator considered is depicted in Figure 10.4, with its design

features, components attached and dimensions. The drawing is extracted
from the 3D model and an isometric view is projected to display sketch
dimensions in mm standard units and inches. The proposed design is of the
following dimensions: length 6 inches (150 mm), width 4 inches (100mm),
and height of 10 inches (254 mm).

10.3 PROPOSED MODEL

The proposed mobile manipulator is of 9 DOF consisting of a 6DOF

Manipulator arm with a two-finger gripper as the end effector mounted
over a 3 DOF Mobile Robot with a tracked wheel chassis. The 3 DOF
mobile robot is an Unmanned ground vehicle (UGV) mounted with a 5 DOF
robotic serial manipulator arm on the top with an end effector used to pick
and place objects in its workspace. The UGV is also equipped with an FPV
camera to obtain the perspective view of the robot to navigate in the envi-
ronment and also to perform object detection to detect and trace objects
around it and identify target objects based on the algorithm. The following
solid model, as shown in Figure 10.5, is used to develop mathematical mod-
elling of the system.
Kinematics and dynamics are key parameters in the analysis of robot
manipulators. In kinematics under motion, time is not considered directly
whereas in dynamics the study of motion considers time. For robot manipu-
lators, there are two approaches for kinematic analysis: the first is forward
or direct kinematics, and the second is inverse kinematics. In forwarding
kinematics, initial joints are known and the final robot position has to find
out whereas in inverse kinematic end-effector positions are known and the
robot’s initial position has to be found out.
Dynamic analysis can be done either with Euler Lagrange or the Newton
Euler approach. With dynamics analysis, one can find out the necessary
 A next-gen IoT-based semi-automatic mobile manipulator 161

Figure 10.5 CAD model of the proposed mobile manipulator.

torque required for robot deployment. When one has fewer links then the
Euler Lagrange approach is better, but in the case of more number links the
newton Euler approach is the most suitable.

10.3.1 Denavit-Hartenberg parameters
The Denavit-Hartenberg (D-H) notation is used to describe the location
of each link of the manipulator in relation to its neighbouring link. Then,
for a given set of joint angles, the resultant pose (position and orienta-
tion) is calculated. Any robot can be described kinematically with values of
four quantities for each link/joint. Two of these quantities describe the link
geometry and the other two describe the link’s connection to its neighbour-
ing link.

10.3.2 Frame Assignment for D-H Parameter

Calculation
The proposed model uses a 5 DOF Robotic Arm with 5 revolute joints.
It consists of a base, shoulder, elbow, wrist and end effector with roll and
pitch, as shown in Figure 10.6. To perform the kinematic analysis the stan-
dard Denavit-Hartenberg convention and methodology are used to derive
the kinematics. This is used to find out the position and orientation of the
end effector with respect to its base.
Similarly, 3 virtual frames are considered to write the D-H parameters for
the 3 DOF mobile robot base. These frames are corresponding to the three
dots of the mobile base, of which two are for translation and one is for rota-
tion (Table 10.1).
162 Mukul Kumar Gupta and Shival Dubey

Figure 10.6 Frame diagram for the manipulator.

Table 10.1 D-H parameter of the manipulator

Joint i Type α (degrees) a (mm) d (mm) θi(degrees)
1 Base 0 0 d1 θ1
2 Shoulder 90 0 0 θ2
3 Elbow 0 a3 0 θ3
4 Wrist 90 a4 0 θ4
5 Gripper 90 0 d5

i α a D θ
1 3pi/2 0 D1 Pi/2
2 3pi/2 0 D2 3pi/2
3 3pi/2 0 0 θ -3pi/2

10.4 ROBOT CONTROL

The control of robotic manipulators is one of the most challenging tasks as

the robot is highly nonlinear in nature. Theoretically, there are various tech-
niques available in nature but in practice mostly PID controller used as a
PID controller is easy to implant and cost-effective in nature [12]. The princi-
pal nonlinear control techniques are adaptive control, robust control, sliding
control, intelligent control, optimal control etc. [13–16]. Stability is also one
of the major tasks in the control of robot manipulators as this is one of the
main objectives of control. Most of the literature uses the Lyapunov-based
approach as this is an energy-based approach. Other theoretical approaches
can be divided into joint space-based and task space-based controllers [17, 18].
 A next-gen IoT-based semi-automatic mobile manipulator 163

Figure 10.7 Relation between joint space and cartesian space.

Nonlinear dynamics and coupling effects have to consider for a better and
more accurate controller design.

10.4.1 Joint space control

The task space is used to train and direct robots to perform certain tasks.
In the space-based joint control, inverse kinematic methods are used to map
activities into joint space. The relation between joint space and cartesian
space is shown in Figure 10.7. The controller is then developed in torque
space based on the joint space information. Because the mapping from joint
space to task space is injective, the robot may be directed to track the trajec-
tory in joint space, and then the trajectory in task space can be followed.
Joint space controllers can be further divided into two groups:

a. classic joint control

b. model-based control

During the early days of robotics, because no computing is required for intri-
cate kinematic and dynamic effects, this solution may be readily conceived,
analysed, and implemented using the single-input single-output classical
PID technique. Traditional joint control is not only simple to use, but also
steady and reliable. This controller is the most popular to date because it is
the one most suited to real-world applications. Moreover, recent nonlinear
control techniques are far too complex for industrial robotic applications.

10.4.2 Task space control

The task space controller is designed differently from the joint space con-
troller. Task space control has some advantages over joint space control,
which are as follows:

• Controllers are designed based directly on task space information so

they are more sensitive to environments.
164 Mukul Kumar Gupta and Shival Dubey

Figure 10.8 Joint space control.

Figure 10.9 Task space control.

• Robots can be properly controlled, even in the case of kinematic

singularity.
• Robot programming can be faster.

Task space control, on the other hand, is less well known than joint space
control, and its design technique is less clearly defined. In task space,
feedback gains are not apparent, and joint space behaviour is difficult to
anticipate.
The joint space is the control action whereas the task space is the opera-
tional space control.
Figure 10.8 shows the basic outline of the joint space control methods.
The main point of the diagram is that torque should be applied to the
manipulator and also that one can control position and velocity. The output
position should tend to match the desired position as the difference between
these two is known as an error. As time tends to infinity, error tends to zero.
Figure 10.9 shows a schematic diagram of the operational space control
methods. Because operational space controllers use a feedback loop to
directly decrease job mistakes, there are various advantages to this tech-
nique. Now, the motion between points can be a straight-line segment in the
task space.

10.4.3 Robotic arm simulation using MATLAB GUI

A 5DOF robotic arm is simulated shown in Figure 10.10 using a MathWorks
open-source GUI and a robotic system toolbox to derive forward and inverse
kinematics. The joint angles are used to determine the cartesian coordinates
A next-gen IoT-based semi-automatic mobile manipulator 165
Figure 10.10 3 link manipulator MATLAB GUI simulation.
166 Mukul Kumar Gupta and Shival Dubey

Figure 10.11 Robot prototype (a) Indoor environment, (b) Outdoor environment.

of the end-effector and vice versa. The robotic arm when placed on the mov-
able mobile robot base, the end effector coordinates alter with respect to the
position of the robot base.

10.5 HARDWARE IMPLEMENTATION

The proposed model, as shown in Figure 10.11, is implemented on hard-

ware using a Raspberry Pi 4 Model B computer and it is teleoperated with
the help of the Internet protocol using SSH on a remote laptop or mobile
device. The Raspberry Pi computer is programmed using Python scripts to
teleoperate the mobile robot using arrow keys and custom designated keys
to operate the robotic arm. The Object Detection Algorithm is performed on
the USB camera connected to the Raspberry Pi to detect the objects in the
robot’s field of view [19].

10.6 CONCLUSION AND FUTURE SCOPE

This chapter describes the research that has been carried out to develop a
working prototype of a Semi-Automatic Mobile Manipulator using IoT
applications. This is developed using a tracked wheel mobile robot. This fea-
ture facilitates the robot to navigate conveniently on all platforms irrespec-
tive of the terrain as the mechanism is designed to climb rocks and navigate
 A next-gen IoT-based semi-automatic mobile manipulator 167

in rugged environments. A 5 DOF robotic manipulator is mounted over the

mobile base to improve the workspace of the robotic arm for pick and place
applications. The developed system covers mechanics of systems design,
dynamic modelling and simulations, and control system design. The mobile
manipulator developed is the redundant one. The design of the proposed
solid model is done in FUSION 360 software and the proposed model has
the manipulator having 5 degrees of freedom on which the mobile base to
be mounted has 3 degrees of freedom. For the future scope of this work,
the development of a fully autonomous bomb disposal robot with improved
object detection algorithms and the addition of an extra manipulator making
it a dual-arm mobile manipulator for added dexterity can be implemented.
IoT with robotics is mostly used for monitoring the position, the obstacle and
other important parameters. It can also be used to control the motion, and
the trajectory of the robot along with the implementation of its path plan-
ning. The use of IoT is desirable in conditions where human intervention is
involved in next to impossible tasks, such as explosive detections, firefighting,
forest fire detection and prediction, etc.

REFERENCES

[1] Rubio, F., Valero, F., and Llopis-Albert, C., “A review of mobile robots:
Concepts, methods, theoretical framework, and applications”. International
Journal of Advanced Robotic Systems, 16(2), pp. 1–22, 2019.
[2] Ashith Shyam, R. B., Hao, Z., Montanaro, U., Dixit, S., Rathinam, A., Gao, Y.,
Neumann, G., and Fallah, S., “Autonomous robots for space: Trajectory learn-
ing and adaptation using imitation”. Frontiers in Robotics and AI, 8, pp. 1–10,
2021.
[3] Sivčev, S., Coleman, J., Omerdić, E., Dooly, G., and Toal, D., “Underwater
manipulators: A review”. Ocean Engineering, 163, pp. 431–450, 2018.
[4] Luo, D., and Yu, L., “From factory to site—Designing for industrial robots
used in on-site construction” In Automating cities (pp. 87–109). Springer,
Singapore, 2021.
[5] Van Wynsberghe, A., Healthcare robots: Ethics, design and implementation.
Routledge, 2016.
[6] Gupta, A., Singh, A., and Verma, V., Amit Kumar Mondal and Mukul Kumar
Gupta published in the advanced robotics titled, developments and clinical
evaluations of robotic exoskeleton technology for human upper-limb rehabili-
tation, Vol 34, Taylor and Francis, April 2020.
[7] Asian Defence, “Bomb Disposal Robot Daksh”. Theasiandefence.blogspot.
com. Retrieved 2010-08-31, 2009.
[8] Naskar, S., Das, S., Seth, A. K., and Nath, A., Application of radio frequency con-
trolled intelligent military robot in defense. In 2011 International Conference on
Communication Systems and Network Technologies (pp. 396–401). IEEE, 2011.
[9] Chowdary, V., Kaundal, V., Sharma, P., and Mondal, A. K., Implantable elec-
tronics: Integration of bio-interfaces, devices and sensors. In Medical big data
and internet of medical things (pp. 55–79). CRC Press, 2018.
168 Mukul Kumar Gupta and Shival Dubey

[10] Chowdary, V., Mondal, A. K., Sharma, A., and Capoor, S., “Pre-deployment
strategy for maximizing barrier coverage in wireless sensor network”.
International Journal of Sensors Wireless Communications and Control, 11(2),
pp. 181–188, 2021.
[11] Meera, C. S., Sairam, P. S., Vineeth, V., Adarsh, K., and Gupta, M. K., “Design
and analysis of new haptic joysticks for enhancing operational skills in excava-
tor control”. Journal of Mechanical Design by ASME, 142(12), pp. 460–467,
2022.
[12] Zhou, H., DC servo motor PID control in mobile robots with embedded DSP.
In International Conference on Intelligent Computation Technology and
Automation (ICICTA) (Vol. 1, pp. 332–336). IEEE, 2008.
[13] Epenetus, A. B., Meera, C. S., Mohan, S., and Gupta, M. K., “Development
and motion control of spatial serial robotic manipulator under varying distur-
bances”. World Journal of Engineering, 16(4), pp. 460–467, 2019.
[14] Gupta, M. K., Kumar, R., Verma, V., and Sharma, A. “Robust control based
stability analysis and trajectory of triple link robot manipulator”. Journal
Européen des Systèmes Automatisés, 54(4), pp. 641–647, 2021.
[15] Kanayama, Y., Kimura, Y., Miyazaki, F., and Noguchi, T., “A stable track-
ing control method for an autonomous mobile robot”. In Proceedings. IEEE
International Conference on Robotics and Automation (pp. 384–389), IEEE,
1990.
[16] Samson, C. and Ait-Abderrahim, K., “Feedback control of a nonholonomic
wheeled cart in cartesian space”. In Proceedings. 1991 IEEE International
Conference on Robotics and Automation (pp. 1136–1137). IEEE Computer
Society, 1991.
[17] Murray, R. M. and Sastry, S. S., Nonholonomic motion planning: Steering
using sinusoids. IEEE Transactions on Automatic Control, 38(5), pp. 700–716,
1993.
[18] Khatib, O., “Real-time obstacle avoidance for manipulators and mobile roots”.
In Autonomous robot vehicles (pp. 396–404). Springer, 1986.
[19] Abbaspour, R., “Design and implementation of multi-sensor based autono-
mous minesweeping robot”. In International congress on ultra-modern tele-
communications and control systems (pp. 443–449). IEEE, 2010.
Chapter 11

Pest identification and classification

using IoT enable technique
Atul B. Kathole
D. Y. Patil Institute of Technology, Pune, India

Sonali D. Patil and Kapil N. Vhatkar

Pimpri Chinchwad College of Engineering, Pune, India

Dinesh Chaudhari
JDIET, Yavatmal, India

Avinash P. Jadhav
DRGIT&R Amravati, Yavatmal, India

CONTENTS

11.1 Introduction ................................................................................ 169

11.2 Literature survey ......................................................................... 171
11.2.1 Related works ................................................................ 171
11.2.2 Problem statement .......................................................... 172
11.3 A novel deep learning framework for IoT-enabled pest
identification and classification ................................................... 174
11.3.1 Proposed architecture and description ............................ 174
11.3.2 IoT-enabled pest detection and smart agriculture ............ 176
11.4 Conclusion .................................................................................. 177
References �������������������������������������������������������������������������������������������� 178

11.1 INTRODUCTION

Identifying and classifying crop pests is one of the vital challenges in the
field of agriculture [1]. Insects are the most important factor in causing dam-
age to crops and reducing crop productivity. Insect classification is known
as a complex task owing to its complex structure and also with the high
similarity in appearance among the distinct species [2]. It is very essential
for recognizing and classifying the insects present in the crops at an earlier
stage by providing highly effective pesticides and also introducing biologi-
cal control methods to prevent the spread of insects that cause crop dis-
eases [3, 4]. The traditional way of identifying insects is considered to be

DOI: 10.1201/9781003407300-11 169

170 Atul B. Kathole et al.

inefficient, time-consuming, and labor-intensive. The vision-based comput-

erized system of image processing has been implemented based on machine
learning in order to accurately perform the classification and identification
of insects so as to overcome the conventional problems in the field of agri-
culture research.
IoT is a well-known revolutionary technology to enable future communi-
cations and computing. The world population is highly dependent on agri-
culture for income and food resources. Therefore, intelligent Information
Technology (IT) technologies are necessary for overcoming the challenges of
traditional agricultural approaches. IoT makes it easier for farmers by pro-
viding a lot of techniques to achieve sustainable and precise agriculture pro-
duction in order to address challenges in the agricultural field [5]. IoT
technology supports farmers to collect information regarding natural sce-
narios such as soil fertility, temperature, moisture, and weather. The online
crop monitoring system enhances agricultural productivity and crop growth,
detects animal intrusion into the field, and is useful in detecting weed, pest,
and water levels [6]. IoT enables farmers to monitor their agricultural field
anywhere in the world at any time. The main intention of the IoT is to
extend the network by concatenating various kinds of connected devices
[7, 8]. IoT has mostly focused on three aspects of the system: cost-saving,
automation, and communication. IoT encourages people to pursue their
routine activities based on the internet and to focus on their principal activ-
ity of growing crops [9].
An automated insect identification system is implemented to interpret
seven geometrical features. It also makes use of deep learning and machine
learning algorithms to obtain better results by considering a smaller number
of insect classes. When involving the machine learning algorithms, the clas-
sification accuracy has mainly relied on the structure of the extracted fea-
tures; thus, the optimal features are selected for machine learning, which
maximizes the level of accuracy in computation [10]. In addition, the accu-
racy needs to be improved by incorporating deep learning algorithms for
categorizing huge image datasets [11]. The deep learning algorithm is used
to perform the automated feature extraction by using the raw data, which
decreases the challenges of the hand-crafted features, and also to address the
highly complex issues related to image classification. Recently, the deep
learning approaches are investigated with the help of Convolutional Neural
Networks (CNNs), which ensure promising solutions for the existing chal-
lenges [12]. When compared to machine learning techniques, deep learning
techniques prove helpful in automatically obtaining the representative fea-
tures from the training data set, which avoids complex image-processing
steps and labor-intensive feature engineering so as to satisfy a wide range of
outdoor conditions [13]. An effective deep CNN model is developed for
classifying the insect species of field-crop insect images, which provides
higher accuracy in classification accuracy. Thus, deep learning techniques
have great potential for pest detection in practical applications. Hence, it is
 Pest identification and classification using IoT enable technique 171

significant to develop a new IoT-enabled pest identification and classifica-

tion model with the help of a deep learning approach.
The main contributions of the research works are given as follows:

• To develop a new IoT-based pest identification and classification model

for accurately detecting the pest in the crop field and reducing their
effects at the earlier stage for improving the crop production in the
agricultural areas.
• To integrate an enhanced deep architecture named CNLSTM for
extracting the deep features from the YOLOv3-based detected images
and classifying the extracted features along with the parameter opti-
mization using the suggested AHBA to identify the type of pest present
in the images.

The remaining section of the developed model is described as follows.

Section 11.2 explains the related works and the problems. In section 11.3,
the proposed pest identification and classification model is depicted. In
­section 11.4, the pest identification and classification model is summarized.

11.2 LITERATURE SURVEY

11.2.1 Related works
In 2021, Kumar et al. [14] introduced an enhanced model for identifying
the pests that were affecting the rice at the time of crop productivity. Here,
the IoT-based mechanism was used for passing the rice pest images to the
cloud storage and has provided the pest information. When the pest was
identified, the information regarding the presence of the pest was sent to
the farmers or owners for taking respective actions. The analysis results
have shown that the proposed approach has minimized the rice wastage in
the productivity field through the continuous monitoring of the pests in the
rice field. In 2020, Chen et al. [15] have implemented a deep learning-based
model for obtaining the insect locations and analyzing the environmental
information from the weather stations in order to obtain the pests’ informa-
tion in the field with the help of an enhanced deep learning approach. The
experimental results have shown that the proposed method have secured
improved identification accuracy. Precise identification of the insects and
pest has decreased the amount of pesticide usage, which has also minimized
the pesticide damage to soil.
In 2021, Turkoglu et al. [16] presented two types of classification
approaches with the help of deep feature extraction that were obtained from
the pre-trained CNN. The proposed model was validated with the help of
diverse diseases and pest images. It was observed that the accuracy scores
were better with the majority of the ensemble model and provided improved
172 Atul B. Kathole et al.

performance when compared with the existing models. In 2019, Liu et al.
[17] have developed an end-to-end method for classifying and detecting the
huge multi-class pests with the help of deep learning. The three major parts
of the proposed framework were a novel module with an attention-based
approach, the developed neural network for ensuring the region proposals,
and a score map for classifying the pest and bounding box regression. The
experimental analysis was carried out to demonstrate the effectiveness of
multi-class pest detection through the proposed model.
In 2021, Li et al. [18] have involved a novel technique for enhancing the
accuracy of small pest detection. The suggested framework was trained with
the help of a transfer learning methodology using the tiny pest training set.
Here, the developed deep learning architecture has provided better perfor-
mance than other approaches. The analysis has shown that the proposed
method has ensured the robust performance in detecting the tiny pests at
varied light reflections and pest densities. In 2018, Yue et al. [19] proposed
an enhanced residual-based network for detecting the problem in the crop
field. The proposed method was correlated with the traditional approaches
and has demonstrated the high power capacity of the developed model for
image reconstruction. The analysis results have shown that the proposed
approach has revealed an enhanced recall rate for pest detection.
In 2021, Wang et al. [20] have developed an efficient deep learning model
in the pest monitoring system for automatically detecting and counting the
pest in the rice planthoppers. Here, the proposed approach was developed
to extract the high-quality regions in the pest images, even the tiny ones. The
analysis results have shown that the suggested approach has improved rec-
ognition recall compared with the state-of-the-art approaches. In 2019,
Thenmozhi et al. [21] implemented an elevated deep learning network for
classifying the insect species through the three available data sets. The sug-
gested approach was validated with the other deep learning architectures
under the insect classification. Further, this model has included transfer
learning for tuning the pre-trained models. The experimental results have
shown that the suggested model was effective in classifying the different
types of insects and also applicable in the agricultural sector for crop
protection.

11.2.2 Problem statement
Plant pests are the most important factor in causing the massive loss in
agricultural production, along with the social, ecological, and economical
implications. It is essential to recognize and classify the insects present in
the crops at the early stage. This is to avoid the insect spread into the crop,
resulting in crop diseases by choosing the efficient biological control and
pesticides approaches. Numerous features and challenges of agriculture pest
detection are reviewed in Table 11.1. Artificial intelligence [14] decreases
rice wastage at the time of production by monitoring the pests at the regular
 Pest identification and classification using IoT enable technique 173

Table 11.1 Features and challenges of agriculture pest detection

Author
[citation] Methodology Features Challenges
Kumar Artificial • It decreases the rice • However, the improved
et al. [14] Intelligence wastage at the time of technique needs to be
production through developed for better
monitoring the pests at performance.
the regular interval of
time.
Chen et al. YOLOv3, • It reduces the damages • Yet, there is requirement
[15] LSTM that are caused in for enhancing the
the environment by perspectives in the
involving enormous images to solve the
usage of pesticides and issues related to
also increases the crop insufficient training
quality. samples.
Turkoglu Ensemble • It ensures high • But, it cannot handle the
et al. [16] learning robustness. imbalance problem of
training data.
Liu et al. CNN, • It is more robust for • On the other hand,
[17] Channel- detecting the tiny pests the size of the
spatial on an image. sample requires to be
attention maximized in diverse
external scenarios for
gaining better results.
Li et al. TPest-RCNN • It helps to improve • However, this model
[18] the performance technically difficult
by maximizing the because of the limits
replacement frequency of computer-vision
of traps. technology based on
visible-range images.

interval of time. However, an improved technique needs to be developed

for better performance. YOLOv3, LSTM [15] reduces the damages that are
caused in the environment by involving an enormous usage of pesticides
and also increases the crop quality. Yet there is a requirement to enhance
the perspectives in the images to solve the issues related to insufficient train-
ing samples. Ensemble learning [16] ensures a high level of robustness.
But it cannot handle the imbalance problem of training data. CNN and
channel-spatial attention [17] are more robust in detecting the tiny pests
on the image. On the other hand, the sample size requires to be maximized
in diverse external scenarios to gain better results. TPest-RCNN [18] helps
to improve the performance by maximizing the replacement frequency of
traps. However, this model has some technical difficulties because of the
limits of computer vision. Deep CNN [19] is used to reduce the density of
the monitoring cameras that are employed for surveillance. But there is a
need to improve the performance of the model against images with motion
174 Atul B. Kathole et al.

blur or noise through the process of DnCNN, Deblur GAN and other image
enhancement approaches. CNN [20] performs well with regard to small pest
detection. However, this model achieves poor levels of accuracy. CNN [21]
has improved potential with regard to pest detection, especially in outdoor
applications. But it requires more time for the processing of a large number
of data. Therefore, a new pest detection model using deep learning and IoT
is required to advance a solution for these abovementioned drawbacks.

11.3 A NOVEL DEEP LEARNING FRAMEWORK FOR

IoT-ENABLED PEST IDENTIFICATION AND
CLASSIFICATION

11.3.1 Proposed architecture and description

A remote monitoring mechanism with IoT devices is the vital technique
involved in diverse applications such as object tracking in smart cities,
healthcare, human surveillance, and modern farming. With the help of IoT,
insect control can be monitored and managed from anywhere in the world.
In [11], the IoT network is used along with a wireless imaging system for
developing the remote greenhouse pest monitoring system. Here, the blob
counting and k-means color clustering algorithm are used in the imaging
system to automatically count the insects and pests present in the trap sheet.
Similarly, in another research study, the IoT-based smart farm field manage-
ment methodology is implemented to monitor crop growth, detecting the
insects in the crop field and also determining the appropriate pesticide to
manage the crop pests. On the other hand, the automated identification of
the insects and pests are the major challenges in the pest monitoring systems.
Hence, the deep learning and machine learning algorithms are used for the
decision-making and object detection in diverse insect control mechanisms.
In [12], the effective performance of the machine learning, deep learning,
and computer vision algorithms are utilized for pest detection, especially
in the tomato farms. This study has shown that the deep learning architec-
ture provides enhanced performance when compared among three consid-
erations. From the literature, it is clear that IoT is more important for pest
monitoring systems, whereas the deep learning techniques are known to be
the optimal approach for detecting and classifying insects and pests from
the crop images. Thus, IoT and deep learning are combined into the pest
identification and classification model in order to offering increased benefits
to the farmers. The architecture has been diagrammatically represented in
Figure 11.1.
A new pest identification and classification model with IoT using the deep
learning architecture is developed to identify and classify the pests in the
crops so as to reduce the usage of fertilizers and increase crop production by
preventing the pests at an earlier stage of crop growth. The IoT technology
 Pest identification and classification using IoT enable technique 175

Figure 11.1 P
 roposed pest identification and classification architecture with IoT
and deep learning approach.
176 Atul B. Kathole et al.

is used for collecting the required crop images from the agricultural fields
through the sensors. These collected images are considered at the object
detection phase, where the YOLOv3 detector is used to detect the pest
regions in the given input images. The detected images from the YOLOv3
are given to the CNLSTM network, where the CNN framework is used to
extract the most essential features of the detected pest images. The pest fea-
tures are passed toward the developed LSTM network, which is used to
classify these features into different pest classes in the agricultural field. The
performance of the classification model is further enhanced by the proposed
AHBA on conducting the optimization in the hidden neurons of the LSTM
network, which intends to achieve the maximization of accuracy in the clas-
sification phase in the proposed optimal pest classification model.

11.3.2 IoT-enabled pest detection and smart

agriculture
Agriculture is defined to be a science of crop production, animal husbandry
and soil cultivation, where the resultant products need to be marketed in
an effective manner. Food demand has been increasing abundantly, in both
qualitative and quantitative aspects, which can be satisfied by incorporat-
ing the computer technology into the agricultural practices. Owing to the
growth in the world’s population, it is also necessary to increase the crop
production regarding the requirements of food in terms of nutrition. Crop
production can be mainly affected by various diseases that are caused due
to certain factors, such as the presence of insects and pests in the crop field.
This needs to be prevented by reducing the disease-affected crop in the agri-
cultural areas. Hence, the IoT technology is required to automatically detect
the insects and pests through the IoT sensor devices, which helps farmers
to monitor and remove the pests that affect the crops during the earlier
stages of crop growth. In the traditional way, the insects and pests are deter-
mined through the manual way by the medical experts, which may be time-
consuming and inaccurate. To avoid these shortcomings, the automatic and
remote identification of pests can be achieved through IoT. IoT-based agri-
cultural systems provide accurate results when determining the pest regions
in the crop field. Here, the sensors need to be properly installed and main-
tained in the agricultural areas. These sensors help farmers to identify those
planted areas that are affected by the insects and pathogens. Initially, the
sensors collect the data; these are then transferred into the centralized plat-
forms through the wireless network. This helps farmers to remotely moni-
tor and protect their crops from insect and pest attacks, also reducing the
possibility of environmental contamination and crop intoxication and also
minimizing the use of pesticides in the crop field. These developments illus-
trate how essential it is to incorporate IoT technology with agriculture in
order to improve crop production. The IoT-based system for pest detection
in smart agriculture is shown in Figure 11.2.
 Pest identification and classification using IoT enable technique 177

Figure 11.2 IoT-based pest detection in smart agriculture.

11.4 CONCLUSION

This research has proposed a novel pest identification and classification

model with implemented AHBA in order to achieve the accurate detection
of pest in the crop field. Initially, pest images were collected through sen-
sors by using IoT technology. The collected images were subjected to the
object detection phase, during which the YOLOv3 detector was utilized
to detect the pest regions in the given input images. The detected images
were obtained from the YOLOv3, and these were further given into the
CNN framework to extract the significant features from the pest images.
The extracted features were considered for the developed CNLSTM net-
work, where the optimal feature classification with proposed AHBA was
performed that has classified into different classes of pests in the agricul-
tural field. Thus, the overall performance of the proposed pest identification
178 Atul B. Kathole et al.

and classification model with implemented AHBA-CNLSTM was superior

to the other conventional techniques. Future research should try to imple-
ment the system on real-time images for better performance analysis when
compared with existing approaches.

REFERENCES

[1] M. Khanramaki, Ezzatollah Askari Asli-Ardeh and Ehsan Kozegar, “Citrus

pests classification using an ensemble of deep learning models,” Computers
and Electronics in Agriculture, Vol. 186, No. 106192, pp. 1–11, July 2021.
[2] M. Turkoglu, D. Hanbay and A. Sengur, “Multi-model LSTM-based convolu-
tional neural networks for detection of apple diseases and pests,” Journal of
Ambient Intelligence and Humanized Computing, pp. 1–11, 2019.
[3] F. Wang, R. Wang, C. Xie, P. Yang and L. Liu, “Fusing multi-scale context-
aware information representation for automatic in-field pest detection and rec-
ognition,” Computers and Electronics in Agriculture, Vol. 169, No. 105222,
pp. 1–11, February 2020.
[4] Thenmozhi Kasinathan and Srinivasulu Reddy Uyyala, “Machine learning
ensemble with image processing for pest identification and classification in
field crops,” Neural Computing and Applications, Vol. 33, pp. 7491–7504,
2021.
[5] M. Ayaz, M. Ammad-Uddin, Z. Sharif, A. Mansour and E.-H. M. Aggoune,
“Internet-of-Things (IoT)-Based Smart Agriculture: Toward Making the Fields
Talk,” IEEE Access, Vol. 7, pp. 129551–129583, 2019.
[6] C.-J. Chen, Y.-Y. Huang, Y.-S. Li, Y.-C. Chen, C.-Y. Chang and Y.-M. Huang,
“Identification of fruit tree pests with deep learning on embedded drone to
achieve accurate pesticide spraying,” IEEE Access, Vol. 9, pp. 21986–21997,
2021.
[7] Y. Ai, C. Sun, J. Tie and X. Cai, “Research on recognition model of crop dis-
eases and insect pests based on deep learning in harsh environments,” IEEE
Access, Vol. 8, pp. 171686–171693, 2020.
[8] A. B. Kathole and Dr. Dinesh N. Chaudhari, “Pros & Cons of Machine learn-
ing and Security Methods,” 2019. http://gujaratresearchsociety.in/index.php/
JGRS, ISSN: 0374-8588, Vol. 21 N4.
[9] A. B. Kathole, Dr. P. S. Halgaonkar, and A. Nikhade, “Machine learning & its
classification techniques,” International Journal of Innovative Technology and
Exploring Engineering (IJITEE) ISSN: 2278-3075, Vol. 8 No. 9S3, pp. 1–7,
July 2019.
[10] A. B. Kathole and Dr. D. N. Chaudhari, “ Fuel analysis and distance predic-
tion using machine learning,” International Journal on Future Revolution in
Computer Science & Communication Engineering, Vol. 5, No 6, pp. 1–11, 2019.
[11] Mohamed Smail Karar, Fahad Alsunaydi, Sultan Albusaymi and Sultan
Alotaibi, “A new mobile application of agricultural pests recognition using
deep learning in cloud computing system,” Alexandria Engineering Journal,
Vol. 60, No. 5, pp. 4423–4432, October 2021.
 Pest identification and classification using IoT enable technique 179

[12] Everton Castelão Tetila, Bruno Brandoli Machado, Gilberto Astolfi, Nícolas
Alessandro de Souza Belete, Willian Paraguassu Amorim, Antonia Railda Roel
and Hemerson Pistori, “Detection and classification of soybean pests using
deep learning with UAV images,” Computers and Electronics in Agriculture,
Vol. 179, No.105836, pp. 1–11, December 2020.
[13] D. J. A. Rustia, Chen-YiLu, Jun-Jee Chao, Ya-Fang Wu, Jui-Yung Chung,
Ju-Chun Hsu and Ta-Te Lin, “Online semi-supervised learning applied to an
automated insect pest monitoring system,” Biosystems Engineering, Vol. 208,
pp. 28–44, August 2021.
[14] S. K. Bhoi, K. K. Jena, S. K. Panda, H. V. Long, R. Kumar, P. Subbulakshmi
and H. B. Jebreen, “An internet of things assisted unmanned aerial vehicle
based artificial intelligence model for rice pest detection,” Microprocessors and
Microsystems, Vol. 80, No. 103607, pp. 1–11, February 2021.
[15] C. J. Chen, Y. Y. Huang, Y. S. Li, C.-Y. Chang and Y.-M. Huang, “An AIoT
based smart agricultural system for pests detection,” IEEE Access, Vol. 8,
pp. 180750–180761, 2020.
[16] M. Turkoglu, B. Yanikoğlu and D. Hanbay, “PlantDiseaseNet: convolutional
neural network ensemble for plant disease and pest detection,” Signal, Image
and Video Processing. Received: 27 August 2020 / Revised: 19 February
2021 / Accepted: 5 April 2021 © The Author(s), under exclusive licence to
Springer-Verlag London Ltd., part of Springer Nature 2021, 1–9, https://doi.
org/10.1007/s11760-021-01909-2
[17] L. Liu, R. Wang, C. Xie, P. Yang, F. Wang, S. Sudirman and W. Liu, “PestNet: An
end-to-end deep learning approach for large-scale multi-class pest detection
and classification,” IEEE Access, Vol. 7, pp. 45301–45312, 2019.
[18] W. Li, D. Wang, M. Li, Y. Gao, W. Jianwei and X. Yang, “Field detection of tiny
pests from sticky trap images using deep learning in agricultural greenhouse,”
Computers and Electronics in Agriculture, Vol. 183, No. 106048, pp. 1–11,
April 2021.
[19] Y. Yue, X. Cheng, D. Zhang, W. Yunzhi, Y. Zhao, Y. Chen, G. Fan and Y. Zhang,
“Deep recursive super resolution network with Laplacian Pyramid for better
agricultural pest surveillance and detection,” Computers and Electronics in
Agriculture, Vol. 150, pp. 26–32, July 2018.
[20] F. Wang, R. Wang, C. Xie, J. Zhang, R. Li and L. Liu, “Convolutional neu-
ral network based automatic pest monitoring system using hand-held mobile
image analysis towards non-site-specific wild environment,” Computers and
Electronics in Agriculture, Vol. 187, No. 106268, pp. 1–10, August 2021.
[21] K. Thenmozhi and U. Srinivasulu Reddy, “Crop pest classification based on
deep convolutional neural network and transfer learning,” Computers and
Electronics in Agriculture, Vol. 164, No. 104906, pp. 1–11, September 2019.
Chapter 12

Framework for leveraging diagnostic

and vehicle data with emphasis
on automotive cybersecurity
Thrilochan Sharma
L. & T. Technology Services Limited, Bangalore, India

K. S. Ashwini
Coventry University, United Kingdom

Prithvi Sekhar Pagala and KNS Acharya

L. & T. Technology Services Limited, Bangalore, India

CONTENTS

12.1 Introduction ................................................................................ 181

12.2 Framework ................................................................................. 184
12.3 Applications and features of the framework ................................ 188
12.4 Conclusion .................................................................................. 192
References ............................................................................................. 193

12.1 INTRODUCTION

Transportation and automobiles have become a pivotal component for a

country’s economy and development [1]. Presently, automobile systems are
evolving into multiple forms according to the requirements of the contem-
porary world. In recent years, automotive systems have achieved a level
of maturity, progressing from mechanical to electromechanical systems
and moving ahead with applications such as connected, autonomous, and
electric vehicles. Advances have been observed in applications which have
increased the scope of data collection, networks both in the vehicle and
outside, with processing and usage for advanced vehicle features [2]. This
has increased confidence in the design of automotive systems which have
assisted in the creation of autonomous vehicles. The Society of Automotive
Engineers (SAE) has drawn up a classification of Level-0 to Level-5 for motor
vehicles and their operations on public roads [3]. Within this system, Level-0
is non-autonomous, as can be seen in today’s standard, non-connected vehi-
cles. Level-1 and Level-2 autonomous vehicles provide relief to the driver by

DOI: 10.1201/9781003407300-12 181

182 Thrilochan Sharma et al.

automating vehicle controls such as the brake, the accelerator, and the steer-
ing. The Advanced Driver Assistance System (ADAS) is a standard feature
in today’s vehicles, which performs advanced operations such as cruise con-
trol, automated braking, lane assist, parking assist, and so on [4]. Level-3
and Level-4 autonomy helps a vehicle to run autonomously on standard
highways, and restricted environments such as university complexes, golf
courses, and industrial parks. Finally, there is Level-5 autonomy where the
system takes over all vehicle controls and requires no human intervention
while driving. The Level-5 vehicles that are currently being designed do not
even have steering or pedals.
The primary objective of the On-board Diagnostics (OBD) system when it
was standardized by SAE in 1979 was to improve in-use emission compli-
ance through the monitoring of the computerized emission control system. In
1988, the California Air Resource Board prepared and implemented the suc-
cessor system, OBD-II. The purpose was to monitor the performance of sen-
sors and actuators affecting the engine emission norms and finally, from
1994 onward, it was applied to all vehicles [5]. Engine monitoring, misfire
warning, fuel system analysis, air conditioning system, etc. were the added
features to monitor the various sub-systems of a vehicle. Most of the moni-
toring aspects of the data collected through the OBD port were indicated
using Malfunction Indicator Lamps (MIL). India’s first official policy, General
Statutory Rule-84, was released in 2009, providing guidelines for Bharat
Standard IV (BSIV)-compliant vehicles [6]. From 2010, all the four-wheelers
in India are mandated by the government to have an OBD port; further,
vehicles manufactured from April 2020 onward must be equipped with the
OBD system as their standard feature in line with Bharat Standard VI [7].
As the vehicles became increasingly complex, this resulted in a huge rise
in the data generation and usage by sensors and actuators. This has increased
the scope for research and data inferences, resulting in the introduction of
new applications which make use of the collected data. Vehicle parameters,
such as acceleration, speed, engine rpm, throttle position and engine load,
have served as critical features in the studies for vehicle and, in particular,
the monitoring of engine health. The analysis of these parameters supplies
information regarding improper gear shifting and speeding. Driving at high
speed at lower gear increases the friction loss [8] and speeding increases the
load on the engine, which leads to increased fuel consumption [9]. The dis-
cussed parameters can also be used to analyze the driver’s behavior. With the
increase in the abundant amount of data and the inferences that are made
from it, every vehicle has become a computer. They can be considered as
mobile nodes when compared to a generic IOT network.
Connected vehicles (CVs) is the concept by which the vehicle can com-
municate bidirectionally with other systems outside of the vehicular sys-
tems, including other vehicles and road infrastructure. CVs have already
been adopted in the market to enable remote and advanced features for
users, along with helping in increasing the safety and quality of vehicle
transportation. Research has confirmed that the network of CVs helps to
 Framework for leveraging diagnostic, vehicle data for cybersecurity 183

decrease the number of accidents and the avoidance of road congestion [10].
Vehicles will be equipped with the necessary hardware to achieve Vehicle to
X (V2X) communication, which typically allows a vehicle to communicate
with nodes in its environment, including other vehicles (V2V), the cloud/
network (V2N), the infrastructure (V2I), and pedestrians (V2P) [11]. A
study conducted in 2014 had forecasted that the number of sensors in a
vehicle would reach a total of 400 by 2020 [12]. Vehicle to Sensor (V2S),
V2I, V2N, and V2V systems are studied and wireless efficient solutions are
provided by previous studies [13]. Inter-vehicle co-operation channel esti-
mation (IVC-CE) is one of the methods which will enhance the performance
and support for channel estimation in the case of V2I applications with a
special focus on safety-related applications has been illustrated [14].
Geographically, the North American automotive market currently holds the
largest share of the global connected vehicles, followed by the European
market [15]. With the advancement of such intelligent vehicles, the focus
shifts to sensory networks within the vehicle which can enable efficient
communication.
Through network connectivity, such a vehicular network provides abun-
dant amounts of different types of data from the various sensors of all road
vehicles. The inclusion of advanced sensors in modern vehicles such as depth
cameras, Lidars, Radars, etc., has led to numerous applications using
advanced technologies such as image processing, object detection, feature
extractions, depth calculation, localization, path planning, etc. Artificial
intelligence (AI), machine learning (ML) and deep learning (DL) all play
major roles in future when these sensors become standardized in automotive
systems. The analysis of health conditions, the prediction of failures, behav-
ioral analysis, and the diagnostics of vehicular systems, etc., are the immedi-
ate applications possible through the implementation of AI, ML and DL.
Advancements in automotive features, with increased data flow in a vehi-
cle, also attract the attention of users with negative intentions posing threats
and exploiting the vulnerabilities of the network using cyberattacks. The
training and testing of ML algorithms for simulating and resolving attack
models to achieve secure networks is one of the important steps in the appli-
cation of AI/ML in CV networks [16]. These scenarios provided a necessity
to protect the automotive systems and the data contained in them, giving
rise to the need for cybersecurity. Cybersecurity has become highly promi-
nent due to the continuous increase of vehicle network interconnection and
intellectualization [17].
Cyber-risks of autonomous, connected vehicles have gained a prominent
attention over the past few years as there as various technical vulnerabilities
that have uncovered by security researchers. Cybercriminals can disrupt the
functionalities of vehicular systems such as steering, braking and the engine
by hijacking the Electronic Control Units (ECUs). V2X has opened the pos-
sibility of the vast spectrum of applications, but also has a number of risks
involved in terms of cyberattacks. Involving multiple devices being paired to
the vehicular network, such as a smartphone, network connectivity hardware,
184 Thrilochan Sharma et al.

Table 12.1 O
 verview of major standards in design and development of automotive
systems
Standard First release Primary goal
Unifying safety standard for all automotive electrical/
ISO 26262 June 2009
electronic systems [17]
Guidance on the applicable design, verification and
ISO/PAS
Jan 2019 validation measures needed to achieve the safety of the
21448
intended functionality (SOTIF) [18]
Baseline for vehicle manufacturers and suppliers to ensure
ISO/SAE
May 2020 that cybersecurity risk management is done efficiently
21434
and effectively [19]

etc., provides certain gateways for the cybercriminals. Standards play a major
role in the reduction of such risks and maintain uniformity across platforms
in terms of automotive software. ISO and the SAE are primarily responsible
in the design and maintenance of automotive standards. Components such as
cybersecurity risks in road vehicles, functional safety throughout the develop-
ment cycle of vehicles, etc., are defined as per the standards mentioned in
Table 12.1.
These standards are the documents that are taken as reference while auto-
motive electronic systems are being designed, developed, and tested. These
standards help the automotive industry to achieve safety and security, result-
ing in reliable automotive systems.
The drawback of the automotive industry currently is that the market is
predominantly occupied by non-connected vehicles which are incapable of
achieving features of a connected vehicle and benefiting from the communi-
cation network of connected vehicles. The worldwide automotive market
was comprised of around 8% CVs in 2018, a figure which is projected to
rise to 23% by 2023. The Indian CV market is projected to be US$32.5
­billion by 2026 from a value of $9.8 billion in 2021 [20]. Such a scenario
creates a number of challenges when a traffic environment is comprised of
both non-connected vehicles and connected vehicles. The application of
converting a non-connected vehicle into an IOT node with additional hard-
ware remains a challenge and has received little research attention. A frame-
work for conversion of a non-connected vehicles into an IOT node is needed
in order to achieve both efficiency and safety.

12.2 FRAMEWORK

The growing population density and traffic conditions demand higher vehic-
ular intelligence in the domain of information and communication technolo-
gies. The implementation of the V2X technology through connected vehicles
has the potential to make road transport both safer and more reliable by
 Framework for leveraging diagnostic, vehicle data for cybersecurity 185

establishing communication with several vehicular and infrastructural enti-

ties. In general, the vehicles that provide a data-rich environment that gives
rise to multiple transport-related applications, making them the building
blocks to achieve V2X communication can be termed as “AutoNodes”.
Each AutoNode can function as an independent network device continu-
ously contributing to the framework. Data obtained from such AutoNodes
can be very useful in enhancing multiple data pipelining levels in vehicles, all
of which can contribute to providing an information-rich travel experience.
Since CVs are equipped with various sensors and data transceiver devices
which contribute to the vehicular IOT network, they can be AutoNodes by
default, as shown in Figure 12.1. As discussed earlier, however, the major
challenge with regard to the emerging automobile market of India is that
the number of non-connected vehicles is greater than the CVs. In such cases,
non-connected vehicles can be converted into connected vehicles, making
them capable of interacting with the proposed framework serving as an
AutoNode. Once non-connected vehicles are converted into AutoNodes,
they can benefit from the CV network by performing both data collection as
well as data reception activities.
In general, the non-connected vehicles’ framework requires external sen-
sory equipment to access vehicular ECU data and it can be converted into

Figure 12.1 Connected vehicle as IOT AutoNode.

186 Thrilochan Sharma et al.

Figure 12.2 Non-connected vehicle conversion into an IOT AutoNode.

an AutoNode as represented in Figure 12.2. This utilizes low-cost hardware

resources, principally comprising a data collection module and a connectiv-
ity interface [21]. In this case, the OBD-II module is used for primary vehicle
data collection, whereas a smartphone enables access to the OBD data
through an application in addition to providing inertial measurement unit
(IMU) data and visual data through in-built cameras. Data acquired from
the OBD module can be transferred to smartphones through wireless, USB
or Bluetooth protocols. The simple low-cost implementation facilitates ease
in access to hardware, improving usability across various demographics.
Vehicle parameters, combined with data acquired from smartphones, pro-
vide a wide range of features that can be used for vehicle and driving behav-
ior analysis. The OBD module outputs data in two main forms: real-time
vehicle data and diagnostics data.
Real-time vehicle data include parameters such as average and instanta-
neous speed, RPM, airflow rate, coolant temperature, accelerator, and brake
pedal positions etc. By contrast, diagnostics data provide information about
vehicle parameters at the time of troublesome events, including diagnostic
trouble codes (DTCs). Similarly, several forms of data can be acquired
through the smartphone interface due to its internal IMU comprising of sen-
sors such as accelerometer, gyroscope, and magnetometer. As smartphones
in the modern age are equipped with good resolution cameras, visual data
during driving tasks act as a notable addition (Figure 12.3).
The OBD, sensor and visual data obtained from the vehicle can be pre-
processed locally through on-board computing components and passed on to
the cloud for further computation. As the data set contains several data types,
the processing needs to be performed either sequentially or in parallel using
triggers between data within the algorithm. The data obtained can be broadly
classified into two types: vehicle-sensor data and visual data. ML and DL algo-
rithms play an important role in data processing as the relationship between
 Framework for leveraging diagnostic, vehicle data for cybersecurity 187

Figure 12.3 Flowchart of overall methodology of the framework [22].

these data types needs to be established. Convolutional Neural Network

(CNN)-based DL architectures can be used to process visual data. Similarly,
classification-based ML algorithms can be used for the post-­processing of
vehicle-sensor data [22].
The framework proposed in this chapter suggests crowdsourcing vehicu-
lar and visual data. However, supervised algorithms require massive amounts
of data to achieve efficient results. In such cases, methods of transfer learn-
ing and reinforcement learning are implemented. Pre-trained CNN-based
DL algorithms, such as YOLO (You Only Look Once) and VGG (Visual
Geometry Group), are popular choices where image data is concerned
[23, 24]. The major computation of visual and vehicle data takes place in
the cloud, which is also responsible for the crowdsourcing and data redistri-
bution to various entities establishing V2X communication.
The inclusion of external sensors and computational equipment in non-
connected vehicles evolves the architecture to a connected vehicle architec-
ture that is capable of exchanging data with the cloud. The cloud, in turn,
is capable of sharing data with viable entities, including infrastructure and
other vehicles. This implies that the difference between the architectures in
terms of physical implementation lies only in the hardware setup. Similarly,
the cloud plays a critical role in the algorithm component of the non-
connected AutoNodes architecture where vehicles would not be able to
188 Thrilochan Sharma et al.

Figure 12.4 AutoNode network of standard and connected vehicle configurations.

communicate directly with each other in the absence of cloud communica-

tion, as shown in Figure 12.4. For in-vehicle communication networks,
ECUs are the major entry points for cybercriminals whereas, in applica-
tions of V2X, connectivity devices such as transceivers and network rout-
ers are the major vulnerabilities. In this case, vehicular cybersecurity and
standards gain a considerable amount of importance as data communica-
tion between vehicles and the cloud need to be protected.

12.3 APPLICATIONS AND FEATURES OF THE

FRAMEWORK

AutoNodes that are implemented through CVs or non-connected vehicles

provide an initiative-taking solution for several supporting applications,
including road safety, traffic and fleet management, ADAS and vehicle main-
tenance. The exponential increase of urbanization demands road safety
and traffic congestion solutions as they have serious consequences toward
the safety of humans as well as increased infrastructure costs. On similar
grounds, vehicle diagnostics and prognosis helps in understanding vehicle
health and aid in environmental wellbeing. The framework proposed in this
chapter benefits multiple stakeholders as the data is predominantly crowd-
sourced. Among the stakeholders are the following:

• Vehicle manufacturers: Major automotive companies in the industry

manufacture vehicles based on geographical requirements. Through
AutoNodes, a comparison of performance of vehicles under different
conditions can provide insightful data in understanding the require-
ment of vehicular features in a geographical location.
• Government: The government can improve infrastructure management
and regular maintenance through the data acquired from AutoNodes.
 Framework for leveraging diagnostic, vehicle data for cybersecurity 189

• Incident analysis: Dangerous situation and accident analysis can be

performed through data from AutoNodes. Traffic-related updates and
alerts can also form an important component of incident analysis.
• Traffic violations: The traffic department largely benefits from the
implementation of AutoNodes as violations can be recorded and
reported in real time supported by data, that would in turn automate
billing and fining systems.
• Vehicle rental and insurance agencies: Driving scores can be formu-
lated by using the driving behavior data obtained from AutoNodes.
• Occupant safety and allied services: In-vehicle visual data can be used
to analyse occupant activity and a response to various driving styles.
This data can be used to enhance safety features for infants and the
elderly in vehicular environments.

Connected vehicles which are equipped with required sensors might help
achieve the above-discussed applications to a high level. But non-connected
vehicles, which are the major occupants of the market, cannot benefit from
the CV network. Using the current framework, our previous research was
performed on four different non-connected vehicles with different drivers
and locations. The inferences that are made from the framework [21, 22]
have been classified into five main categories, as shown in Figure 12.3:

• Anomaly detection
• Driving behavior analysis
• Driving situation analysis
• Driver assistance features
• Environment logging

The feature of anomaly detection uses the data from IMU and the cam-
era sensors of the smartphone for the detection of potholes, as shown in
Figure 12.5(a), whereas the OBD interface data of the car are used for the
analysis of sensor data of the vehicle to extract the speed of the car, the
accelerator pedal position, etc. Accelerometer data perception of pothole
detection in the current approach can be observed in Figure 12.5(a). The
data undergoes filtering and thresholding methods to render negative peaks,
where the vehicle encounters the pothole. A trigger-based system is devel-
oped based on data-driven and image-driven results that are used for vali-
dation. After validation, the location of potholes has been extracted in the
form of longitude and latitude and plotted on a map for representation, as
marked in Figure 12.6. Real-time vehicular data obtained from the OBD
module, such as vehicle speed and accelerator pedal position, shown in
Figure 12.5(b) and (c), are utilized to understand ride quality.
The driving behavior feature has a number of sub-features that can be
further improved by considering precautionary suggestions based on vehicle
parameters on regular routes. Parameters such as catalytic converter, overall
fuel consumption and vehicle speed have been used for this purpose.
190 Thrilochan Sharma et al.

Figure 12.5 (a) Pothole validation through accelerometer data (b) Accelerator
pedal position and speed through potholes (c) Values of speed of
vehicle with respect to pothole presence on the road [21, 22].

The performance of the catalytic converter depends on the mean O2 sensor

values from the vehicle obtained from the OBD module. On similar grounds,
average fuel consumption during the drive is established using the “L/100km”
method. One important sub-feature of driving behavior analysis is the rec-
ognition of over-speeding instances during the drive. The parameters used to
detect these instances are torque, vehicle speed and engine RPM. The param-
eters compass bearing, along with vehicle speed values, help in detecting
sharp turns which forms another sub-feature (Figure 12.6).
Driving situation analysis in this context focuses on traffic detection and
analysis based on OBD and visual data. The data-driven approach indicates
a reduction in speed whereas the image data processed using the YOLO-V3
architecture is used to detect classes of objects on the road. The pre-trained
model detects more than 80 classes of objects, including stop signs, cars,
motorcycles, pedestrians, buses, and animals. The feature extractor in the
YOLO-V3 architecture is DarkNet-53, which is 1.5 times faster when com-
pared to previous versions. Situational analysis is performed based on these
parameters, as shown in Figure 12.7(b).
Framework for leveraging diagnostic, vehicle data for cybersecurity 191
Figure 12.6 Geotagged locations of potholes detected [21].
192 Thrilochan Sharma et al.

Figure 12.7 ( a) Pothole detected through smartphone camera sensor (b) Result
of situational driving analysis [21, 22].

12.4 CONCLUSION

The chapter talks about the overview of automotive electronic systems and
the evolution of OBD to establish the scale of data generated by automo-
biles. The complexity of systems has increased multifold and the possibilities
of applications at scale using the huge amount of data provided by vehicles
has been established. Advanced features such as ADAS, diagnostics, appli-
cations using image processing, etc., are discussed in detail. The proposed
framework conclusively, with the help of the results presented, establishes
 Framework for leveraging diagnostic, vehicle data for cybersecurity 193

that non-connected vehicles can have multiple advanced features of con-

nected vehicles and enable any AutoNode to have ADAS along with the
aspects of standards and security. The discussed methodology emphasizes
on the applications of taking previous research by authors on road anomaly
detection, behavioral analysis etc. Further extension to the current meth-
odology with the implementation of AI, ML and DL algorithms trained
with the help of collected data, can bring in a lot of additional inferences.
AutoNodes, along with V2X infrastructure established on the large scale,
will have a very substantial impact on the advanced features of an automo-
bile in achieving safe, eco-friendly, and efficient drive cycles.

REFERENCES

[1] R. Menaka and K. Ashath, “A study on role of automobile industry in India

and its customers satisfaction,” Shanlax International Journal of Management,
vol. 2, no. 4, pp. 2321–46433, 2015.
[2] A. Oktav, “New trends and recent developments in automotive engineering
chapter 331 new trends and recent develelopments in automotive engineering
Akın OKTAV.” [Online]. Available: https://www.researchgate.net/publication/
321621798
[3] “J3016_202104: Taxonomy and Definitions for Terms Related to Driving
Automation Systems for On-Road Motor Vehicles – SAE International.” https://
www.sae.org/standards/content/j3016_202104/ (accessed Jul. 18, 2022).
[4] A. Ziebinski, R. Cupek, D. Grzechca, and L. Chruszczyk, “Review of advanced
driver assistance systems (ADAS),” AIP Conference Proceedings, vol. 1906,
2017, doi: 10.1063/1.5012394.
[5] B. J. Shinde, S. S. Kore, and S. S. Thipse, “Comparative study of on board diag-
nostics systems-EOBD, OBD-I, OBD-II, IOBD-I and IOBD-II,” International
Research Journal of Engineering and Technology, 2016, Accessed: Jul. 18,
2022. [Online]. Available: www.irjet.net
[6] “Ministry of Shipping, Road Transport and Highways G.S.R 84(E),” Feb. 9, 2009.
https://morth.nic.in/sites/default/files/notifications_document/Notification_
No__S_O_84E_dated_09-02-2009_Part-1_0.pdf (accessed Jul. 18, 2022).
[7] “India Bharat Stage VI Emission Standards,” ICCT POLICY UPDATES,
Apr. 2016. https://theicct.org/sites/default/files/publications/India%20BS%20
VI%20Policy%20Update%20vF.pdf (accessed Jul. 18, 2022).
[8] Y. Ko and K. Hosoi, “Measurements of power losses in automobile drive
train,” SAE Technical Papers, 1984, doi: 10.4271/840054.
[9] C. M. C. G. Fernandes, P. M. T. Marques, R. C. Martins, and J. H. O. Seabra,
“Gearbox power loss. Part II: Friction losses in gears,” Tribology International,
vol. 88, pp. 309–316, Apr. 2015, doi: 10.1016/J.TRIBOINT.2014.12.004.
[10] D. Singh and M. Singh, “Internet of vehicles for smart and safe driving,” 2015
International Conference on Connected Vehicles and Expo, ICCVE 2015 –
Proceedings, pp. 328–329, 2016, doi: 10.1109/ICCVE.2015.93.
[11] K. Jadaan, S. Zeater, and Y. Abukhalil, “Connected vehicles: An innovative
transport technology,” Procedia Engineering, vol. 187, pp. 641–648, 2017, doi:
10.1016/J.PROENG.2017.04.425.
194 Thrilochan Sharma et al.

[12] M. Pinelis, “Automotive sensors and electronics: trends and developments in

2013,” Sensors Electronics Expo, vol. 2013, 2013.
[13] N. Lu, N. Cheng, N. Zhang, X. Shen, and J. W. Mark, “Connected vehicles:
Solutions and challenges,” IEEE Internet of Things Journal, vol. 1, no. 4, pp.
289–299, Aug. 2014, doi: 10.1109/JIOT.2014.2327587.
[14] Y. Yang, D. Fei, and S. Dang, “Inter-vehicle cooperation channel estimation
for IEEE 802.11p V2I communications,” Journal of Communications and
Networks, vol. 19, no. 3, pp. 227–238, 2017, doi: 10.1109/JCN.2017.000040.
[15] P. Sharma, D. Austin, and H. Liu, “Attacks on Machine Learning: Adversarial
Examples in Connected and Autonomous Vehicles,” 2019 IEEE International
Symposium on Technologies for Homeland Security, HST 2019, 2019, doi:
10.1109/HST47167.2019.9032989.
[16] M. Raya and J. P. Hubaux, “The Security of Vehicular ad Hoc Networks,”
SASN’05 - Proceedings of the 2005 ACM Workshop on Security of Ad Hoc and
Sensor Networks, vol. 2005, pp. 11–21, 2005, doi: 10.1145/1102219.1102223.
[17] P. Gliwa, “Safety and ISO 26262,” Embedded Software Timing, pp. 289–295,
2021, doi: 10.1007/978-3-030-64144-3_11.
[18] “ISO - ISO/PAS 21448:2019 – Road vehicles – Safety of the intended function-
ality.” https://www.iso.org/standard/70939.html (accessed Jul. 18, 2022).
[19] M. Tschersich, “How to prepare automotive for future challenges iso 21434 –
a standard for cybersecurity engineering,” VDI Berichte, vol. 2017, no. 2310,
pp. 29–30, 2017, doi: 10.51202/9783181023105-29.
[20] “India Connected Car Market Size, Share, Forecast Report – 2025.” https://www.
marketsandmarkets.com/Market-Reports/india-connected-car-market-15076
5405.html (accessed Jul. 18, 2022).
[21] K. S. Ashwini, G. Bhagwat, T. Sharma, and P. S. Pagala, “Trigger-Based Pothole
Detection Using Smartphone and OBD-II,” Proceedings of CONECCT 2020 - 6th
IEEE International Conference on Electronics, Computing and Communication
Technologies, 2020, doi: 10.1109/CONECCT50063.2020.9198602.
[22] S. Sabde, P. S. Pagala, K. Mudaliar, and K. S. Ashwini, “Scalable Machine
Learning and Analytics of the Vehicle Data to derive Vehicle Health and Driving
Characteristics,” Proceedings of IEEE International Conference on Disruptive
Technologies for Multi-Disciplinary Research and Applications, CENTCON
2021, pp. 146–151, 2021, doi: 10.1109/CENTCON52345.2021.9687913.
[23] W. Lan, J. Dang, Y. Wang, and S. Wang, “Pedestrian Detection Based on YOLO
Network Model,” in IEEE International Conference on Mechatronics and
Automation, 2018.
[24] S. Arjapure and D. R. Kalbande, “Deep Learning Model for Pothole Detection
and Area Computation,” in Proceedings – International Conference on
Communication, Information and Computing Technology, ICCICT 2021,
2021. doi: 10.1109/ICCICT50803.2021.9510073.
Chapter 13

An array of Fibonacci series-based

wide-band printed antennas
for IoT/5G applications
Kalyan Sundar Kola and Anirban Chatterjee
National Institute of Technology, Ponda, Goa

CONTENTS

13.1 Introduction ................................................................................ 195

13.2 Proposed single antenna .............................................................. 197
13.2.1 Formation of the proposed geometry .............................. 197
13.2.2 Depiction of single antenna ............................................ 199
13.2.3 Contour length and surface area computation ................ 200
13.2.4 Time domain analysis ..................................................... 202
13.3 Proposed array ............................................................................ 202
13.3.1 Construction of power divider network .......................... 202
13.3.2 Development of array ..................................................... 204
13.3.3 Benefits of the Wilkinson power divider ......................... 204
13.4 Results and discussion ................................................................. 206
13.5 Conclusion .................................................................................. 210
Acknowledgement �������������������������������������������������������������������������������� 210
References �������������������������������������������������������������������������������������������� 210

13.1 INTRODUCTION

As internet connection spreads among many devices, the Internet of Things

(IoT) has become an increasingly popular subject for discussion. Ericsson
claims that by 2025, more than 5 billion gadgets will simultaneously be
linked to the internet. Over the same period, 5G is becoming increasingly
popular as a replacement for LTE (4G), which suffers from low data rates,
limited bandwidth, and poor quality of service (QoS) in metropolitan areas.
5G, by contrast, delivers fast data rates, substantial capacity, and excellent
spectrum efficiency. Outdoor, urban, and interior systems will all be needed
for the new 5G technology. Accordingly, IoT and 5G technologies are in
high demand because of these benefits, whose architecture is depicted in
Figure 13.1(a) and (b), respectively. For 5G technology, low-­cost and light-
weight printed antennas [1, 2] are in high demand in this communication age
for the connection of IoT devices. For this reason, a wide-­band microstrip

DOI: 10.1201/9781003407300-13 195

196 Kalyan Sundar Kola and Anirban Chatterjee

Figure 13.1 A
 rchitectural representation of (a) IoT-based applications (b) 5G
communications.

patch antenna is an ideal choice for such uses. The printed antenna and its
array have been studied in a number of publications.
In literature [3], printed antennas were studied by Carver et al. in both
theoretical and practical terms. Furthermore, Werner et al. [4] study the
progress of using fractal geometry in antenna design. The authors of that
study developed antenna arrays using fractal geometries [4]. In another
study, Al-­Sehemi et al. [5] have suggested and fully analyzed a broadband
waterproof antenna based on IoT applications and Ashyap et al. [6] devel-
oped a fabric material laminated C-­shaped printed antenna for medical IoT
applications. Cowsigan et al. [7] have presented a SIW cavity-­backed printed
antenna for IoT applications and an ESP8266 antenna module for IoT
applications is provided by Roges et al. [8]. As described by Sharma et al.[9],
future wireless applications may well benefit from a six-­element MIMO
antenna. In the IoT context, Mushtaq et al. [10] proposed an array of
T-­shaped slotted printed antennas. According to Wang et al. [11], a circu-
larly polarized printed meta-­material antenna for 5G indoor applications
has been developed. An IoT application-­oriented printed antenna that is
5G-­enabled has been built and thoroughly tested by De et al. [12] And Singh
et al. [13] introduced a dual-­band microstrip patch antenna for 5G applica-
tions based on a SIW antenna structure. In addition, many studies have sug-
gested using an antenna array to enhance parametric outcomes. Wang et al.
[14] introduced a planar array of U-­slotted patch antennas. Clover-­leaf-­
shaped geometries inspired by nature have been published by Kola et al.
[15]. The author has also developed and analyzed a variety of microstrip
patch antenna arrays, such as Christmas-­ tree, hybrid fractal, and tulip
flower geometries [16–19]. Drawing on this background, this chapter
 An array of Fibonacci series-based WB antennas for IoT/5G applications 197

reports an IoT/5G applications-­

based 2-­
element linear printed array of
monopole antennas.
Among the new features of the proposed antennas are the following:

• The sole antenna is derived from the Fibonacci series of successive

seventh terms.
• The sole antenna is capable of offering wide impedance bandwidth of
2.3 GHz.
• It also offers below –35dB of x-­pol. suppression along the main lobe
of direction.
• It also offers quite good peak-­gain at both of the resonating frequencies.
• The array’s feed network has been designed using the Wilkinson power
splitter [20] to achieve low loss, high isolation, and improved band-
width responsiveness.
• In addition, the array has a broad impedance bandwidth, a high gain,
and a shallow x-­pol. level at the ideal radiation point.

In addition to the advantages mentioned above, the single antenna and the
array have aperture [24, 25] and radiation efficiencies [26–30] of over 63%
and 92%, respectively. At appropriate resonating frequencies, both anten-
nas give more than 30 dB/m correction factors [31] and so have modest
electromagnetic interference effects. As a result of their specifications, both
antennas are appropriate for use in IoT and 5G networks.

13.2 PROPOSED SINGLE ANTENNA

Seven terms of the Fibonacci sequence were considered in the development

of the single antenna. Detailed geometry creation, antenna construction,
perimeter and area calculations, and time-­domain analysis have all been
completed here.

13.2.1 Formation of the proposed geometry

The step-­wise formation of the proposed ‘Fibonacci series’-based geometry
is described in Figure 13.2. There are seven successive Fibonacci squares
in the one quadrant construction, and the side length is calculated [21] as
follows.

kS 1    kS , 1  s  6 (13.1)

where, φ is the ‘golden ratio’ (≈1.67) and k1 = 1 mm.

Initially, a square geometry A is considered whose side length is kept as k1,
which is shown in Figure 13.2(a). In step-­1, the square A is scaled with φ
 198
Kalyan Sundar Kola and Anirban Chatterjee
Figure 13.2 Step-wise formation of proposed Fibonacci series-based geometry.
 An array of Fibonacci series-based WB antennas for IoT/5G applications 199

and this obtains the scaled geometry B, whose side length becomes k2 which
is presented in Figure 13.2(b). Further, the geometry B is scaled by φ and
obtains the resultant geometry C after the completion of step-­2, whose side
length is k3 as shown in Figure 13.2(c). Similarly, the geometries D, E, F and
G have been achieved following the completion of steps-­3,4,5,6, and those
are represented in Figure 13.2(d)–(g), respectively. The side length of the
resultant geometries, achieved from steps-­3,4,5,6 becomes k4, k5, k6, k7,
respectively. In step-­7, all the resultant geometries obtained from the preced-
ing steps are Boolean and they have achieved the Fibonacci square-­based
geometry H whose overall side length becomes k7. This is shown in Figure
13.2(h). Further, a structure I is formed by the combination of four arrow
shaped geometries where the arrows make an angle of θ1, θ2, θ3, θ4 with
respect to the generating arrow, as shown in Figure 13.2(i). The length and
width of the thinned arrow geometry are fixed as rL and rW, respectively, as
indicated in Figure 13.2(i). In order for the structure to resonate at the
desired frequency, the geometry I was cut out of each corner of H. After
step-­8, we got the one-­quadrant Fibonacci square-­based slotted structure J,
which is shown in Figure 13.2(j). To produce a symmetric radiation pattern,
the four values of J are positioned in a four-­quadrant of the xy-­plane and
then overlaid them to form the 2D-­symmetric structure K, as shown in
Figure 13.2(k). To make sure that all of the parts of the geometry K are con-
nected, a circle and four semicircles are Boolean with the K. This is how the
final proposed geometry M, shown in Figure 13.2(l), was made. The numeric
values of the used notations are enlisted in Table 13.1.

13.2.2 Depiction of single antenna

To build the printed antenna, as shown in Figure 13.3, the final recom-
mended geometries of the step-­wise construction are laminated on the top
layer of Rogger 5880 PCB material. Figure 13.3(a) depicts the antenna’s
frontal perspective. A microstrip line feed has been used to excite the
antenna construction, and the optimal length of the feed is Lf, as indicated
in Figure 13.3(a). A partial ground plane is used underneath the substrate
layer of the antenna in order to create a broad impedance bandwidth. The
partial ground plane has also been etched with a rectangular slot (SL × SW),
which runs parallel to the feed line and ensures correct impedance matching

Table 13.1 Dimension details of the single antenna (unit: mm)

Exponent Esteem Exponent Esteem Exponent Esteem Exponent Esteem
k1 1 K2 2 K3 3 K4 5
K5 8 K6 13 K7 21 rL 11
rW 0.5 r 1 W 42 L 42
SL 5 SW 2.29 Lf 20.9 Lg 19.65
200 Kalyan Sundar Kola and Anirban Chatterjee

Figure 13.3 Proposed antenna (a) front-view (b) side-view.

at the required frequency range. Figure 13.3(a) shows that the radius of
the connecting circle is set at r. The vertical view of the antenna is reflected
in Figure 13.3(b). It is evident from the vertical perspective of the antenna
that the suggested antenna is a single-­layered construction. The proposed
structure is simulated in the SONNET EM simulator [22], and, further, the
results are cross-­verified by CST microwave studio [23]. Table 13.1 lists all
of the dimensions’ numerical values.

13.2.3 Contour length and surface area computation

The effective perimeter and the surface area of the suggested structure can
be calculated as follows:
Contour length:
The contour length of the superimposed geometry H is k7 mm. Hence, the
perimeter of H is computed as:

PH  4  k7 mm (13.2)
 An array of Fibonacci series-based WB antennas for IoT/5G applications 201

In order to increase the electrical length of J, four slotted based geometry (I)
has been etched from geometry H where the contour length of I is computed
as follows:

PI  4  2   rL  rW  mm (13.3)

quadrant geometry J
Therefore, the effective contour length of the one-­
becomes:

PJ  PH  PI  4  k7  2  rL  rW  mm (13.4)

Similarly, the contour length of the proposed four-­quadrant geometry K is

computed as follows:

PK  4  PJ mm (13.5)

Also, the geometry K adds one circle and four semicircles, which make the
structure shorter from an electrical point of view. The following equation
shows how to figure out the effective length of the final geometry M’s contours:

PM  PK   6    r  mm (13.6)

After putting all the parameter values in equations (13.2)–(13.6) and the
computed effective contour length of the final proposed geometry becomes
685.15 mm.

Surface area:
The radiating surface area of the suggested superimposed structure H is
derived as follows:

AH = K7 2 mm2 (13.7)

After subtracting the structure I, the overall area of the one-­quadrant struc-
ture J is reduced and it becomes

AJ  K7 2   4  rL  rW  mm2 (13.8)

Now the surface area of the four-­folded geometry K is as follows:

AK  4  AJ 2 mm2 (13.9)

Further, one circle and four numbers of semi-­circles are added with the
geometry K, which improve the surface area of the structure. Therefore, the
overall surface area of the final geometry M is computed as follows:

 
AM  AK  3    r 2 mm2 (13.10)
202 Kalyan Sundar Kola and Anirban Chatterjee

After putting all the parameter values in equations (13.7)–(13.10) and the
computed surface area or the radiating of the final proposed geometry
becomes 16.85 cm2.

13.2.4 Time domain analysis

Time domain characteristics, such as input–output signal quality, scattering
parameters, and group delay, are examined here. The identical antennas are
placed in two different ways on the platform of CST microwave studio, as
illustrated in Figures. 13.4(a) and (b), one horizontally and the other verti-
cally. The inter-­element spacing has been maintained at 0.3 meters for both
arrangements. Figure 13.4(c) shows the desired output signal values when
a sinusoidal signal is supplied to the antenna’s port. Antenna isolation is
shown in Figure 13.4(d) for two different configurations. As can be seen in
Figure 13.4(d), excellent isolation of less than −22 dB was obtained from
both arrangements throughout the entire input frequency range. As seen in
Figure 13.4(e), the phase response of the antennas shows linear phase fluc-
tuation, indicating a well-­designed system. Figure 13.4(f) depicts the group
delay characteristics visually. Group delay is below 2ns in both situations,
which is desirable and widely acceptable for wideband applications, par-
ticularly for IoT and 5G communications.

13.3 PROPOSED ARRAY

This section explains the proposed power divider, the building of the array,
and the benefits of the Wilkinson power divider network.

13.3.1 Construction of power divider network

A three-­port power divider network [20] is presented in Figure 13.5.
Basically, it’s a Wilkinson power divider network where the source power is
forwarded to the sink ports by maintaining an equal magnitude. It follows
the transmission line theory principle. The power is driven from the source
port using a transmission line whose characteristics impedance is Z0 and
transferred to the quarter-­wavelength junction whose characteristics imped-
ance is 2 Z0. Further, the power is equally delivered to the output ports
through the Z0 transmission lines. For isolation, the 2Z0 resistor is fixed in
the quarter-­wavelength junction, as shown in Figure 13.5.
The transmission line’s [1] input impedance (Zin) can be determined as
follows:

ZL  jZ0 tan  l
Zin  Z0 (13.11)
Z0  jZL tan  l
 An array of Fibonacci series-based WB antennas for IoT/5G applications
Figure 13.4 T
 ime domain analysis of the single antenna: (a) side-by-side (b) face-to-face (c) input–output signal (d) impedance (e) S 21
magnitude (f) S 21 phase (g) group delay.

203
204 Kalyan Sundar Kola and Anirban Chatterjee

Figure 13.5 A 3-way power divider network.

where ZL and Z0 are indicated as load and characteristics impedance of the

antenna and l is considered as the length of the transmission line, respectively.

13.3.2 Development of array
A two-­element array, shown in Figure 13.6, has been suggested to enhance
parametric results. It’s a linear array with a 0.5 (R) separation between
the elements. A shared partial ground plane is implemented to accomplish
the broad impedance bandwidth necessary for IoT and 5G applications.
As part of the array’s feed network, the Wilkinson power divider is used.
50Ω–70Ω–50Ω transmission lines make up the network, which sends out
an equal amount of power to all sink ports of the antennas. The 50Ω trans-
mission line’s length and width are set as λ/2 and 2.29 mm, respectively.
The 70Ω transmission line’s length and width are indeed an integer mul-
tiple of λ/4 and 1.31 mm, respectively. As seen in Figure 13.6, a quarter-­
wavelength transmission line’s comprehensive description is reflected.
Data about the transmission lines’ dimension notations are presented in
Table 13.2.

13.3.3 Benefits of the Wilkinson power divider

Two types of power divider networks may be utilized to build the array
geometry: the conventional and the Wilkinson power divider networks.
Using the Wilkinson power divider network, we may reduce the overall size
of the array while still achieving the goals mentioned above of low loss [1]
and high isolation [1]. Stable radiation patterns may also be achieved by
 An array of Fibonacci series-based WB antennas for IoT/5G applications 205

Figure 13.6 Proposed 2-element linear array geometry.

Table 13.2 Dimension details of the proposed array (unit: mm)

Exponent Esteem Exponent Esteem Exponent Esteem Exponent Esteem

WA 150.70 LA 98.30 R1 2 R2 18.71

R3 2.02 R4 1 R5 1 R6 8.60
R7 2 R8 21.93 S1 9.94 S2 2
S3 10.44 S 1 R 25.74 -- --

using this method. Figure 13.7 shows the results of a performance com-
parison between arrays using standard and Wilkinson feeds [20]. Figures
13.7(a) and (b) show the return-­loss characteristics of a standard feed and
its related arrays. Furthermore, the Wilkinson power divider network and its
related arrays’ return loss are shown in Figures 13.7(c) and (d). A Wilkinson
power divider network may be plainly shown to provide a superior return
loss and impedance bandwidth than an array that uses the standard feed.
Table 13.3 displays the parametric results side by side in a tabular manner.
Table 13.3 clearly shows that the Wilkinson power divider-­based suggested
array has superior parameters with regard to gain, radiation efficiency, and
other metrics.
206 Kalyan Sundar Kola and Anirban Chatterjee

Figure 13.7 C
 omparative performance of array geometries (a) conventional feed,
(b) resultant S 11 of (a), (c) Wilkinson power feed (d) resultant S 11 of (c).

Table 13.3 Comparative performance of the arrays

Gain
Feed-­type Size [mm2] fr [GHz] S11 [dB] BW [GHz] [dB] η [%]
Conventional 150.70 × 2.46,3.63 −19.65, 0.6, 0.14 4.22, 88, 87
98.30 −13.45 4.64
Wilkinson 150.70 × 2.46,3.63 −24.82, 1.82 7.17, 97, 93
102.30 −16.70 7.82

13.4 RESULTS AND DISCUSSION

The proposed printed antenna and its array, based on the Fibonacci series
and geared toward IoT/5G applications, have been modeled with the help of
the SONNET EM simulator [22] and cross-­verified by the CST microwave
studio [23].
The partial ground plane-­based single antenna has resonated in two dif-
ferent frequencies, i.e., 2.40 and 3.60 GHz. It offers a satisfactory return
loss of 32.56 and 19.12 dB, respectively, as depicted in Figure 13.8(a).
 An array of Fibonacci series-based WB antennas for IoT/5G applications 207

Figure 13.8 S imulated results of single antenna: (a) S 11 , (b) gain (c) E- and H-field
patterns for 1st resonance (d) E- and H-field patterns for 2nd reso-
nance (e) 3D-radiation pattern of 1st resonance (f) (e) 3D-radiation
pattern of 2nd resonance.

It also covers a wide-­impedance bandwidth of 2.30GHz (1.55–3.85 GHz).

The single antenna can satisfy the needs of IoT-­based household appliances
and 5G communications based on resonating frequencies. The sole antenna
gives desired peak gain of 4.26 and 5.92 dBi at two resonances, as shown in
Figure 13.8(b). The E-­and H-­field properties of the antenna for two reso-
nating frequencies have been characterized in Figure 13.8(c) and (d), respec-
tively. It can be clearly observed that the sole antenna gives <–35 dBi
cross-­pol. discrimination in both the frequencies, which indicates the excel-
lent design. 3D radiation patterns of the antenna are presented in Figure
13.8(e) and (f), respectively, in which the right-­hand scale indicates the max-
imum magnitude achieved from the antenna. Additionally, it was discovered
208 Kalyan Sundar Kola and Anirban Chatterjee

Table 13.4 Parametric performances of the antenna and the array

Single antenna Antenna array
Parameter(s) 1st Resonance 2nd Resonance 1st Resonance 2nd Resonance
fr [GHz] 2.40 3.60 2.46 3.63
S11 [dB] −32.56 −19.12 −24.82 −16.70
BW [GHz] 2.30 1.82
Gain [dBi] 4.26 5.92 7.17 7.82
Directivity [dBi] 5.63 7.15 8.26 9.14
x-­pol. [dBi] −38 −39 −29 −32
Rad. eff. [%] 97 96 97 93
App. eff. [%] 74.90 64.77 75.27 63.68
CF [dB/m] 33.57 35.43 30.87 33.60

that the antenna produces a radiation pattern, which is symmetric in the

required directions. Table 13.4 has an entry for every numeric value of the
parameter. The single antenna gives 97% and 96% radiation efficiency for
2.40 and 3.60 GHz, respectively. It can be observed that the antenna is
highly efficient for IoT and 5G applications.
An antenna’s aperture efficiency [24, 25] is computed as follows:

D
ap  (13.12)
Dmax

4    Area
Dmax  (13.13)
2

where D, Area are indicated as the directivity and overall surface area of the
antenna, and λ is the free-­space wavelength.
The overall surface area of the sole antenna is 83.8 × 83.8 mm2 and the
directivity is 5.63 and 7.15 dBi, respectively. As a result, the estimated aper-
ture effectiveness of the antenna becomes 97% for a frequency of 2.40 GHz
and 96% for a frequency of 3.60 GHz.
One of the most essential aspects of an antenna’s electromagnetic impact
is its electromagnetic compatibility (EMC). Using the surface equivalence
technique, the antenna’s correction factor (CF) [31] is determined, and the
CF is then calculated as follows:

EINC  9.73 
CF   20 log   (13.14)
Vr   Ga 

where, |EINC|→ incident field strength, |Vr|→ antenna terminal voltage, and
Ga→ antenna’s gain. The single antenna gives the correction factor of 33.57
 An array of Fibonacci series-based WB antennas for IoT/5G applications 209

and 35.43 dB/m at two resonating frequencies, and these are widely accept-
able for the desired applications.
With the help of the Fibonacci series-­based single antenna, a two-­element
array has been proposed and developed to improve parametric outcomes.
Figure 13.9(a) shows that the suggested array has a return loss of 24.82 dB
and 16.70 dB, respectively, at 2.46 and 3.63 GHz. The array also has a large
impedance bandwidth of 1.82 GHz, covering the entire IoT and 5G range.
It can be seen in Figure 13.9(b) that the antenna offers 7.17 and 7.82 dBi
gains, respectively, which are broadly accepted for the intended applica-
tions. In Figure 13.9(c) and (d), the E-­and H-­fields of the array are described.
One noticeable feature of the array is that the x-­pol. suppression is relatively

Figure 13.9 S imulated results of array: (a) S 11 , (b) gain (c) E- and H-field patterns
for 1st resonance (d) E- and H-field patterns for 2nd resonance (e)
3D-radiation pattern of 1st resonance (f) (e) 3D-radiation pattern
of 2nd resonance.
210 Kalyan Sundar Kola and Anirban Chatterjee

shallow, with a value below −29 dBi that is desired and commonly accepted.
Figure 13.9(e) and (f) show that the array provides a steady radiation pat-
tern at the correct frequencies. The antenna array’s simulated outcomes are
enlisted in Table 13.4. The radiation efficiency of the array is shown to be
97% and 93%, respectively. The array’s aperture is 150.7 × 98.3 mm2 and
its directivity is 8.26 and 9.14 dBi. For the two different frequencies of
operation, the array’s aperture efficiency was determined as 75.27% and
63.68%, respectively. The array’s computed CF values are 30.87 and 33.6
dB/m, which are desired. Both antennas can support IoT-­based home appli-
cations and 5G communications with acceptable performance parameters.

13.5 CONCLUSION

An array of a couple of printed radiators has been designed and its paramet-
ric behavior has been thoroughly analyzed. The single monopole radiator of
the array is derived from the Fibonacci series based on seven square geome-
tries. A four-­fold architecture has been studied in order to generate symmet-
ric radiation patterns at the necessary frequency. The resulting fourfolded
shape was used to etch four numbers of thinned slotted structures, which
provided the necessary resonant properties. The construction incorporates
a partial ground plane with a rectangular slot in order to produce a broad
impedance bandwidth. It has been analysed in the time domain to determine
its signal quality, isolation, group delay, and other characteristics. The single
antenna has a broad impedance bandwidth, strong gain, a shallow cross-­pol.
level, and great radiation efficiency. The two-­element linear array was cre-
ated to enhance parametric results. The Wilkinson power divider network is
also used to construct its feed and achieve low loss, a wide impedance band-
width, and intense isolation. The array’s parametric findings are likewise
promising. There is considerable potential for IoT-­based home applications
and 5G communications using these antennas.

ACKNOWLEDGEMENT

This chapter is an outcome of a project (File No. SB/S3/EECE/226/2016)

under the SERB Extra Mural Research Funding, DST, GoI.

REFERENCES

[1] C.A. Balanis, Antenna Theory: Analysis and Design, Wiley, 1997.
[2] R.L. Haupt, Antenna Arrays: A Computational Approach, Wiley-­IEEE Press,
2010.
 An array of Fibonacci series-based WB antennas for IoT/5G applications 211

[3] K.R. Carver, and J.W. Mink, “Microstrip Antenna Technology,” IEEE
Transactions on Antennas and Propagation, vol. AP-­29(1), pp. 2–24, 1981.
[4] D.H. Werner and S. Ganguly, “An Overview of Fractal Antenna Engineering
Research,” IEEE Antennas and Propagation Magazine, vol. 45(1), pp. 38–57,
2003.
[5] G.A. Abdullah, A.A. Ahmed, T. D. Nikolay, T.A. Nikolay, and L. A. Gabriela,
“Design of a Flexible Waterproof Antenna for Internet of Things Applications,”
Journal of Electromagnetic Waves and Applications, vol. 35(7), pp. 874–887,
2021.
[6] Y.I.A. Adel et al., “C-­shaped Antenna Based Artificial Magnetic Conductor
Structure for Wearable IoT Healthcare Devices,” Wireless Networks, vol. 27,
pp. 4967–4985, 2021.
[7] S.P. Cowsigan, D. Saraswady, “Substrate Integrated Waveguide Cavity Backed
Antenna for IoT Applications,” Journal of Ambient Intelligence and Humanized
Computing, pp. 1–6, 2021. doi: 10.1007/s12652-­020-­02742-­0.
[8] R. Roges, and P.K. Malik, “Planar and Printed Antennas for Internet of Things-­
enabled Environment: Opportunities and Challenges,” International Journal of
Communication Systems, vol. 34(15), pp. 1–32, 2021.
[9] D. Sharma, B.K. Kanaujia, and S. Kumar, “Compact Multi-­standard Planar
MIMO Antenna for IoT/WLAN/Sub-­6 GHz/X-­band Applications,” Wireless
Networks, vol. 27, pp. 2671–2689, 2021.
[10] A. Mushtaq, A. Rajawat, and S.H. Gupta, “Design of Antenna Array Based
Beam Repositioning for IoT Applications,” Wireless Personal Communications,
vol. 122, pp. 3205–3225, 2022.
[11] Z. Wang, T. Liang, and Y. Dong, “Metamaterial-­Based, Compact, Wide Beam-­
Width Circularly Polarized Antenna for 5G Indoor Application,” Microwave
and Optical Technology Letters, vol. 63(8), pp. 2171–2178, 2021.
[12] A. De, B. Roy, and A.K. Bhattacharjee, “Miniaturized Dual Band Consumer
Transceiver Antenna for 5G-­ Enabled IoT-­ Based Home Applications,”
International Journal of Communication Systems, vol. 34(11), pp. 1–14, 2021.
[13] U. Singh, and R. Mishra, “A Dual-­ Band High-­ Gain Substrate Integrated
Waveguide Slot Antenna for 5G Application,” Progress In Electromagnetics
Research C, vol. 119, pp. 191–200, 2022.
[14] H. Wang, X.B. Huang, and D.G. Fang, “A Single Layer Wideband U-­Slot
Microstrip Patch Antenna Array,” IEEE Antennas and Wireless Propagation
Letters, vol. 7, pp. 9–12, 2008.
[15] K.S. Kola and A. Chatterjee, “A Two-­element Array with Clover-­leaf Shaped
Antennas for X-­band Applications,” 7th International Conference on Signal
Processing and Integrated Networks (SPIN), Noida, India, pp. 349–354, 2020.
[16] K.S. Kola and A. Chatterjee, “A Linear Array of Christmas-­ tree Shaped
Antennas for DBS Applications,” International Conference on Computer,
Electrical & Communication Engineering (ICCECE), Kolkata, India, pp. 1–6,
2020.
[17] K. S. Kola and A. Chatterjee, “A Printed Array of High-­gain Fractal Antennas
for X-­band Applications,” International Conference on Communication,
Computing and Industry 4.0 (C2I4), Bangalore, India, pp. 1–6, 2020.
[18] K. S. Kola and A. Chatterjee, “A 1 x 2 Array of High-­gain Radiators for Direct
Broadcast Satellite (DBS) Services under Ku-­band,” 8th International Conference
on Signal Processing and Integrated Networks (SPIN), pp. 297–302, 2021.
212 Kalyan Sundar Kola and Anirban Chatterjee

[19] K. S. Kola and A. Chatterjee, “An Array of Tulip-­flower Shaped Printed Radiators
for Direct Broadcast Satellite (DBS) Applications,” Advanced Communication
Technologies and Signal Processing (ACTS), vol. 2021, pp. 1–6, 2021.
[20] E.J. Wilkinson, “An N-­ way Hybrid Power Divider,” IRE Transactions on
Microwave Theory Techniques, vol. 8(1), pp. 116–118, 1960.
[21] E.O. Kreyszig, Advanced Engineering Mathematics, New York, Wiley, 1983.
[22] High Frequency Electromagnetic Software SONNET (v14), North Syracuse,
NY 13212, April, 2013.
[23] CST STUDIO SUITE, (v2018), CST AG, Germany, 2018.
[24] Z. Ma, and A.E. Vandenbosch, “Low-­Cost Wideband Microstrip Arrays With
High Aperture Efficiency,” IEEE Transactions on Antennas and Propagation,
vol. 60(6), pp. 3028–3034, 2012.
[25] A. Vosoogh, and P.S. Kildal, “Simple Formula for Aperture Efficiency
Reduction Due to Grating Lobes in Planar Phased Arrays,” IEEE Transactions
on Antennas and Propagation, vol. 64(6), pp. 2263–2269, 2016.
[26] E.H. Newman, P. Bohley, and C.H. Walter, “Two Methods for the Measurement
of Antenna Efficiency,” IEEE Transactions on Antennas and Propagation, vol.
AP-­23(4), pp. 457–461, 1975.
[27] D.M. Pozar, and B. Kaufman, “Comparison of Three Methods for the
Measurement of Printed Antenna Efficiency,” IEEE Transactions on Antennas
and Propagation, vol. 36(1), pp. 136–139, 1988.
[28] G.S. Smith, “An Analysis of the Wheeler Method for Measuring the Radiating
Efficiency of Antennas,” IEEE Transactions on Antennas and Propagation, vol.
25(4), pp. 552–556, 1997.
[29] R. Chair, K.M. Luk, and K.F. Lee, “Radiation Efficiency Analysis on Small
Antenna by Wheeler Cap Method,” Microwave and Optical Technology
Letters, vol. 33(2), pp. 112–113, 2002.
[30] M.A. Moharram, and A.A. Kishk, “MIMO Antennas Efficiency Measurement
Using Wheeler Caps,” IEEE Transactions on Antennas and Propagation, vol.
64(3), pp. 1115–1120, 2016.
[31] C.R. Paul, Introduction to Electromagnetic Compatibility, 2nd edition, Wiley,
2006.
Chapter 14

Deep learning IoT platform

for dental disease detection
Sindhu P. Menon
School of Computing and Information Technology, REVA University,
Bangalore, India

Pramod Kumar Naik and Baskar Venugopalan

Department of Computer Science and Engineering, Dayananda Sagar
University, Bangalore, India

CONTENTS

14.1 Introduction�������������������������������������������������������������������������������� 213

14.2 Literature survey������������������������������������������������������������������������� 215
14.3 Suggested design������������������������������������������������������������������������� 216
14.3.1 YOLO object detection interpretation���������������������������� 221
14.4 Overall network structure����������������������������������������������������������� 224
14.5 Results and discussion����������������������������������������������������������������� 226
14.5.1 Performance measure����������������������������������������������������� 228
14.5.2 IoU (Intersection Over Union)���������������������������������������� 228
14.5.3 Precision and recall curve����������������������������������������������� 228
14.5.3.1 Analysis of bounding boxes using
histogram������������������������������������������������������� 230
14.5.3.2 Analysis of bounding boxes using different
type pf graphs������������������������������������������������ 231
14.5.3.3 Metrics analysis���������������������������������������������� 232
14.6 Conclusion���������������������������������������������������������������������������������� 235
References��������������������������������������������������������������������������������������������� 235

14.1 INTRODUCTION

Dental anatomy speaks about the structure of the tooth, its appearance and
how it is classified. If the subject concerns the structure and categories of the
tooth, then this is described as macroscopic anatomy. The portion above the
enamel junction is called the realm and is the crown of the tooth. This crown
appears clearly if an eruption occurs. There can be either one or many roots
in a single tooth.
In light of present-­day eating habits, most of us are prone to dental dis-
eases [24]. This is to the result of poor healthcare, low standards of living,

DOI: 10.1201/9781003407300-14 213

214 Sindhu P. Menon et al

unhygienic conditions and practices, and the use of drugs, tobacco and alco-
hol. In most situations, however, these diseases could be prevented if regu-
larly monitored. The monitoring of these diseases can also play a vital role
in controlling diabetes and also cerebrovascular and cardiovascular
diseases.
Deep learning is widely used for object classification mainly when large
data sets are involved. Algorithms such as CNN and RNN are used to pre-
dict diseases related to tooth using various forms of images [5].
Tooth loss normally occurs for baby teeth; once they are replaced by
human adult teeth the possibility of tooth loss due to some disease is much
reduced. Sometimes this will occur as a result of diseases such as dental
avulsion, cavities, and gum disease. Tooth plaque is a sticky layer of bacte-
rium that forms on teeth when we eat or drink. During this process, bacteria
transforms the food into acids, which we term tooth plaque. The acids
which are formed during this process destroy enamel.
Fluorosis is caused when teeth are overexposed to Halide (Halogen ion)
[6]. This condition normally takes place during the initial eight years of
childhood. It’s a serious condition which affects the teeth. The normal result
is that the teeth will lose color as a result. Periodontal Disease is to the result
of poor oral hygiene. This leads to infection in the gums and affects the bone
of the jaw. This can be stopped. In serious cases, it leads to tooth loss. This,
in turn, can lead to respiratory issues [23]. Professionals may even perform
surgeries in the most serious cases. If not treated, dental diseases can lead to
heart disease, stroke, diabetes, some cancers, Alzheimer’s, osteoporosis and
even some metastatic diseases.
“You Only Look Once” (YOLO) is used by [25] for object detection.
Deep learning algorithms which use object detection are as follows. One is
based on classification and the other on regression. The occurrence of objects
show up more often when the region of interest is identified which is proven
in the classification algorithm [27]. When CNN is applied on the region of
interest, the chances of object occurrence are high [25].
In this the detector must be run each time in the region of interest and this
slows down the computation. This is a major drawback of this algorithm.
Examples of such algorithms are R-­CNN. The regression-­based algorithms
are faster when compared to this. In this algorithm, there is no choice of
attention-­grabbing ROI among the images [27]. Here, it considers the entire
image and identifies the bounding boxes, thereby predicting the classes to
which they belong. YOLO is one such regression-­ based algorithm. The
YOLO detector is extraordinarily quick, hence its use in many applications,
such as self-­driving cars and others where period object detection is needed
[25]. This chapter applies YOLO to identify the various tooth-­related disor-
ders. It is structured as follows. Section 14.2 presents the literature review of
our work. Section 14.3 explains the methodology used in predicting the vari-
ous tooth diseases using YOLO. In Section 14.4, the results are discussed.
Finally, in Section 14.5 we give our conclusion and outline future work.
 Deep learning IoT platform for dental disease detection 215

14.2 LITERATURE SURVEY

A literature survey or review is the section which shows the varied [1] analy-
sis and research made within the field of our interest. In [22] author classi-
fied dental diseases using Convolutional Neural Networks. Diseases such as
periapical infection, periodontitis and cavity were considered [21]. In order
to obtain improved accuracy, transfer learning with a VGG16 pre-­trained
model is used. Training and testing was carried ot on 251 RVG x-­ray images.
A general accuracy of 88.46% was achieved. The limitations of this work
are that: For training purposes, the system requires large labelled data sets;
only when transfer learning was applied was an increase in accuracy identi-
fied; only x-­ray images were used for classification of diseases; the length of
the prediction time was not addressed; finally, the system seems to be more
appropriate for use in a laboratory setting rather than being a suitable prod-
uct for ordinary people [2].
The classification of skin was done using traditional methods wherein
features were extracted manually. Later, in [18] the authors have applied
CNN to extract features. The advantage of deep nets is that they need lesser
pre-­processing when compared with traditional methods [7]. Traditional
methods have used various clustering techniques and Support Vector
Machine for feature extraction. By contrast, better architectures such as
deep nets have the ability to analyse images better.
Research has shown that only a comparatively small proportion of those
people with tooth problems such as decay and fluorosis and so on visit a
dentist. This is because of the lack of awareness among the population [31].
Hence awareness programs regarding oral health have to be conducted at
regular intervals for the early diagnosis of diseases and the prevention of
dental problems. Deep Convolution Neural Networks (CNN) were applied
to medical problems with an input data size of over 10,000 images, which
require [11] higher cognitive process. They were able to achieve an efficiency
of 84%. Exploiting network cascades is a general task in the multitasking
environment. The authors have shown that [4] these networks are fast and
accurate during segmentation. Another framework was introduced for
detecting objects deep [32] in the network. This framework is completely
different from bounding box or sliding windows. Here, the corner point and
the middle point are linked to the detection of objects. Hence, scale or ratio
variation will not have any effect on this system.
IoT could be used for faster and cheaper [14] innovations for customers
in healthcare. Diagnosis and location of intervertebral discs was done using
filets and stacked autoencoders [16]. The data set for validation contained
102 images. Features were extracted using unsupervised learning techniques
and the results, when compared, were reasonably good. Studies have shown
that people with diabetes [19] and those with high levels of glycaemia are at
a higher risk of being diagnosed with periodontitis. Other diseases, like car-
diovascular and macrovascular complications, could also arise in patients
216 Sindhu P. Menon et al

with periodontitis. The Retinex algorithm was used [20] to scan the image
in order to study the serology of an infection caused by pathogens. Various
regression techniques were used in [29] to extract features to diagnose AD/
MCI through neuroimaging. Dimensionality was a major problem in this
approach. To overcome this issue, regression methods were used for feature
selection, thereby reducing the dimensionality.
A survey was conducted by researchers in [30] where they have listed the
various opportunities and research issues present in healthcare. A frame-
work was used in [33] to demonstrate the optimization of general purpose
algorithms. The authors have compared the various NFL theorems, to show
how they underperform when the data set is small. AIDS patients generally
have a low count of CD4 lymphocyte. Such patients are more prone to IRIS
if they have had [10] an infection and have undergone treatment for the
same. Such patients can be given treatment before HAART is initiated.

14.3 SUGGESTED DESIGN

The objective of this work is to train the images using YOLO3, to identify
the tooth and the type of image (Figure 14.1) and then, later, to classify it
into a proper class. The end product will be an app stored on the smartphone
or tablet. YOLO is designed in such a way that at a time it will try to focus
on each and every object which is available inside the images. In the faster
R-­CNN [10] family we try to send a feature which we will extract from the
CNN and we will append the feature map with our proposed region.
Suppose there is an input image (Figure 14.4). We will try to send the
input image with multiple objects in a pre-­trained network (CNN). After

Figure 14.1 Tooth detection procedure.

 Deep learning IoT platform for dental disease detection 217

Figure 14.2 YOLO grid system.

sending the data it will try to give you different feature maps from different
layers and send them into forward connection. Based on the previous fea-
ture map, we extract different objects in an input image.
The entire image will be split into grids (SxS) (Figure 14.2) where every
box will have its own centre point (centroid). With respect to every box we
are going to have two bounding boxes, which can have both vertical and
horizontal coordinates (X,Y).
With respect to the image in Figure 14.2, we have a single class where
every bounding box will have a width (W), a height (H) and confidence (C).
The model [34] then calculates the class score of class in each grid, leading
to the output of each grid being 5 × 1; which means 5 is the number of
classes and 1 is the grid. For each bounding box in each grid the process is
repeated, leading this way class confidence of all 98 (7(X-­axis) × 7(X-­axis)
× 2(1 + 1 1 on the X-­axis and the Y-­axis) grids in the image is generated.
Whenever the algorithm performs the first label of operation then it will
be able to get the bounding boxes in the first layer. We then have to set the
threshold for bounding boxes for the sum of the class as in Figure 14.3. If
the result is less than the set threshold, then it chooses the next layer, and all
the scores will be arranged in descending order [the Higher the confidence-­
First it comes [9]. YOLO uses NMS(Non maximal suppression algorithm).
This eliminates all the unnecessary bounding boxes and keeps the boxes
which are required and then it sets scores to zero for the redundant box.
In every bounding box we have 1-­Class decay. The algorithm tries to
extract confidence of decay of each and every bounding box (Figure 14.6)
and it then takes the confidence score from each and every bounding box1
218 Sindhu P. Menon et al

Figure 14.3 YOLO’s bounding box plotting.

Figure 14.4 IoT Web cloud.

class i.e decay and arranges it in descending order. After arranging the boxes,
it compares the top 2 scores as in Figure 14.5.
The comparison can be made by the equation given below

IOU  box1 box2   0.5 then set 0 score to  box2 

After evaluating all the bounding box scores, the value which is most accu-
rate will be mapped and the output will be predicted. The classes included
for training are fluorosis, decay, oral thrush, chipped tooth and fillings. For
training of the model on Dental images [21], a pre-­trained YOLO model,
 Deep learning IoT platform for dental disease detection 219

Figure 14.5 Comparison of scores.

Figure 14.6 System network architecture for smart dental health SaaS system.
220 Sindhu P. Menon et al

trained on the COCA and VOC data set, is used to improve the accuracy in
achieving multi-­object detection.
Optimizer Initialization: For the case study, an Adam Optimizer is being
used. Adam is efficient when working with large problems which have many
features. The combination of ‘gradient descent with momentum’ algorithm
and the ‘RMSP’ algorithm results in this optimizer [1].
Warm-­up steps are performed as it helps the network to slowly adapt to
the data. However, theoretically, the main reason for warm-­up steps is to
allow adaptive optimizers (e.g. Adam, RMSProp, …) to compute correct
statistics of the gradients.
Three layers encompass the network architecture of the Smart Dental
Health SaaS system shown in Figure 14.7

1. Input Image Layer

2. Intelligent Service Layer
3. Service Layer.

The outermost layer is the service layer, which coordinates with the regis-
tered dentists, healthcare centers, health providers and product suppliers.
These centers provide attractive offers and attract customers through their
infrastructure and technology.
This architecture helps in building positive loops with patients. This can
be accomplished by managing the patients’ data, recognize groups of
patients who have shown considerable improvement in health conditions
and informing them of the same, thereby creating a positive environment.
The core layer is called Smart Dental Service Layer, where the analysis
results of dental symptoms are obtained by data processing [2, 28].
The middleware acts as an intermediate layer for the API. The API inter-
acts with the middleware. All user request and logic support are provided by
the cloud APIs.

Figure 14.7 Dental image system logical data flow diagram.

 Deep learning IoT platform for dental disease detection 221

The images are captured using a high-­end camera or mobile. These devices
do the local computing and display results to the end user through 3G, 4G
or Wi-­Fi networks. The service layer assists users in checking the health
condition of their tooth. The data captured by these devices is fed to the
service layer. YOLO algorithm is then applied on these images to identify the
area of infection, if any. If the service layer confirms any defect in the tooth,
then medical advice may have to be taken. The dental images are stored
using Mongo database. The flow of services is shown in Figure 14.10. The
application has three layers: the input layer, the service layer and the client
layer.
HTTP service is accomplished using allocation and scheduling resources
through load-­balancing techniques. It also helps in implementing the con-
nections between the algorithms and the cloud. The logical flow diagram is
shown in Figure 14.7.
Steps in the flow diagram are:

1. Sample images are acquired and fed into the system.

2. The sample images are enhanced for accurate and common ground
predictions.
3. The colour texture is matched on to a single scale.
4. Classification of disease and coarse localization is done by using MS
VoTT software, where disease labeling and ROI (Region of Interest)
are determined.
5. Further preparation like train and test split of images and model
weights are added and segregated for model training.
6. The data is fed into YOLO and trained on training images [8].
7. The trained model is then loaded on to the application software for
consumption
8. The user then uploads the picture to the trained and validated model.
9. Predictions are made and the image is sent back to the user

14.3.1 YOLO object detection interpretation

Unlike the RCNN series, YOLO (Figure 14.12) classifies all problems related
to detection as a regression problem. It uses box regression and network for
classification [3, 12]. The specific method is: divide the image into N * N
grids, where the function of the grid is to predict the positions (x, y, w, h) of
Yb boxes, confidence, and class probability. The output dimension is N * N
* (Y * 5 + Ca), where Ca indicates the number of categories. The respon-
sibility of each grid is to predict a set of class probabilities. The number of
grids will not have any impact on the class probabilities.

1. Prediction of the Bounding box

YOLO foretells that anchor boxes are used by bounding boxes (Figure
14.12). This network figures 4 coordinates for each bounding box: ax,
222 Sindhu P. Menon et al

ay, aw, and ah. (tx, ty) represents [28] the coordinates of the upper-­
left corner of the current grid offset from the upper-­left corner of the
image and it is the distance, with pw, ph representing the width and
height of the prior box (prior). During training, a sum of squared er-
ror loss is used. True value by said gradient is the true value minus the
predicted value. YOLO uses logistic regression to predict a score for
each bounding box. The algorithm only matches an optimal prior box
for each truth value.
2. Multi-­label prediction
Each bounding box may contain multiple types of objects, that is,
multi-­label prediction. So logistic (sigmoid) is used instead of Soft-
Max, because SoftMax indicates that each box has only one type of
object, and sigmoid can predict multi-­label classification. In fact, a
[35] sigmoid classifier is applied to the results of each classification
calculation in logits to determine whether the sample belongs to a
certain category. During training, a binary cross-­entropy loss function
is used for class prediction [13].
3. Combine the characteristics of different convolution layers to extract
more fine-­grained information and make multi-­scale predictions
YOLO uses three different scales to predict the boxes. The network
uses a similar feature pyramid concept to extract features from differ-
ent scales. Several new convolution layers will be added to the original
basic feature extractor. Finally, a 3-­dimensional tensor will be used
to represent the bounding box, objectness, and class predictions. The
next step is to predict 3 boxes, so the tensor is N * N * [3* (4 + 1 +
80)]. This represents 80 class predictions,1 objectness prediction and
4 bounding boxes. Then we retrieve a feature set from the previous
network and merge it with the up sampled features. The objective is
to get an approximation of what would have been obtained if it was
done at a higher rate. The next step is to append some additional con-
volutional layers so that an alike tensor could be produced by process-
ing this feature map. Repeat this process to predict the size of the final
box. In the past, YOLO had difficulty in predicting small objects, but
now it has improved significantly through multi-­scale prediction.
4. Network structure (DarkNet53 = Darknet19 + ResNet)
Combine the residual thought to extract deeper semantic information.
Continuous 3 × 3 and 1 × 1 convolutional layers are still used. Predic-
tion will be performed on three different scales by up sampling. For
example, the 8 * 8 feature map up sampling and the 16 * 16 feature
maps are added and calculated again, so that smaller objects can be
predicted. A convolutional layer with a step size of 2 is used instead
of the pooling layer because the pooling layer will lose information.
5. Predict more targets
YOLO still uses k-­means clustering to determine template boxes and
predict 9 bounding boxes for each grid, which can improve recall.
 Deep learning IoT platform for dental disease detection 223

6. Loss function
Following the reference from [24], performance can be improved if we
add pre-­trained convolutional networks. Input resolution is increased
to 448*448 from 224*224. This is done because visual images which
are fed should be fine-­grained so that object detection becomes ac-
curate. The network consists of four convolution layers, two of which
are fully connected with weights initialized at random. The next step
is to compress the values in between 0 and 1. This can be achieved by
normalizing the width and height of the bounding box image. Next
parameterize the bounding box x and y coordinates to be the offsets of
a particular grid cell location so they are also bounded between 0 and
1. Leaky rectified linear activation function is applied on the interme-
diate layers and a linear activation function is used for the final layer.

 x, if x  0
  x  
0.1x, otherwise

The next step is to get an optimum model by finding out the sum of squared
error Figure 14.8. This results in zero confidence score and high loss [15].
In order to reduce the loss, we use two parameters, λcoord and λnoobj. These
parameters are used to determine the error based on the deviations. If the

Figure 14.8 YOLO detection using convolution.

224 Sindhu P. Menon et al

deviations are large, they cause greater effect in small boxes rather than the
large ones. To overcome this, the square root of the bounding boxes height
and width are considered instead of using them directly.

n2 M

1  a  a   b  b  

2 2
 cod obj
ij
^i i
^
i
x 0 y 0
n2

1     

M
 2 2
  coord obj
ij ci  c^i hi  h^i
x 0 y 0
n2 M s2 B

 
 C  C^ 2   
 1 C  C  
2
obj obj
 1 ^
ij
 i i
 noobj ij i i
x 0 y 0 i 0 j 0
s2

1   p c   p c 
2
obj
 ij i ^i

i 0 cClasses

During the training phase, for each grid cell, the algorithm predicts a
number of bounding boxes. Each object is overlooked by one predictor dur-
ing the training phase. This predictor later predicts the objects based on
IOU. The one which has the highest IOU w.r.t ground truth will be the pre-
diction. With this the predictors get specialized. The overall recall starts
improving as predictions are able to predict the aspect ratios, sizes or vari-
ous classes of objects better. The assumptions made during training are as
follows. The object which is in cell i is denoted as 1obj
ij . It also says that the
prediction is due to the jth bounding box present in cell i. Another fact
observed here is that the penalization of classification error happens by the
log function only if the grid cell contains that particular object. In the case
that the predictor has the greatest IOU of any grid cell, then the bounding
box coordinate gets penalized.

14.4 OVERALL NETWORK STRUCTURE

The basic network model is Google Net, but instead of using the inception
module, we have used 1 * 1 and 3 * 3 convolutional layers alternately.
Convolutional layer extracts features from the fully connected layer, a total
of 24 convolutional layers, 2 fully connected layers.
The network consists of 24-­Convolutional + 2 fully connected layers [35].
Sub-­network: This is a pre-­trained classification network, the input image
size is 224 * 224. It contains first 20 convolutional layers + 1 global average
pooling + 1 fully connected. Sub-­network: The second one is a target detec-
tion network, the input image size is 448 * 448. Loss function (square sum
loss function) including 4 parts: box center position x, y loss + box width
and height w, h loss + confidence loss + classification loss.
 Deep learning IoT platform for dental disease detection 225

YOLO advantages
• High speed. Seen as a regression problem, no complicated pipeline is
needed. Have a global understanding of the image.
• Other than RCNN, the features of the whole image is considered to
predict the bounding box, instead of RCNN, only the features of the
candidate bounding box can be used to predict the box.

The number of candidate boxes is much smaller, only 7 * 7 * 2 = 98. The

RCNN has 2000 selective searches, which are computationally intensive.
Representation of Bounding boxes is given in Figure 14.10.
In our model (in Figure 14.9), YOLO has used dimension clusters as
anchor boxes to predict bounding boxes. Let us assume that the cell is at an
offset from the top left corner of the image by a dimension (cx, cy) and the
width and height of the bounding box is pw, ph.

bx   t x  cx (14.1)

by   t y  cxy (14.2)

bw = pw etw. (14.3)

bh = pheth (14.4)

The performance parameters used to compute loss during training is the

sum of the squared error. The gradient is the error term computed as the
difference between ground truth value i.e. tˆ* and prediction which is
tˆ* – t*. If the equations above are inverted, it displays the ground truth
value. To predict the objectness score for each bounding box, YOLO uses
logistic regression. The score should have a value of 1 if the bounding box

Figure 14.9 Bounding boxes.

226 Sindhu P. Menon et al

Figure 14.10 Bounding box illustration.

prior overlaps a ground truth object by more than any other bounding box
prior. The cluster centroids can be used to estimate the width and height of
the box.
The sigmoidal function is used to predict the centre coordinates of the
box relative to the location of the filter. The value of threshold is 0.5. Each
ground truth object is assigned one bounding box prior. Sometimes there
will be no loss incurred if the assignment between bounding box prior and
ground truth doesn’t happen.
Bounding boxes (Figure 14.10) are boxes that are imaginary boxes. They
are around the objects checked for collision. There are two types of coordi-
nate systems, a 2D coordinate system and a 3D coordinate system, that are
used to detect objects. In Figure 14.10, the square boxes represent the
bounding boxes which, in turn, show the affected disease part of the tooth.
The accuracy of the prediction is also shown along with the bounding boxes.
In Figure 14.13, the digits along with the bounding boxes 0.90, 0.88, 0.72,
0.89 etc. represents accuracy of 90%, 88%, 72% and 89%

14.5 RESULTS AND DISCUSSION

Figures 14.11–14.15 show how the trained model is predicting dental

disease.
 Deep learning IoT platform for dental disease detection 227

Figure 14.11 Predicted results of dental disease fluorosis.

Figure 14.12 Predicted results of dental disease chipped tooth.

Figure 14.13 Predicted results of dental decay.

Figure 14.14 Predicted results of dental disease fillings.

The predicted model gives the results with predicting bounding boxes
with accuracy mentioned over it. Diseases like fluorosis, chipped tooth,
tooth decay, oral thrush and fillings are predicted by the proposed system. If
the value of the bounding box is 0.9 then this says that it is 90% accurate.
228 Sindhu P. Menon et al

Figure 14.15 Predicted results of dental disease oral thrush.

14.5.1 Performance measure
The metrics used for calculation of performance measures are IoU and
Precision and Recall. The same is illustrated in Figure 14.17.

14.5.2 IoU (Intersection Over Union)

• In object detection models, IoU is used to evaluate the performance of
the model.
• IoU (Figure 14.16) is the division of the union of two bounding boxes
for the ground truth to the predicted bounding box.

14.5.3 Precision and recall curve

Precision: It shows how many times the model is correct in positive predic-
tion. It gives an idea how much we can trust our algorithm when it predicts
“true”. The proportion of predictions guessed properly as “true” versus all
the positive predictions.
Recall: Recall tells us the percentage of all the correct output classified by
the model.
If we want to know the best threshold value of our proposed system then
we have to plot a Precision Recall curve (PR curve). Therefore, we have to

Figure 14.16 Representation of intersection over union.

 Deep learning IoT platform for dental disease detection 229

Figure 14.17 Precision and recall curve.

calculate the precision and recall values of the algorithm for different values
of threshold. We then use those values to plot a graph. In Figure 14.17, the
blue curve is the best for our proposed system.
In Figures 14.17–14.18, the curve of Precision and Recall is drawn. We
have obtained mAP (mean Average Precision) of the respective diseases. It is
evident that the mAP value is maximum for three diseases i.e. oral thrush,
chipped tooth, fillings is about 99.5% and fluorosis is of 99.1% and tooth
decay is of 99.3% respectively.
The higher the mAP value the greater the accuracy in respective disease
prediction. We can conclude that the average mAP value of all diseases is
about 99.4%.
An 80% confidence interval means that we are confident at 80% that the
real value is in the specified interval. Large samples output mean with more
precision when compared with smaller data. As a result, when a large sam-
ple is taken, the confidence could be smaller. As we increase the threshold of
confidence the precision will go up (Figure 14.19). The average confidence
of all the classes in the above graph is 86% at 1.0.
Using the F1 score curve shown in Figure 14.20, the balance between
precision and recall can be visualized and a design point can be deter-
mined using a curve graph. The F1 curve shows that optimal value for
precision and recall is obtained when confidence is 0.506. We would want
our model to have cases which have higher confidence value. The maxi-
mum value of F1 occurs at 0.81. Hence it is ideal to select confidence of
0.79 as it results in F1 value of 0.8 which is pretty close to the maximum
F1 value.
230 Sindhu P. Menon et al

Figure 14.18 F1 and confidence curve.

14.5.3.1 Analysis of bounding boxes using histogram

To describe the spatial location of an object we use a bounding box in object
detection.
The bounding box is usually rectangular in shape where it has x-­and y-­
coordinates. Another commonly used bounding box representation is the
 Deep learning IoT platform for dental disease detection 231

Figure 14.19 Precision and confidence curve.

(x, y) (x, y)-axis coordinates of the bounding box center, and the width and
height of the box, as shown in the histogram in Figure 14.21.

14.5.3.2 Analysis of bounding boxes using different type pf

graphs
A bounding box has four values to represent it, i.e. x_center, y_center, width,
height. The center of the bounding box is x_center and y_center is the nor-
malized coordinates.
232 Sindhu P. Menon et al

Figure 14.20 Histogram plotting for bounding boxes.

We consider pixel values of x and y to normalize the coordinates. This

becomes the center of the bounding box on the x-­and y-­axes. Next, the
value of x is divided by the width of the image and value of y is divided by
the height of the image. They are also normalized.
In Figure 14.24, we can see the different types of representation of the
bounding boxes. Each graph takes two parameters (either x and y) or height
and width to represent the bounding boxes of the dental data. We can also
see the range of instances of each disease by representing it in the bar graph.
The range covered by all the bounding boxes in the data set is also shown as
a summary.

14.5.3.3 Metrics analysis
The graphs in Figures 14.22–14.24 show the performance of each metric of
the YOLO algorithm.
Recall: This parameter indicates the number of correct instances that
were [26] obtained.
Deep learning IoT platform for dental disease detection 233
Figure 14.21 Different graph plotting for bounding boxes.
234 Sindhu P. Menon et al

Figure 14.22 Performance of recall.

Figure 14.23 Performance of metric mAP.

mAP: To measure the accuracy of object detectors, the metric used is

Average [17] Precision.
In this we try to find the area covered by the precision recall curve. Since
the values of recall and precision lie between 0 and 1, the value of mAP will
also lie between 0 and 1.


AP  pr  re  dr
0
(14.5)

The x-­axis represents the epochs count and the y-­axis represents the
respective metric values evaluated during the training of the data set.
 Deep learning IoT platform for dental disease detection 235

Figure 14.24 Performance of metric precision.

The recall, precision and mAP gradually increase with the increase in epochs
count. It further leads to an increase in the accuracy of the overall model.

14.6 CONCLUSION

The present results demonstrates the prediction of tooth diseases such as

tooth decay, tooth filling, fluorosis, chipped tooth and oral thrush through
an app. We have seen that YOLO is very good at object detection and
produces accurate results. Recall of 91.3% was obtained, as shown in the
graph. Precision of around 89% was obtained with 45 step counts. We have
observed that as the number of epochs was increased, the accuracy used to
increase. This study has its own limitations, one of them being that a high-­
speed network is necessary for the app to work well. At times, dentists may
have to perform some physical tests for the diagnosis of the disease which
cannot be provided by our system. In addition to this, the app on the mobile
can be used in the diagnosis and to evaluate the disease. Currently, many
nations may not have such access to the technology which is required for
our system to work. In future, we hope to overcome all these flaws and build
a more robust system.

REFERENCES

1. Abar S, Abe T, Kinoshita T, “A next generation knowledge management system

architecture,” In 18th international conference on advanced information net-
working and applications, AINA 2004, (vol. 2, pp. 191–195). IEEE, 2004.
2. Birogul S, Temür G, Kose U, “YOLO object recognition algorithm and buy-­sell
decision model over 2D candlestick charts,” IEEE Access, vol. 8, pp. 91894–
91915, 2020.
236 Sindhu P. Menon et al

3. Chen, CH, Handbook of pattern recognition and computer vision, World

Scientific, 2015.
4. Dai J, He K, Sun J, “Instance-­ aware semantic segmentation via multi-­
task network cascades,” In Proceedings of the IEEE conference on com-
puter vision and pattern recognition. pp. 3150–3158, 2018. doi: 10.1109/
CVPR.2016.343
5. Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, Thrun S,
“Dermatologist-­level classification of skin cancer with deep neural networks,”
Nature, vol. 542, no. 7639, pp. 115–118, 2017.
6. He K, Zhang X, Ren S, Sun J, “Deep residual learning for image recognition,”
In Proceedings of the IEEE conference on computer vision and pattern recog-
nition. pp. 770–778, 2016. doi: 10.1109/CVPR.2016.90
7. Henriques J, Neves N, de Carvalho P, “XV Mediterranean Conference
on Medical and Biological Engineering and Computing,” In MEDICON
Proceedings of MEDICON 2019, Coimbra, Portugal, vol. 76, Springer Nature,
September 26–28, 2019.
8. Hua, G., Jégou H, “Computer Vision–ECCV”, 2016 Workshops: Amsterdam,
The Netherlands, 2016, Proceedings, Part II (Vol. 9914). Springer, 2016.
9. Ibadov, SR, Kalmykov, BY, Ibadov RR, Sizyakin, RA, “ Method of automated
detection of traffic violation with a convolutional neural network,” In EPJ Web
of Conferences, vol. 224, pp. 04004. EDP Sciences, 2019.
10. Khanagar SB, Alfouzan K, Awawdeh, M, Alkadi L, Albalawi F, Alfadley A,
“Application and performance of artificial intelligence technology in detec-
tion, diagnosis and prediction of dental caries (DC)—A systematic review,”
Diagnostics, vol. 12, no. 5, p. 1083, 2022. doi: 10.3390/diagnostics12051083
11. Kwasigroch A, Mikołajczyk A, Grochowski M, “Deep convolutional neural
networks as a decision support tool in medical problems–malignant melanoma
case study,” In Polish Control Conference, Springer, Cham, pp. 848–856, Jun
18, 2017.
12. Li, S, Gu, X, Xu, X, Xu, D, Zhang, T, Liu Z, Dong Q, “Detection of con-
cealed cracks from ground penetrating radar images based on deep learn-
ing algorithm,” Construction and Building Materials, vol. 273, pp. 121949,
2021.
13. Liu S, Li, X, Gao M, Cai Y, Nian R, Li P, Yan T, Lendasse A, “Embedded online
fish detection and tracking system via YOLOv3 and parallel correlation filter,”
In OCEANS 2018 MTS/IEEE Charleston, pp. 1–6, IEEE, 2018, Oct.
14. Metcalf D, Milliard ST, Gomez M, Schwartz M, “Wearables and the internet of
things for health: Wearable, interconnected devices promise more efficient and
comprehensive health care,” IEEE Pulse, vol. 7, no. 5, pp. 35–39, 2016. http://
doi:10.1109/MPUL.2016.2592260.
15. Nie G, Liu Y and Wang Y, “DovCut: a draft based online video compositing
system,” In Proceedings of the 15th ACM SIGGRAPH Conference on Virtual-
Reality Continuum and Its Applications in Industry, Volume 1, pp. 129–136,
2016 Dec.
16. Oktay AB, Akgul YS, “Diagnosis of degenerative intervertebral disc disease
with deep networks and SVM,” In International Symposium on Computer and
Information Sciences, Springer, Cham. pp. 253–261, Oct 27, 2016.
17. pastel.archives-ouvertes.fr
 Deep learning IoT platform for dental disease detection 237

18. Pang Z, Chen Q, Tian J, Zheng L, Dubrova E, “Ecosystem analysis in the design
of open platform-based in-home healthcare terminals towards the internet-of-
things. In 2013 15th international conference on advanced communications
technology (ICACT), IEEE, pp. 529–534, Jan 27, 2013.
19. Preshaw PM, Alba AL, Herrera D, Jepsen S, Konstantinidis A, Makrilakis K,
Taylor R,“Periodontitis and diabetes: a two-way relationship,”Diabetologia, vol.
55, no.1, pp. 21–31, 2012. https://dx.doi.org/10.1007%2Fs00125-011-2342-y.
20. Provenzi E, De Carli L, Rizzi A, Marini D, “Mathematical definition and analy-
sis of the Retinex algorithm,” JOSA A., vol. 22, no. 12, pp. 2613–2621, 2005.
21. Prajapati SA, Nagaraj R and Mitra S, “Classification of dental diseases
using CNN and transfer learning,” In 2017 5th International Symposium on
Computational and Business Intelligence (ISCBI), IEEE, pp. 70–74, Aug, 2017.
22. Prajapati SA, Nagaraj R, Mitra S, “Classification of dental diseases using CNN
and transfer learning,” In 2017 5th International Symposium on Computational
and Business Intelligence (ISCBI), IEEE, pp. 70–74, Aug 11, 2017.
23. Pussinen PJ, Jousilahti P, Alfthan G, Palosuo T, Asikainen S, Salomaa V,
“Antibodies to periodontal pathogens are associated with coronary heart dis-
ease,” Arteriosclerosis, Thrombosis, and Vascular Biology, vol. 23, no.7, pp.
1250–1254, 2003. http://dx.doi.org/10.1161/01.ATV.0000072969.71452.87.
24. Ren S, He K, Girshick R, Zhang X, Sun J, “Object detection networks on
convolutional feature maps,” IEEE Transactions On Pattern Analysis and
Machine Intelligence, Aug 17, vol. 39, no. 7, pp. 1476–1481, 2016. http://
dx.doi.org/10.1109/TPAMI.2016.2601099.
25. robocademy.com
26. Sathya, B. and Neelaveni, R., “Transfer learning based automatic human iden-
tification using dental traits-an aid to forensic odontology,” Journal of Forensic
and Legal Medicine, 76, pp. 102066, 2020. http://dx.doi.org/10.1016/j.
jflm.2020.102066.
27. Smys, S., Tavares, J.M.R.S., Balas, V.E. and Iliyasu, A.M., “Computational
Vision and Bio-Inspired Computing,” 2019.
28. southern-sweepers.com.
29. Suk HI, Lee SW, Shen D, “Deep sparse multi-task learning for feature selection
in Alzheimer’s disease diagnosis,” Brain Structure and Function, vol. 221, no.
5, pp. 2569–2587, 2016. http://doi:10.1007/s00429-015-1059-y.
30. Sundaravadivel P, Kougianos E, Mohanty SP, Ganapathiraju MK, “Everything
you wanted to know about smart health care: Evaluating the different tech-
nologies and components of the internet of things for better health,” IEEE
Consumer Electronics Magazine, Dec 13, vol. 7, no. 1, pp. 18–28, 2017. https://
doi.org/10.1109/MCE.2017.2755378.
31. Varghese CM, Jesija JS, Prasad JH, Pricilla RA, “Prevalence of oral diseases
and risks to oral health in an urban community aged above 14 year,” Indian
Journal of Dental Research, 2019, vol. 30, no.6, pp.844, http://doi:10.4103/
ijdr.IJDR_42_18.
32. Wang X, Chen K, Huang Z, Yao C, Liu W, “Point linking network for object
detection,” arXiv preprint arXiv:1706.03646. Jun 12, 2017.
33. Wolpert DH, Macready WG, “No free lunch theorems for optimization,” IEEE
Transactions On Evolutionary Computation, vol. 1, no. 1, pp. 67–82, 1997.
http://doi.org.10.1109/4235.585893.
238 Sindhu P. Menon et al

34. Almalki, YE, Din, AI, Ramzan, M, Irfan, M, Aamir, KM, Almalki, A, Alotaibi,
S, Alaglan, G, Alshamrani, HA, Rahman, S, “Deep learning models for clas-
sification of dental diseases using orthopantomography X-ray OPG images”,
Sensors, vol. 22, no. 19, p. 7370, 2022.
35. Zhao, K. and Ren, X., “May. Small aircraft detection in remote sensing
images based on YOLOv3”, IOP Conference Series: Materials Science and
Engineering, vol. 53, no. 1, pp. 012056, IOP Publishing, 2013.
Index

Pages in italic refer figures.

Adaptive Honey Badger Algorithm Cardiovascular disease, 115, 124, 127,

(AHBA), 171, 176–178 214, 215
Advanced Driver Assistance System Chronic kidney disease, 115, 125
(ADAS), 182 CISCO, 5, 7
Additive Manufacturing, 24 Cloud Computing, 1, 2, 7, 19, 23, 28,
Aerial robots, 156 29, 37, 39, 40, 41, 43, 44, 49,
AIDS, 216 50, 57, 92
Amazon web services (AWS), 49 Clustering, 118–120, 174, 215, 222
Analytics, 1, 6, 9, 10, 25, 28, 50, 92, Computed Tomography (CT), 116
114 Computer Vision, 32, 33, 115, 121,
Apache, 77 173, 174
Application Layer, 4, 8, 68–70, 74, 77, Congestion, 8, 41, 42, 46, 47, 143, 154,
78 183, 188
Application Programming Interface Connected vehicles (CVs), 182–185, 189
(API), 8 Constrained Application Protocol
Array, 58, 94, 195–197, 199, 201, 202, (CoAP), 3, 70, 74
204–206, 209, 210 Convolutional Neural Networks
Arduino Mega, 93 (CNN), 170, 215
Artificial Intelligence, 10, 13, 14, 26–29, Convolution Neural Long Short-Term
34, 50, 53, 57, 113, 115, 143, Memory (CNLSTM), 171,
183 176–178
Augmented Reality, 22 Correction factor (CF), 197, 208
Automotive Cybersecurity, 181 Covid-19, 126
Automotive Standards, 184 CST microwave studio, 200, 202, 206
Automotive Systems, 181, 183, 184 Cross-layer DoS attack, 80, 85
Autonomous Robot, 24 Cyber-Physical Manufacturing System
AutoNode, 185 (CPMS), 55
Autonomy, 23, 25, 46, 49, 182 Cyber Physical Production System
(CPPS), 19
Bandwidth, 25, 38, 39, 41, 46, 78 Cyber Physical System (CPS), 18, 23,
Big data, 19, 25, 28, 29 28, 29, 54, 57, 58
Blockchain, 9, 10, 39, 56, 114, 115, Cybersecurity, 2, 23, 26, 183, 184, 188
122–124, 127
Bluetooth, 70, 186 Data lifecycles, 50
Bounding box, 172, 214, 215, 217, 218, Data sovereignty, 46, 47
221–228, 230, 232 Decision Tree, 118

239
240 Index

Deep learning (DL), 121, 154, 170, 172, Industrial AI, 13, 28, 29, 30, 34
174, 183, 214 Industry 4.0, 1, 16–19, 21–26, 28, 34,
Degree of freedom (DOF), 103, 156, 54–57, 59, 60, 62, 63
159–161, 164, 167 India, 93, 95, 108, 114, 148, 182, 185
Denial of Service (DoS), 68, 70, 72, Information Technology (IT), 16, 37,
74, 77 47, 48, 98, 170
Density-Based Spatial Clustering of Integrated circuit (IC), 144
Applications with Noise, 119 International Telecommunication Union
Diabetes Mellitus (DM), 115, 124 (ITU), 96
Diagnostics, 183, 186, 188, 192 Internet of Manufacturing Services
Diagnostic trouble codes (DTCs), 186 (IoMs), 21
Digital production, 19 Internet of Medical Things, 122
Distributed Denial of Service, 68, 77 Internet of People (IoP), 21
Internet of Vehicles, 9, 38
Edge Computing, 1, 6–8, 37–43, 45–50, Intersection Over Union (IOU), 228
158 Interworking Proxy Entities (IPE), 5
Edge Security, 46 IPv4, 3
Electrocardiograms (ECGs), 122 IPv6, 3, 74, 76
Electrooculograms (EOGs), 122 IRIS, 216
Electroencephalograms (EEGs), 122
Electroglottographs (EGGs), 122 Japan, 27, 55, 143
Electromagnetic compatibility (EMC),
208 Kappa, 118
Electromyograms (EMGs), 122 Kidney disease, 125
Electronic Control Unit (ECU), 68, 183 k-means clustering, 119, 222
Embedded systems, 40, 53, 93 k-Nearest Neighbour, 118
Ensemble learning, 173
ESP8266, 145, 146, 196 Latency, 7–9, 38, 41, 42, 44, 47
Event–Condition–Action (ECA), Legged Robots, 156
139–141 Light Dependent Resistor (LDR), 144
Local area network (LAN), 41, 42
Fibonacci, 197, 198, 199, 206, 209, 210 Loss function, 222–224
Fifth Generation, 27 Low Power and Lossy Networks (RPL),
Fog computing, 1, 6–8, 10, 37, 43, 70
44, 49 Long short-term memory (LSTM), 173
Fractal, 196 Long-Term Evolution (LTE), 195
Fused deposition modeling (FDM), 24
FUSION, 120, 126, 167 MAC, 3
F1 Score, 118, 229 Machine Learning (ML), 28, 47, 81,
113, 115, 116, 143, 170, 174,
GitHub, 68 183
Glomerular Filtration Rate, 125 Magnetic Resonance Imaging (MRI),
116
Healthcare, 8, 9, 47, 61, 113–116, 122, MANET, 76
124, 127, 174, 213, 215, 216, Manipulators, 156, 157, 160, 162
220 MATLAB, 164
HTTP, 70, 76, 77 Message Queue Telemetry Transport
Hybrid Robot, 157 (MQTT), 3, 70
Mobile edge computing (MEC), 8, 39
IEEE, 3, 74
Industrial Revolution, 14, 15, 16 Near-Field Communication (NFC), 96
Industrial wireless protocols, 25 Network Layer, 4, 68–70, 76, 78, 79, 81
 Index 241

Network Time Protocol (NTP), 78 Software Defined Networks (SDN), 8

NodeMCU, 145, 146, 148–150 SONNET, 200, 206
Non maximal suppression (NMS), 217 Stochastic Neural Analog
Reinforcement Calculator, 27
On-board Diagnostics (OBD), 182 Structured Query Language (SQL), 139,
Optimization, 21–23, 34, 47, 57, 118, 148
171, 176, 216 Supervisory control and data
acquisition (SCADA), 139
Passive infrared (PIR), 146, 158 Supervised learning, 116, 117, 121
Periodontal Disease, 214 Support vector machines (SVM), 118
Positron Emission Tomography (PET),
116 Tooth loss, 214
Precision, 23, 228, 229, 234, 235 Tooth plaque, 214
Pulse width modulation (PWM), 144 Transport Layer Security (TLS), 76
Python, 166 Transportation, 3, 8, 47, 92, 93, 106,
181, 182
Quality of Service, 8, 70, 195 Triple Space Computing (TSC), 77

RAM, 74 Ultrasonic sensor, 144, 146, 148, 150,

Random Forest, 118 152, 158
Raspberry Pi, 93, 166 Unmanned ground vehicle (UGV), 160
Reactive Metabolites (RM), 124 Unsupervised learning, 116, 118, 120
Recall, 172, 222, 228 USB, 76, 146, 166, 186
Received Signal Strength Indicator
(RSSI), 76 VANETS, 10
Reinforcement Learning, 116, 120, 187 Visual Geometry Group (VGG), 187,
Respiratory Syndrome Coronavirus-2 215
(SARS-CoV-2), 126 VIRTUS, 77
RISC, 146
Received Signal Strength Indicator Wheeled Mobile Robots (WMRs), 156
(RSSI), 76 Wide area network (WAN), 40
Wide-band, 195
Selective laser melting (SLM), 24 Wi-Fi, 49, 70, 146, 148, 150, 221
Semi-supervised learning, 116, 120 Wilkinson power divider, 202, 204,
Sensors, 1–5, 18, 21, 23, 33, 37, 43, 46, 205, 210
47, 55, 57, 67, 70, 92, 94, 114, Wireless Sensor Networks (WSN), 3,
122–124, 127, 134, 135, 143, 10, 75
144, 146, 148, 150, 154, 158, World Health Organizations (WHO),
159, 176, 177, 182, 183, 185, 114
189 WPAN, 3, 74
Service Layer, 4, 5, 69, 220, 221
Servo motor, 144, 145, 148–150, 152 XAMPP, 144, 148, 150
Simple Object Access Protocol (SOAP),
139 You Only Look Once (YOLO), 87, 214
Smart cities, 44, 61, 93, 106, 143, 174 YOLOv3, 171, 173, 176, 177
Society of Automotive Engineers (SAE),
181 Zigbee, 3, 70