Full download ebooks at ebookmeta.

com

Advances in Signal and Data Processing: Select

Proceedings of ICSDP 2019 1st Edition S. N.
Merchant (Editor)

https://ebookmeta.com/product/advances-in-signal-and-data-
processing-select-proceedings-of-icsdp-2019-1st-edition-s-n-
merchant-editor/

OR CLICK BUTTON

DOWLOAD NOW

Download more ebook from https://ebookmeta.com

More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Advances in Signal Processing and Communication

Engineering: Select Proceedings of ICASPACE 2021 Pradip
Kumar Jain

https://ebookmeta.com/product/advances-in-signal-processing-and-
communication-engineering-select-proceedings-of-
icaspace-2021-pradip-kumar-jain/

Advances in Manufacturing and Industrial Engineering

(Select Proceedings of ICAPIE 2019) 2nd Edition Singari

https://ebookmeta.com/product/advances-in-manufacturing-and-
industrial-engineering-select-proceedings-of-icapie-2019-2nd-
edition-singari/

Recent Advances in Artificial Intelligence and Data

Engineering: Select Proceedings of AIDE 2020 (Advances
in Intelligent Systems and Computing, 1386)

https://ebookmeta.com/product/recent-advances-in-artificial-
intelligence-and-data-engineering-select-proceedings-of-
aide-2020-advances-in-intelligent-systems-and-computing-1386/

Advances in Systems Engineering: Select Proceedings of

NSC 2019 (Lecture Notes in Mechanical Engineering) V.
H. Saran (Editor)

https://ebookmeta.com/product/advances-in-systems-engineering-
select-proceedings-of-nsc-2019-lecture-notes-in-mechanical-
engineering-v-h-saran-editor/
Signal Processing in Medicine and Biology: Innovations
in Big Data Processing 1st Edition Iyad Obeid

https://ebookmeta.com/product/signal-processing-in-medicine-and-
biology-innovations-in-big-data-processing-1st-edition-iyad-
obeid/

Intelligent Multi-Modal Data Processing (The Wiley

Series in Intelligent Signal and Data Processing) 1st
Edition Soham Sarkar

https://ebookmeta.com/product/intelligent-multi-modal-data-
processing-the-wiley-series-in-intelligent-signal-and-data-
processing-1st-edition-soham-sarkar/

Advances in Civil Engineering: Select Proceedings of

ICACE 2020 1st Edition Scott Arthur

https://ebookmeta.com/product/advances-in-civil-engineering-
select-proceedings-of-icace-2020-1st-edition-scott-arthur/

Communications Signal Processing and Systems

Proceedings of the 11th International Conference on
Communications Signal Processing and Systems Vol 1
Qilian Liang
https://ebookmeta.com/product/communications-signal-processing-
and-systems-proceedings-of-the-11th-international-conference-on-
communications-signal-processing-and-systems-vol-1-qilian-liang/

Advances in Thermofluids and Renewable Energy Select

Proceedings of TFRE 2020 1st Edition Pinakeswar Mahanta

https://ebookmeta.com/product/advances-in-thermofluids-and-
renewable-energy-select-proceedings-of-tfre-2020-1st-edition-
pinakeswar-mahanta/
Lecture Notes in Electrical Engineering 703

S. N. Merchant
Krishna Warhade
Debashis Adhikari Editors

Advances
in Signal
and Data
Processing
Select Proceedings of ICSDP 2019
Lecture Notes in Electrical Engineering

Volume 703

Series Editors

Leopoldo Angrisani, Department of Electrical and Information Technologies Engineering, University of Napoli
Federico II, Naples, Italy
Marco Arteaga, Departament de Control y Robótica, Universidad Nacional Autónoma de México, Coyoacán,
Mexico
Bijaya Ketan Panigrahi, Electrical Engineering, Indian Institute of Technology Delhi, New Delhi, Delhi, India
Samarjit Chakraborty, Fakultät für Elektrotechnik und Informationstechnik, TU München, Munich, Germany
Jiming Chen, Zhejiang University, Hangzhou, Zhejiang, China
Shanben Chen, Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai, China
Tan Kay Chen, Department of Electrical and Computer Engineering, National University of Singapore,
Singapore, Singapore
Rüdiger Dillmann, Humanoids and Intelligent Systems Laboratory, Karlsruhe Institute for Technology,
Karlsruhe, Germany
Haibin Duan, Beijing University of Aeronautics and Astronautics, Beijing, China
Gianluigi Ferrari, Università di Parma, Parma, Italy
Manuel Ferre, Centre for Automation and Robotics CAR (UPM-CSIC), Universidad Politécnica de Madrid,
Madrid, Spain
Sandra Hirche, Department of Electrical Engineering and Information Science, Technische Universität
München, Munich, Germany
Faryar Jabbari, Department of Mechanical and Aerospace Engineering, University of California, Irvine, CA,
USA
Limin Jia, State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing, China
Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences, Warsaw, Poland
Alaa Khamis, German University in Egypt El Tagamoa El Khames, New Cairo City, Egypt
Torsten Kroeger, Stanford University, Stanford, CA, USA
Qilian Liang, Department of Electrical Engineering, University of Texas at Arlington, Arlington, TX, USA
Ferran Martín, Departament d’Enginyeria Electrònica, Universitat Autònoma de Barcelona, Bellaterra,
Barcelona, Spain
Tan Cher Ming, College of Engineering, Nanyang Technological University, Singapore, Singapore
Wolfgang Minker, Institute of Information Technology, University of Ulm, Ulm, Germany
Pradeep Misra, Department of Electrical Engineering, Wright State University, Dayton, OH, USA
Sebastian Möller, Quality and Usability Laboratory, TU Berlin, Berlin, Germany
Subhas Mukhopadhyay, School of Engineering & Advanced Technology, Massey University,
Palmerston North, Manawatu-Wanganui, New Zealand
Cun-Zheng Ning, Electrical Engineering, Arizona State University, Tempe, AZ, USA
Toyoaki Nishida, Graduate School of Informatics, Kyoto University, Kyoto, Japan
Federica Pascucci, Dipartimento di Ingegneria, Università degli Studi “Roma Tre”, Rome, Italy
Yong Qin, State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing, China
Gan Woon Seng, School of Electrical & Electronic Engineering, Nanyang Technological University,
Singapore, Singapore
Joachim Speidel, Institute of Telecommunications, Universität Stuttgart, Stuttgart, Germany
Germano Veiga, Campus da FEUP, INESC Porto, Porto, Portugal
Haitao Wu, Academy of Opto-electronics, Chinese Academy of Sciences, Beijing, China
Junjie James Zhang, Charlotte, NC, USA
The book series Lecture Notes in Electrical Engineering (LNEE) publishes the
latest developments in Electrical Engineering - quickly, informally and in high
quality. While original research reported in proceedings and monographs has
traditionally formed the core of LNEE, we also encourage authors to submit books
devoted to supporting student education and professional training in the various
fields and applications areas of electrical engineering. The series cover classical and
emerging topics concerning:
• Communication Engineering, Information Theory and Networks
• Electronics Engineering and Microelectronics
• Signal, Image and Speech Processing
• Wireless and Mobile Communication
• Circuits and Systems
• Energy Systems, Power Electronics and Electrical Machines
• Electro-optical Engineering
• Instrumentation Engineering
• Avionics Engineering
• Control Systems
• Internet-of-Things and Cybersecurity
• Biomedical Devices, MEMS and NEMS
For general information about this book series, comments or suggestions, please
contact [email protected].
To submit a proposal or request further information, please contact the
Publishing Editor in your country:
China
Jasmine Dou, Editor ([email protected])
India, Japan, Rest of Asia
Swati Meherishi, Editorial Director ([email protected])
Southeast Asia, Australia, New Zealand
Ramesh Nath Premnath, Editor ([email protected])
USA, Canada:
Michael Luby, Senior Editor ([email protected])
All other Countries:
Leontina Di Cecco, Senior Editor ([email protected])
** This series is indexed by EI Compendex and Scopus databases. **

More information about this series at http://www.springer.com/series/7818

S. N. Merchant Krishna Warhade
• •

Debashis Adhikari
Editors

Advances in Signal and Data

Processing
Select Proceedings of ICSDP 2019

123
Editors
S. N. Merchant Krishna Warhade
Department of Electrical Engineering Department of Electronics and
Indian Institute of Technology Bombay Communication Engineering
Mumbai, Maharashtra, India MIT World Peace University
Pune, Maharashtra, India
Debashis Adhikari
School of Electrical Engineering
MIT Academy of Engineering
Pune, Maharashtra, India

ISSN 1876-1100 ISSN 1876-1119 (electronic)

Lecture Notes in Electrical Engineering
ISBN 978-981-15-8390-2 ISBN 978-981-15-8391-9 (eBook)
https://doi.org/10.1007/978-981-15-8391-9
© Springer Nature Singapore Pte Ltd. 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

The present era of technological advancements is moving at a very fast pace. It is

therefore essential to keep oneself abreast with all the recent developments in
various engineering domains. The book discusses various state–of–the-art devel-
opments in the diverse area of communications, data processing and signal pro-
cessing, and the endeavor has been to bring together some of these developments in
a concise platform that would benefit students, researchers, academicians and
industry people. The chapters presented in the book have been selected on the basis
of relevance and mathematical deliberations on the topics. Apart from the
above-listed domains, this book has additionally included topics on social issues
providing advanced technological solutions.
Chapter “Deep Semantic Segmentation for Self-driving Cars” introduces the
technique of semantic segmentation of urban scene for a self-driving car that
comprises three sub-systems in navigation, viz. lane finding, urban scene under-
standing and geo-positioning. In Chapter “Shot Boundary Detection Using
Artificial Neural Network,” hybrid video shot boundary detection process using
feature extraction by mean log difference is discussed in combination with artificial
neural network techniques. A system for leaf parameter analysis is proposed in
Chapter “Custard Apple Leaf Parameter Analysis, Leaf Diseases, and Nutritional
Deficiencies Detection Using Machine Learning,” where detection of N, P and K
deficiencies and leaf diseases is accomplished using K-nearest neighbors (k-NN)
and support vector machine (SVM) algorithms. A typical problem of recognizing
and removing the rain streaks on photographs by an improved convolutional neural
network (CNN) architecture is discussed in Chapter “Single Image Rain Removal
Using Convolutional Neural Network.” A study of voice samples for two disorders—
hypo and hyper—along with normal voice samples is considered in Chapter
“A Robust Approach of Estimating Voice Disorder Due to Thyroid Disease” to create
a databank for three classes—normal, hypo and hyper. A combined classifier,
i.e., SVM and HMM (hidden Markov model), was utilized.
In Chapter “Face Recognition Using Golden Ratio for Door Access Control
System,” a combination of Viola–Jones face detection method with characteristics
of extracting golden ratio is discussed to improve the security of sensitive places

v
vi Preface

through facial recognition. This book has included chapters on topics such as
underwater detection of objects, adaptive background subtraction models for shot
detection, acoustic classification and evaluation of bird species using support vector
machine and artificial neural networks. In Chapter “Feature-Based Model for
Landslide Prediction Using Remote Sensing and Digital Elevation Data,” a study to
generate landslide susceptible maps and landslide hazard zonation maps is pre-
sented using the digital elevation model for the prediction of future landslides.
Chapter “Emotion Recognition using Gamma Correction Technique Applied to
HOG and LBP Features” discusses techniques on emotion recognition using
gamma correction when applied to histogram of oriented gradient (HOG) and LBP
features. The book also includes diverse areas on images and speech signal pro-
cessing, besides the above-mentioned topics such as analysis of vocal tract
parameters of speech, 3D reconstruction of plant features with non-destructive plant
growth monitoring systems, digital image watermarking by fusion of wavelet and
curvelet transform, and content-based image retrieval (CBIR) techniques.
This book also features many recent advancements on machine learning algo-
rithms. Chapter “Automatic Gear Sorting Using Wireless PLC Based on Computer
Vision” discusses conversion of wired PLC into wireless PLC by interfacing the
PLC with the Wi-Fi module to enable real-time surveillance and control of the
system of equipment sorting via Wi-Fi module interfacing with PLC. Chapter
“Machine Learning Feature Selection in Archery Performance” discusses machine
learning feature selection in Archery performance using Boruta algorithm, Chapter
“Skin Lesion Classification Using Deep Learning” deliberates on skin lesion
classification using deep learning, Chapter “Deep Learning-Based Paperless
Attendance Monitoring System” deals with deep learning-based paperless atten-
dance monitoring system, and Chapter “Image Analytics to Detect Cigarette in an
Image Using Deep Learning” deals with image analytics to detect cigarette in an
image using deep learning.
In the domain of antenna design and communication, Chapter “Frequency and
Pattern Reconfigurable Antenna for WLAN and WiMAX Application” proposes an
innovative bow tie frequency and pattern reconfigurable antenna for WLAN and
WiMAX applications. Chapter “Design of a Power Efficient Multiband Patch
Antenna” gives a design consideration of multiband patch antenna. Chapter “A
Frequency Reconfigurable Antenna for Sub-GHz and TV White Space Applications”
discusses a frequency reconfigurable antenna for sub-GHz and TV white space.
Chapter “Comparative Analysis of Least Squares Method and Extended Kalman
Filter for Position Estimation in GPS Receiver” provides a comparative analysis of
position estimation techniques in a GPS receiver by using the least squares
(LS) method and extended Kalman filter method (EKF). Fair scheduling
non-orthogonal random access for 5G networks is presented in Chapter “Fair
Scheduling Non-orthogonal Random Access for 5G Networks.” Chapter “An
Improved Carrier Frequency Offset Estimation Under Narrowband Interference in
OFDM Cognitive Radio” surveys various techniques to estimate carrier frequency
offset (CFO) for OFDM cognitive radio.
Preface vii

This book has also included few interesting power-related chapters. Chapter
“Trends in Energy Management System for Smart Microgrid—An Overview”
reviews several energy management systems developed based on different strategic
approaches available for microgrid on demand-side management. Chapter
“Discontinuous PWM Techniques to Eliminate Over-Charging Effects in Four-
Level Five-Phase Induction Machine Drives” presents equivalent circuit modeling
of a Li-ion battery cell and its state of charge estimation using the Kalman filter
algorithm in MATLAB Simulink. A Transition Based Odd/Full Invert (TBO/FI)
coding scheme, which focuses on crosstalk avoidance and low dynamic power
consumption in NoC links, is also discussed. Chapter “Efficient Design of Drone
Flight Control Using Delay Tolerant Algorithm” presents a study to reduce the
human error parameter in the probable causes for drone crashes.
Various social issues are also presented by authors with their technical solutions.
In Chapter “IRIS: An Application for the Visually Impaired Using Google Cloud
API,” the authors present the design considerations of a cost-effective and efficient
visual aid which proposes a smart stick (IRIS) to help the user in obstacle detection
and navigation. Chapter “Implementation of Hand Gesture Recognition System to
Aid Deaf-Dumb People” considers a time system for hand gesture recognition that
acknowledges hand gestures and then converts them into text and voice. Statistical
validity of pre-smoking and post-smoking impact on heart rate variability among
middle-age men is presented in Chapter “Statistical Validity of Presmoking and
Postsmoking Impact on Heart Rate Variability Among Middle Age Men.” Chapter
“Analysis of Chronic Joint Pain Using Soft Computing Techniques” analyzes the
chronic joint pain remedies using soft computing techniques.
A critical evaluation of each submitted chapter by at least two expert reviewers
was carried out. The authors re-submitted with all suggested alterations given by the
expert panel. The book would definitely be of immense help to passionate
researchers, students and industry persons.

Mumbai, India S. N. Merchant

Pune, India Debashis Adhikari
Pune, India Krishna Warhade
Contents

Deep Semantic Segmentation for Self-driving Cars . . . . . . . . . . . . . . . . 1

Abhilash Sachin Kulkarni, Jyothi S. Nayak, Aditi Desai, Jahnavi Singh,
and Shraddha Murali
Trends in Energy Management System for Smart Microgrid—An
Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Harini Vaikund and S. G. Srivani
IRIS: An Application for the Visually Impaired Using Google
Cloud API . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Kajal Naik, Nikita Sawant, Gauravi Kamat, Siddhi Kandolkar,
and Niyan Marchon
Shot Boundary Detection Using Artificial Neural Network . . . . . . . . . . 45
Neelam Labhade-Kumar, Yogeshkumar Sharma, and Parul S. Arora
Custard Apple Leaf Parameter Analysis, Leaf Diseases,
and Nutritional Deficiencies Detection Using Machine Learning . . . . . . 57
Appasaheb Gargade and Shridhar Khandekar
Discontinuous PWM Techniques to Eliminate Over-Charging Effects
in Four-Level Five-Phase Induction Machine Drives . . . . . . . . . . . . . . . 75
J. Balakrishna, Teegala Bramhananda Reddy, and Marapu Vijaya Kumar
State of Charge Estimation Using Extended Kalman Filter . . . . . . . . . . 101
Kshitija A. Gaikwad and Vijaykumar Bhanuse
Frequency and Pattern Reconfigurable Antenna for WLAN
and WiMAX Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Anuradha A. Palsokar and Swapnil L. Lahudkar
Implementation and Analysis of Low Power Consumption Full Swing
GDI Full Adders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Deepgandha Shete and Anuja Askhedkar

ix
x Contents

Single Image Rain Removal Using Convolutional Neural

Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
P. Musafira and K. S. Shanthini
Ring Oscillator-Based Physical Unclonable Functions . . . . . . . . . . . . . . 147
Shruti Sakhare and Dipti Sakhare
A Robust Approach of Estimating Voice Disorder Due
to Thyroid Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Namrata V. Kanase, Satyajit A. Pangoankar, and Ashish R. Panat
Smart Glasses: Digital Assistance in Industry . . . . . . . . . . . . . . . . . . . . 169
Trupti Sutar and Savita Pawar
Implementation of Hand Gesture Recognition System
to Aid Deaf-Dumb People . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Supriya Ghule and Mrunalini Chavaan
Robust Underwater Animal Detection Adopting CNN with LSTM . . . . 195
Harishchandra Jagtap and Mrunalini Chavaan
Face Recognition Using Golden Ratio for Door Access
Control System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Prajakta S. Gaikwad and Vinayak B. Kulkarni
Efficient Design of Drone Flight Control Using Delay Tolerant
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Priyanka Dange and Bhairavi Savant
Adaptive Background Subtraction Models for Shot Detection . . . . . . . . 249
Dattatraya A. Jadhav, Yogeshkumar Sharma, and Parul S. Arora
Automatic Gear Sorting Using Wireless PLC Based
on Computer Vision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Yogesh Darekar and Smita Kulkarni
Statistical Validity of Presmoking and Postsmoking Impact
on Heart Rate Variability Among Middle Age Men . . . . . . . . . . . . . . . . 271
S. R. Rathod and C. Y. Patil
Transition Based Odd/Full Invert Coding Scheme for Crosstalk
Avoidance and Low Power Consumption in NoC Links . . . . . . . . . . . . 279
M. Vinodhini and N. S. Murty
Feature-Based Model for Landslide Prediction Using Remote
Sensing and Digital Elevation Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
Litesh Bopche and Priti P. Rege
Acoustic Classification of Bird Species . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Rashmika Patole and Priti Rege
Contents xi

Analysis of Chronic Joint Pain Using Soft Computing Techniques . . . . 321

Shailaja Suresh Patil and Shubhangi B. Patil
Emotion Recognition using Gamma Correction Technique Applied
to HOG and LBP Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
Vishal D. Bharate, Devendra S. Chaudhari, and Mayur D. Chaudhari
Scalable and Rapid Fault Detection of Memories Using MBIST
and Signature Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Midhun Sasikumar, Ramesh Bhakthavatchalu, K. N. Sreehari,
and Arjun S. Kumar
Analyzing Vocal Tract Parameters of Speech . . . . . . . . . . . . . . . . . . . . . 369
Sharada Vikram Chougule
Effect on Quality of 3D Model of Plant with Change in Number
and Resolution of Images Used: An Investigation . . . . . . . . . . . . . . . . . 377
Abhipray Paturkar, Gourab Sen Gupta, and Donald Bailey
Comparative Analysis of Least Squares Method and Extended
Kalman Filter for Position Estimation in GPS Receiver . . . . . . . . . . . . . 389
Jyoti S. Kavathekar and Ashwini M. Deshpande
Fair Scheduling Non-orthogonal Random Access for 5G Networks . . . . 405
Mansi N. Purohit and T. K. Ramesh
Analysis of Tree-Based Classifiers for Web Attack Detection . . . . . . . . . 421
Deshmukh Surbhi and Kshirsagar Deepak
Implementation of Random Pulse Width Modulation Techniques
for the Open-End Winding Five-Phase Motor Drives to Reduce
Acoustic Noise and Harmonic Distortion . . . . . . . . . . . . . . . . . . . . . . . . 429
J. Balakrishna, Teegala Bramhananda Reddy, and Marapu Vijaya Kumar
A Frequency Reconfigurable Antenna for Sub-GHz and TV White
Space Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Sanjeev Kumar, Rohit Khandekar, and Neela Rayavarapu
Human Activity Recognition Using Positioning Sensor and Deep
Learning Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Aarati Mohite, Priti Rege, and Debashish Chakravarty
A Discriminative Model for Multiple People Detection . . . . . . . . . . . . . . 491
Smita S. Kulkarni and Sangeeta Jadhav
Jal Sanchay—A Novel Approach for Water Usage Monitoring . . . . . . . 499
Ashish Srivastava, Mandar R. Nalavade, and Debashis Adhikari
xii Contents

An Algorithm for Skew Angle Estimation and It’s Application

Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Unnati Raju Kulkarni, Hemant Goraksh Ghuge, Revati Anand Kulkarni,
and Kirti Vasant Thakur
Analysis of Accuracy of Supervised Machine Learning Algorithms
in Detecting Denial of Service Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . 519
Deepa Krishnan
An Improved Carrier Frequency Offset Estimation Under
Narrowband Interference in OFDM Cognitive Radio . . . . . . . . . . . . . . 531
Vibha Patel and Krishna Warhade
Digital Image Watermarking by Fusion of Wavelet and Curvelet
Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
Jyotsna S. Gaikwad and Usha Verma
Machine Learning Feature Selection in Archery Performance . . . . . . . . 561
Uma Mahajan, Anup Krishnan, Vineet Malhotra, Deep Sharma,
and Sharad Gore
Skin Lesion Classification Using Deep Learning . . . . . . . . . . . . . . . . . . . 575
Aditya Bhardwaj and Priti P. Rege
Vehicle-to-Vehicle Driver Safety-Related Data Transmission
and Reception Using Li-Fi Technology . . . . . . . . . . . . . . . . . . . . . . . . . . 591
Snehal Pacharne and Vinayak Kulkarni
A Novel Approach for CBIR Using Four-Layered Learning . . . . . . . . . 607
Shweta Salunkhe, S. P. Gaikwad, and S. R. Gengaje
Design of a Power Efficient Multiband Patch Antenna . . . . . . . . . . . . . . 623
Punam Deotare and Debashis Adhikari
Brain Activity Analysis for Stress Recognition . . . . . . . . . . . . . . . . . . . . 635
Aishwarya Wakale and Usha Verma
Deep Learning-Based Paperless Attendance Monitoring System . . . . . . 645
Pallavi Derkar, Jitesh Jha, Mayuresh Mohite, and Rushikesh Borse
Image Analytics to Detect Cigarette in an Image Using
Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
Abhijeet Kharade, Kumar Abhishek, Debaraj Dwibedi, Siddharth Mehta,
Hemanth Meruga, Pratap Gangula, D. Narayana, and Rushikesh Borse
About the Editors

Dr. S. N. Merchant is currently a Professor (Emeritus Fellow) in the Department

of Electrical Engineering at IIT, Mumbai. He obtained his BTech, MTech and PhD
degrees from the department of Electrical Engineering, IIT Mumbai. His broad area
of research interest are wireless communication, wireless sensor network, signal
processing, multimedia communication and image processing. His noteworthy
contribution has been in solving state of the art signal and image processing
problems faced by the Indian Defense forces. He serves on the editorial board of
two international journals: International Journal of Ultra-Wideband Communication
& Systems and International Journal of Distributed Sensor Networks. He is a fellow
of Institution of Electronics and Telecommunication Engineers, IETE and is a
recipient of 10th IETE Prof SVC Aiya Memorial Award for his contribution in the
field of detection and tracking. He is also the recipient of the 9th IETE Prof SVC
Aiya Memorial Award for “Excellence in Telecom Education”. He is the winner of
2013 VASVIK Award in the category of Electrical & Electronic Sciences and
Technology.

Dr. Krishna Warhade is currently the Dean Research, Development, Innovation

and Consultancy and Professor and Head of the Department, department of
Information Technology at MIT College of Engineering and Dr. Vishwanath
Karad MIT World Peace University, Pune, India. He has obtained his BE
(Electronics) and ME (Instrumentation) from Shri Guru Gobind Singhji Institute of
Engg. & Tech., Nanded, under Swami Ramanand Teerth Marathwada University,
Nanded. He obtained his PhD from IIT, Mumbai. His research interests include
healthcare, video segmentation, wavelets, digital signal processing, digital Image
processing, digital filter design and ICT in agriculture. He has worked on various
collaborative projects with Bhaba Atomic Research Center (BARC), Blackberry
and Konkan Railway Corporation Limited (KRCL).

Dr. Debashis Adhikari is currently the Professor and Dean, School of Electrical
Engineering at MIT Academy of Engineering, Pune. He has obtained his BTech
from the Institute of Radio Physics & Electronics, University of Calcutta, ME and

xiii
xiv About the Editors

PhD from the Defense Institute of Advanced Technology (DRDO) (DU), Pune. His
broad area of research interest is wireless communication and antenna systems. He
has vast experience on surface to air missile systems, troposcatter communication,
SATCOM systems of the Indian Air Force. He has been actively involved in many
modification projects on SAM systems of the IAF. He is a fellow of IETE and is a
regular reviewer of many reputed international journals.
Deep Semantic Segmentation
for Self-driving Cars

Abhilash Sachin Kulkarni, Jyothi S. Nayak, Aditi Desai, Jahnavi Singh,

and Shraddha Murali

Abstract A self-driving car comprises three subsystems in the navigation, viz. lane
finding, urban scene understanding and geopositioning. This paper introduces the
technique of semantic segmentation for urban scene understanding with various
implementations in recent years and proposes a novel approach to fast and accurate
semantic segmentation. The architecture of the model is designed using VGG16 as
encoder, adopting hierarchical feature fusion technique to perform downsampling
followed by a lightweight decoder to perform upsampling. Skip connections are
introduced between the encoder and the decoder to improve the information flow. This
new approach outperforms the previous implementations by giving 94% accuracy
and 8.7 frames per second (FPS) simultaneously.

Keywords Semantic segmentation · FCN · CRFs · KITTI · VGG16

A. S. Kulkarni (B) · J. S. Nayak · A. Desai · J. Singh · S. Murali

Department of Computer Science and Engineering, BMS College of Engineering, Basavangudi,
Bengaluru, Karnataka, India
e-mail: [email protected]
J. S. Nayak
e-mail: [email protected]
A. Desai
e-mail: [email protected]
J. Singh
e-mail: [email protected]
S. Murali
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2021 1

S. N. Merchant et al. (eds.), Advances in Signal and Data Processing,
Lecture Notes in Electrical Engineering 703,
https://doi.org/10.1007/978-981-15-8391-9_1
2 A. S. Kulkarni et al.

1 Introduction

Semantic segmentation is a method of scene understanding in which classification

is performed on every single pixel of an image. Semantic segmentation is used
in autonomous vehicles to locate frontal objects such as roads, dividers, vehicles,
pavements, etc. It is a vital subsystem of the vehicle’s navigation system.
In the early days, deep neural networks were used for semantic segmentation.
This is termed as the classical method [1], and approaches based on neural networks
were used. Some of the networks include random forest classifier, condition random
fields (CRFs) and texton forest. CRFs, a sequential modelling algorithm, are used
for structured prediction problems. CRFs can identify and recognize patterns after
they are trained.
Another approach uses fully convolution networks [2] such as pretrained VGG or
ImageNet. FCNs have convolution layers that use kernels and perform computations
across all regions. The computation is divided into multiple regions and generates a
feature map. This feature map is then upsampled using deconvolution process. These
networks are used for segmentation, and another network is employed to perform
upsampling or deconvolution operation.
The aforesaid methods do not exploit the structure of the data. These methods tend
to establish a structure in the data and learn from them. Condition random fields or
CRFs [3] incorporate structure in the model. CRFs can be used to exploit the structure
in segmentation of automotive scenes and for post processing after segmentation.
The most widely used architecture is the use of dilated convolution after the
condition random fields network for considering contextual information. Another
popular approach for using structure in data is the spatio-temporal approach. In this
approach, the spatial and temporal information of the data is used to train the model.
This is most widely used in video segmentation. For example, the use of long short-
term memory (LSTMs) in fully convolution networks allow different layers to learn
at different rates controlled by clockworks involving small clock signals or clock
pulses [4]. The clockworks method is depicted in Fig. 1.

Fig. 1 Clockworks method

Deep Semantic Segmentation for Self-driving Cars 3

2 Related Work

This section discusses a few models of semantic segmentation:

2.1 SegNet

SegNet is a neural network that comprises an encoder, a decoder and a classification

layer that performs pixel-wise classification. The encoder has thirteen convolution
layers. Pretrained VGG16 is generally modified by replacing the fully connected
layers with fully convolution layers. The feature maps generated by the layers are
retained. For training, the model is initialized with the pretrained weights since
VGG16 is trained for classification problems on ImageNet data.
The encoder network is followed by a decoder network. The decoder network
has thirteen layers, and each layer corresponds to a layer in the encoder network.
Pixel-wise class probabilities are produced by feeding the decoder output directly
into a Softmax classifier. This produces a segmented image [5]. This is depicted in
Fig. 2.

2.2 DeepLab

There are various methods of performing semantic segmentation using DeepLab [6]
such as atrous convolution and atrous spatial pyramid pooling. Atrous convolution is
used in deep convolution neural networks (DCNNs) for computation of undecimated
wavelet transforms. Atrous convolution is used for its efficiency in computation. The
response of layers can be computed at any resolution and can be seamlessly integrated
with training and applied thereafter once trained.
Atrous spatial pyramid pooling is based on multiscale image representation tech-
nique and DCNNs. It has two approaches, viz. multiscale processing and spatial
pyramid pooling method which is widely used in recurrent-CNNs. Multiscale image

Fig. 2 SegNet architecture

4 A. S. Kulkarni et al.

representation can handle scale variability in semantics segmentation. R-CNN can

efficiently and accurately classify regions of arbitrary scale. The model can be trained
on datasets that contain objects of varying size.

2.3 RefineNet

Dilated convolution is a technique employed for faster convolution. Dilated convolu-

tion introduces zeros or holes in the convolution kernel between each pixel, thereby
significantly increasing the effective receptive field of the kernel. However, dilated
convolution is computationally expensive as it has to be applied on a huge number
of feature maps that are of high resolution involving volumes of memory. Compara-
tively, DeepLab performs the same computation with predictions one-eighth the size
of input image.
This approach uses ResNet-101 blocks for the encoder, while the decoder
comprises RefineNet blocks. The decoder concatenates high-resolution feature maps
generated by encoder and preceding RefineNet blocks [7]. The architecture of
RefineNet is depicted in Fig. 3.

2.4 Fast Semantic Segmentation

Real-time semantic segmentation mandates execution speed to be traded against

accuracy of segmentation. Fast semantic segmentation features a network that is

Fig. 3 RefineNet network and its components

Deep Semantic Segmentation for Self-driving Cars 5

structured as an encoder-decoder pair where the encoder has CNN to detect high-level
objects in the input image and the decoder takes this information from the output of
encoder and performs prediction for each pixel in the original input. Fire module and
dilated convolution [8] is featured in the model which helps in improving the accuracy
of the model. RefineNet module with ELU is also included in the architecture.

2.5 U-Net for Biomedical Imaging

U-Net is widely used in the field of biomedical image segmentation and was devel-
oped by Olaf Ronnieberger et al. U-Net is based on architecture by Ciresan et al.
[9] that uses sliding window setup for training the model. This setup allows for the
receptive fields to obtain the context of a region around the pixels by providing local
regions around the pixel as inputs. This redundancy results in a trade-off between
accuracy and context.
U-Net consists of an encoder and a decoder. The encoder contains four convolution
layers and a point-wise convolution layer, and the decoder is a standard decoder
network symmetric to the encoder network. Skip connections are introduced in the
network between the encoder and the decoder for improving the information flow
between them.

2.6 PSPNet

PSPNet aggregates different scales of global information by adding multi-scale

pooling on top of the backend model [10]. Bilinear interpolation is used for upsam-
pling, and 3 × 3 convolution fuses different levels of feature after concatenation.
Bilinear interpolation in PSPNet learns less number of parameters during training
as compared to deconvolution operation reducing the amount of data needed to train
the network.

3 Proposed Architecture

The proposed model will have an encoder-decoder architecture similar to U-Net.

The image is fed to the encoder block that performs downsampling or convolution
operation producing a feature map. This feature map is then fed into the decoder
block that performs upsampling or deconvolution operation and produces a pixel-
wise classification or a segmented image. The high-level architecture of the proposed
semantic segmentation network is depicted in Fig. 4.
The subsystems involved in the encoder and decoder blocks are as follows:
6 A. S. Kulkarni et al.

Fig. 4 High-level architecture

3.1 VGG16 (Encoder)

It is envisaged that the VGG16 network will form the backbone of the encoder block
[11] since it offers very high accuracy with low loss. In the proposed architecture, the
fully connected layers of the network are converted to fully convolutional network
(FCN) in VGG16. The third, fourth and seventh fully connected layers are replaced
with convolution layers with the convolution factorization principle resulting in the
split of the computation into multiple steps applying the hierarchical feature fusion
technique. These layers are rescaled for compatibility as a skip layer that introduces
skip connection. Point-wise convolution is then applied to the rescaled layers to
reduce the number of classes. The final layer of the VGG16 is a point-wise (1 ×
1) convolution, and the number of filters is equal to the number of masks, i.e., the
number of classes.

3.2 Hierarchical Feature Fusion

The proposed model will incorporate hierarchical feature fusion or HFF [12] tech-
nique, recently adopted in efficient spatial pyramid (ESP) block, to identify the
dominant features without increasing the computation cost and intensity. ESP block
uses reduce-split-transform method to handle downsampling. The reduce phase uses
point-wise convolution to reduce the number of features, hence reducing the compu-
tation cost of the next phase. The convolution operation is then split into multiple
operations where dilated convolution is applied with varying kernel sizes and holes.
This produces multiple feature maps. HFF is applied in the transform phase. HFF
concatenates the feature maps, generated by multiple dilated convolution operations
in the split phase, hierarchically to obtain a feature map with the dominant feature(s).
The use of hierarchical feature fusion in ESP block is depicted in Fig. 4.
Deep Semantic Segmentation for Self-driving Cars 7

3.3 Lightweight Decoder

The decoder employs a lightweight decoder network such as lightweight RefineNet

[13]. This network performs the deconvolution operation at a faster rate and provides
the desired speed in semantic segmentation.

3.4 Experimental Dataset

The following open-source datasets are available for training semantic segmentation
networks:
1. Cityscapes dataset for semantic urban scene understanding [14]
2. Berkeley DeepDrive Video dataset (BDDV)
3. KITTI dataset [15]
4. ImageNet dataset
5. PASCAL VOC dataset
6. MNIST.
The dataset used for the project is KITTI pixel-level semantic segmentation dataset
from KITTI Vision Benchmark Suite. The dataset contains training set and testing
set. The training set contains original images, segmented images and calibration
files. There are 289 original images and corresponding segmented images. Example
images are shown in Fig. 5.

3.5 Implementation

3.5.1 Modification to VGG16

VGG16 as Encoder
Layers 3, 4 and 7 (fully connected layers) are replaced with fully convolution layers.
L2 regularization (ridge regression) is used as a kernel regularizer, and the kernel is
initialized with normal initialization. The kernel size and stride is 1, and the number
of classes is 2. Padding remains the same.

Fig. 5 Original image and corresponding segmented image

8 A. S. Kulkarni et al.

Decoder
Upsampling layer (transposed_conv) is placed corresponding to the 3rd, 4th and 7th
layers of encoder. L2 regularization is used as kernel regularizer, and the kernel is
initialized with zeros. The kernel size is 4, and strides is 2. The number of classes is
2, and padding remains the same.

Skip Connections
The output of the upsampling layers is rescaled, and skip connections are added
between the encoder and decoder. The VGG layers are resized for compatibility as
skip layers. The skip connections are added between layer 3 of encoder and the corre-
sponding layer 3 of decoder and between layer 4 of encoder and the corresponding
layer 4 of decoder.

Generator Function
Generator function is used to supply the images to the model. The function gets the
path of the files in the dataset and loads the images randomly using cv2 package. The
function generates batches of the images and resizes the images to same dimensions.
The function returns an array with the images.

Optimization Function
The optimization function uses regularization loss and Softmax cross entropy loss
to calculate loss and Adam optimizer.

3.5.2 Incorporation of Hierarchical Feature Fusion in the Model

The fully connected layers in third, fourth and seventh layers of VGG16 are replaced
with fully convolution layers with dilation rate for dilated convolution. The convolu-
tion operation is split into multiple steps, and dilated convolution is performed with
dilation rate from 2 ton. The feature maps generated is then concatenated hierarchi-
cally, and another convolution operation produces the final feature map which is fed
into the next layer. The flowchart of the proposed architecture is shown in Fig. 6.

4 Experimental Setup and Results

After a critical assessment of the networks, U-Net has been chosen for this project
owing to its accuracy in biomedical image segmentation and the same has been
modified for our data. Semantic segmentation training was performed using KITTI
dataset. Salient outcomes are appended below.
Deep Semantic Segmentation for Self-driving Cars 9

Fig. 6 Flowchart of the

architecture

4.1 Architecture

The encoder has the standard U-Net architecture with multiple repeating convolution
layers with rectified linear unit (RLUs) and max-pooling. The image is downsampled
through these layers, and a feature map is generated.
The decoder uses the feature map generated by the encoder and performs upsam-
pling or deconvolution operation to obtain a feature mask. This feature mask is the
segmented image. The concatenation layer of the decoder only concatenates images
with identical length and width. Hence, concatenation layer is modified to handle
images with rectangular dimensions. The decoder architecture is depicted in Fig. 7.

4.2 Results

Training the model and its fine tuning resulted in the following: Hyperparameters
for training are summarized in Table 1.
Accuracy and loss at the end of training are summarized in Table 2.
10 A. S. Kulkarni et al.

Fig. 7 Training process and testing process

Table 1 Hyperparameter for

Hyperparameter Value
training U-Net
Validation split 10%
Number of epochs 20
Batch size 32
Optimization function ADAM

Table 2 Accuracies and

Training Validation
losses at the end of training
Accuracy 88.79% 86.02%
Loss 0.83 0.85

4.2.1 System Testing

Training System
The training system uses the VGG16 architecture as encoder and a lightweight
decoder and runs training for 25 epochs. The process of training is shown in Fig. 7.
In the training system, the model is loaded using the predefined configurations,
and model is compiled with the optimization function. Image generator is used to get
batches of images from the training dataset. These images (original and segmented)
are fed into the model in every epoch. The model is trained on the images for 25
epochs using the same images. After the final epoch, a trained model is obtained
with minimum loss, and the model and the weights are saved for testing.
The hyperparameters for training are summarized in Table 3.
Deep Semantic Segmentation for Self-driving Cars 11

Table 3 Hyperparameters for training

Hyperparameter Value
Learning rate 0.009
Batch size 32
Epochs 25
Optimization function ADAM

Table 4 Results of training against the hardware/platforms used

Platform/hardware Training time (min) Training accuracy Training losses
Amazon 9 0.96 0.15
Sagemaker
GCP 5 0.94 0.14
Lenovo Thinkpad 200 0.94 0.15

Each epoch runs training on 289 original images and 289 segmented images.
The training time, training accuracy and training losses are shown against the
hardware/platform used as shown in Table 4.

Testing System
The testing system uses the VGG16 trained model generated by the training system.
The testing is performed on a test dataset containing 290 non-segmented images.
The process of testing is shown in Fig. 7. In the testing system, the saved model is
loaded into the instance. Then, the saved weights are loaded from the trained model,
and the model is compiled. The image generator is used to make batches of images
from the testing dataset. These images are then fed into the model, and the output is
a set of segmented images.

4.2.2 Evaluation Metrics

The following evaluation metrics are used to evaluate the model during training:
1. Training Accuracy (TA): Accuracy obtained on the training set after every epoch.
2. Training Loss (TL): Loss calculated after every epoch on the training set.
3. Validation Accuracy (VA): Accuracy obtained on the validation set.
4. Validation Loss (VL): Loss (MSL) calculated on the validation set.
5. Training Time Per Epoch (TTPE): The time taken for training in one epoch.
6. Total Training Time (TTT): The total time taken for training. Table 5 shows the
results of evaluation of the metric on the model.
The following evaluation metrics are used to evaluate the model during testing:
1. Total Testing Time: Total time taken to run segmentation on images
12 A. S. Kulkarni et al.

Table 5 Result of metric evaluation on the model

Epoch TA TL VA VL TTPE (s)
1 0.61 1.049 0.53 0.943 508
5 0.73 0.677 0.69 0.775 489
10 0.79 0.420 0.77 0.501 482
15 0.85 0.218 0.79 0.230 486
20 0.91 0.165 0.89 0.181 481
25 0.95 0.140 0.92 0.149 482

Fig. 8 Cross-entropy loss function and loss function with the regularization function

2. Frames Per Second (FPS): The ratio of the total number of test images and the
total testing time. The higher the FPS, the faster the model.

The mean total testing time obtained was 33 s after running tests 15 times.
Hence, the speed obtained was approximately 9 FPS.
The improvement in training and validation accuracy was achieved by customizing
the optimization function and choosing an optimum batch size. The optimization
function calculates the total loss and uses ADAM optimizer to minimize the loss in
each step. This helps in increasing the training and validation accuracies at each step.

Loss Calculation
The loss calculation is performed by summing cross-entropy loss (Softmax loss),
and L2 norm (regularization) that are calculated based on the difference between the
actual probabilities and the predicted probabilities. The cross-entropy loss function
is shown in Fig. 8.
The loss function along with the regularization function is shown in Fig. 8.

4.3 Performance Analysis

For analyzing performance, accuracy, total training time, total testing time, FPS and
size of the model are considered. The results are shown in Table 6.
Modified VGG gives 94% accuracy compared to 86% of U-Net and 95% of
PSPNet, but it gives 8.7 FPS compared to 3.7 of U-Net and 4.4 of PSPNet. The
modified VGG16 outperforms U-Net and PSPNet in terms of FPS, but there’s a
Deep Semantic Segmentation for Self-driving Cars 13

Table 6 Results of three semantic segmentation networks

Model Accuracy (approx. Total training time Total testing time FPS Size (MB)
%) (25 epochs, min) (290 images, s)
Mod. VGG16 94 29 33 8.7 956
U-Net 86 21 78 3.7 128
PSPNet 95 24 66 4.4 398

heavy price to pay in terms of its size. The model performs accurately on the test
images and provides accurate mask for the same.

5 Conclusion

The application of semantic segmentation systems that can perform at higher speeds
without compromising accuracy is an imperative in autonomous vehicles. The
encoder-decoder network as implemented in this paper utilizing VGG16 with HFF
and lightweight decoder is expected to provides the requisite speed and accuracy for
real-time deployment in autonomous vehicles.

References

1. Mennatullah S, Sara E, Martin J, Senthil Y (2017) Deep semantic segmentation for automated
driving: taxonomy, roadmap and challenges
2. Jonathan L, Evan S, Trevor D (2015) Fully convolutional networks for semantic segmentation
3. Guosheng L, Chunhua S, Anton VDH, Ian R (2016) Efficient piecewise training of deep
structured models for semantic segmentation
4. Mohsen F, Mohammad HS, Mohammad S, Mahmood F, Reinhard K, Fay H (2016) STFCN:
spatio-temporal FCN for semantic video segmentation
5. Vijay B, Alex K, Roberto C (2017) SegNet: a deep convolutional encoder-decoder architecture
for image segmentation
6. Liang-Chieh C, George P, Florian S, Hartwig A (2017) Rethinking atrous convolution for
semantic image segmentation
7. Guosheng L, Anton M, Chunhua S, Ian R (2017) RefineNet: multi-path refinement networks
for high-resolution semantics segmentation
8. Simon J, Michal D, David V, Adriana R, Yoshua B (2017) The one hundred layers tiramisu:
fully convolutional densenets for semantic segmentation
9. Ciresan DC, Gambardella LM, Giusti A, Schmidhuber J (2012) Deep neural networks segment
neuronal membranes in electron microscopy images
10. Hengshuang Z, Jianping S, Xiaojuan Q, Xiaogang W, Jiaya J (2017) Pyramid scene parsing
network
11. Karen S, Andrew Z (2014) Very deep convolution networks for large-scale image recognition
12. Sachin M, Mohammad R, Anat C, Linda S, Hannaneh H (2018) ESPNet: efficient spatial
pyramid of dilated convolution for semantic segmentation
13. Vladimir N, Chunhua S, Ian R (2018) Light-weight refinenet for real-time semantic segmen-
tation
14 A. S. Kulkarni et al.

14. Marius C, Mohamed O, Sebastian R, Timo R, Markus E, Rodrigo B, Uwe F, Stefan R, Bernt
S (2016) The cityscapes dataset for semantic urban scene understanding
15. Andreas G, Philip L, Christoph S, Raquel U (2013) Vision meets robotics: the KITTI dataset
Trends in Energy Management System
for Smart Microgrid—An Overview

Harini Vaikund and S. G. Srivani

Abstract Smart Grid (SG) is an integrated grid to improve efficiency, consistency,

and security with the incorporation of conventional and renewable energy sources,
through automated control and modern communication technology. The development
of Microgrid (MG) is a great solution for the incorporation of sustainable energy
resources inside the smart grid environment. The emergence of Microgrid (MG) by
decomposition of the grid is a combination of Distributed Energy Resources (DERs),
Energy Storage System (ESS), loads and Control devices. This makes MG a single
and controllable power supply system that can enhance versatility, dependability
and furthermore incorporate the benefits of distributed generation. In MG, an EMS
is fundamental for the ideal utilization of distributed energy sources in smart, secure,
reliable and synchronized ways. The need for understanding the energy utilized is
increasing since effective Energy Management is more challenging in Microgrid
(MG). This paper reviews several Energy management systems developed based on
different strategic approaches available for Microgrid on Demand-Side Management.

Keywords Microgrid · Energy management system · Demand-side management

1 Introduction

A Microgrid is a cutting edge distributed power system utilizing local sustain-

able energy sources designed through different smart grid initiatives. Renewable
power resources like wind, solar, microturbines, latest generation technologies like
combined heat and power (CHP) technology and fuel cell technologies become part
of a Microgrid. Renewable energy resources normally have intermittency problems.
To solve the intermittency problem electric vehicle technology, flywheel storage

H. Vaikund (B)
Department of Electrical and Electronics, Dr. Ambedkar Institute of Technology, Bengaluru, India
e-mail: [email protected]
S. G. Srivani
Department of Electrical and Electronics, R V College of Engineering, Bengaluru, India
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2021 15

S. N. Merchant et al. (eds.), Advances in Signal and Data Processing,
Lecture Notes in Electrical Engineering 703,
https://doi.org/10.1007/978-981-15-8391-9_2
16 H. Vaikund and S. G. Srivani

systems and Storage devices like battery energy storage systems can be used [1].
Due to their better coordination and control microgrid is preferred compared to the
distributed generation sources. MG can be operated without the presence of the main
grid (called as an islanded mode) so providing security for a local community. If it
is connected to the grid it is grid-tied mode.
Three significant objectives of MG is
• Reliability—Cyber, physical
• Sustainability—environmental considerations
• Economics—efficiency, cost optimizing [2]. It is capable of operating with grid
and/or without the grid (island mode). Figure 1 shows a MG architecture consisting
of various DERs, critical and responsive loads. The main grid and the MG are
linked through Point of Common Coupling (PCC). MG’s advantage is, during
grid-connected mode—power trading with the main grid is done and system
stability shifts to the islanded mode when there are disturbances in the main grid.
Microgrid Central Controller (MGCC) and Local Controllers (LCs) controls and
coordinates the whole MG operation.
There are different classifications of MGs based on
1. Power type—AC and DC
2. Supervisory Control—Centralized and Decentralized
3. Operation Mode—Grid-connected and islanded
4. Phase—Single and Three Phases
5. Application—Residential/Commercial/Industrial and Utility/Municipality/
Military.
A few focal points of MGs are a decrease in Green House Gases (GHG) outflows,
voltage level improvement, power supply decentralization, Demand Response (DR)
and incorporation of cogeneration. It additionally lessens losses in line and blackouts
in transmission and distribution (T&D) systems [3, 4]. On different operating modes,
the MGs to protect the grid, managing the load connected to the system and the

Fig. 1 Microgrid architecture

Trends in Energy Management System for Smart … 17

voltage and frequency adjustment is a requirement so resynchronization process and

facilitation of the generation-side and load-side management is necessary.
Some restrictions of MGs are a high upfront cost of RERs, efficient use of power
sources, problems due to control of MGs and non-availability of protected and admin-
istrative standards, and client security. Solving EMS problems is the key focus
for researchers because of the non-availability of continuous RERs and increased
integration of controllable loads [3].
In this paper, the Energy Management System (EMS) and the need for it is
presented in Sect. 2. Demand-Side Management (DSM), necessity and different opti-
mization techniques based on many mathematical optimization techniques which are
used to solve the strategies are reviewed and discussed in Sect. 3. Finally, the paper
is concluded in Sect. 4.

2 Energy Management System (EMS)—Introduction,

Architecture, and Strategies

An energy management system (EMS) monitors, controls, optimizes the performance

of the transmission and generation system and also saves overall operating cost in
electric utility grids [5]. EMS is defined as “a computer system comprising a soft-
ware platform providing basic support services and a set of applications providing
the functionality needed for the effective operation of electrical generation and trans-
mission facilities to assure adequate security of energy supply at minimum cost” [3].
Energy management is a venture to control all kinds of energy.
A basic illustration of MG EMS is shown in Fig. 2. The microgrid EMS consists
of some components. Modules for efficient execution of decision-making strategies
are carried out by Human Machine Interfaces (HMI), DERs/load forecasting and
Supervisory Control and Data Acquisition (SCADA) modules by transferring optimal
decisions to the storage, generation and load units. Optimizer to determine the optimal
power flow trajectories and for electricity demand use of forecasting modules and

Fig. 2 An illustration of a microgrid energy management system

18 H. Vaikund and S. G. Srivani

renewable energy sources power production are the main components of MG [6].
The local controller has the following functions—supervising the renewable energy
sources (RES), energy storage (ES) and based on trajectories set by the optimizer
collect the measurement data by activating the actuators.
In MG the main challenge is the reduction of energy balances caused due to
the dynamic nature of electricity consumption and uncertainties in energy supply
from renewable-based Distributed Generators (DGs). Small size imbalances can be
handled by droop control or frequency control. In large supply-demand imbalances,
these methods fail and which necessitates the need for the development of energy
management strategies for microgrids [7]. These procedures give numerous advan-
tages to live energy-saving, frequency regulation, low cost reduction, GHG emission
reduction, and client privacy.
EMS of a MG incorporates both Demand-side and Supply-side Management,
while fulfilling system constraints, to realize a reliable, cost-effective and sustain-
able working of MG. Scheduling DERs and loads, losses and system outages mini-
mization, control of irregularity and unpredictability of RERs are the strategies of
MG.
To achieve proficient and optimal operation of MG based on mathematical tech-
niques various approaches [3, 8, 9] many researchers have been using to solve energy
management strategies. An overview of these solution approaches and control strate-
gies is described in the following sections. Table 1 gives some of the EMS based on
strategic approaches used to solve the strategies.

3 Demand-Side Management

The expression “demand-side management” (DSM), also known as “energy demand

management,” stands for a various energy consumption related work. It incorporates
the controls and alteration of energy use (e.g., energy preservation, efficiency, and
storage) but also the practices that are engaged with the procedures [10].
Demand management refers to any strategy that reduces energy consumption or
remodels the use of energy on the customer’s side of the electricity grid as an alterna-
tive to increase supply capacity. Demand management strategies may include preser-
vation programs, shift tasks, and an increase in the strategic task. Load control is a
feature of demand-side management. The process of transforming conventional MG
into the green system is facilitated by a powerful tool called Demand-side manage-
ment (DSM). DSM is a cost-effective choice for balancing the energy imbalances
caused by unreasonable utilization of power by adjusting the energy utilization to
the supply. The main focal point of DSM strategies is moving the energy demand
from peak hours to off-peak hours to lessen the peak demand [7].
A vital role is played by DSM for the development of industrial power, planning
of energy and protection of the environment. In DSM there is a concept of Demand
Response (DR), which refers to “changes in electric usage by end-use customers from
Trends in Energy Management System for Smart … 19

Table 1 EMS based on different strategic approaches

Different methods Approaches
Energy management system Classical method • Linear programming
based on • Mixed integer linear
programming
• Nonlinear programming
• Mixed nonlinear programming
• Dynamic programming and
rule-based methods
Meta-heuristic approach • Genetic algorithm
• Particle swarm optimization
• Differential evolution
• Ant colony optimization
• Gravitational search algorithm
• Modified bacterial foraging
• Artificial bee colony
• Tabu search
• Imperialist competition
algorithm
Artificial intelligence • Fuzzy logic
methods • Neural networks
• Multi-agent system
• Game theory
• Markov decision process
• Adaptive intelligence technique
Stochastic and robust • Stochastic optimization
programming approaches • Robust optimization
• Chance constrained
programming
• Scenario-based optimization
Model predictive control • Model predictive control
Other methods • Rolling horizon optimization
• Hierarchical control
• Homeostatic control
• Predictor corrector proximal
multiplier algorithm
• Mesh adaptive direct search

their normal consumption patterns in response to changes in the price of electricity

over time” [8].
Electricity Market can have the following advantages by the introduction of DSM
[10]:
• Supply and demand information trading can be done instantly, transparent trans-
actions can be facilitated and rapidly improve the development of an electricity
price mechanism.
• The efficient market operation and good control of power of market is possible.
20 H. Vaikund and S. G. Srivani

• Improve power system reliability by addressing peak hours demand congestion

problems.
• Lessen the cost of power generation, transmission, and distribution.
• A new diagnosis for Energy management and emission reduction can be
facilitated.
For improving energy efficiency by balancing supply and demand the combined
work of advanced DSM and distributed energy resources (DERs) in a MG along
with a good communication plan can play a vital role. The DSM by using developed
optimal methods and information technologies can achieve a reduction in energy
consumption, energy handling efficiency enhancement and cost reduction.
Due to the heterogeneous nature of the MG network which comprises of ESS,
diesel generators, solar power, wind turbines, and electric vehicles, the specific objec-
tive function need to be defined, unlike the classical power grid where optimization is
based on the centralized objective function. To control and enhance SG systems that
include a MG distribution network it is natural to implement distributed analytical
methods. In this regard, a few open issues in microgrids can be dealt with utilizing
various strategies [11].

3.1 Classical Methods

An optimization method was proposed in [12] for the DSM EMS based on Linear
Programming (LP) approach. EMS is for a given customer’s hourly electricity prices.
The proposed EMS which is developed using optimization algorithm based in LP,
the customer can use their own plan to
• Control load and prices in the energy distribution system
• Forecast future energy use
• Efficiency improvement
• Losses reduction.
During the demand for energy excess energy which is stored in the battery can be
utilized. In [13] focus is on both DSM and SSM load scheduling problem for users
in SG using LP models to
• Minimize energy cost in DSM
• Maximize the load factor in SSM.
For the operation of intended appliances, the LP model offers various flexibilities
to preset if the appliance has to operate without interruption and to define multiple
time intervals. Uncertainty aspect of the MG system with RERs is addressed by
designing and experimentally testing an adaptive online MG EMS in [14]. Interfacing
with controller and incorporation with other modules and communication between
them is tested using the designed EMS architecture. Optimization module is based
on Mixed Integer Linear Programming (MILP).
Trends in Energy Management System for Smart … 21

3.2 Meta-Heuristic Approach

To solve the objective function the interior point method and for demand response,
particle swarm optimization and artificial immune systems are applied in [15]. For
both customer and utility in the MG environment the DR management proposed, a
service provider who carries out the optimization decides a common solution. It has
the following features
• For energy balance problems application of DR techniques
• Decrease the revelation of the supplier to the market instability by improving the
cost of both supplier and consumer
• Utilized newest and extensively used realistic tools
• Easy adaptation in the energy industry to develop real-world applications.
The main focus of [16] is to develop a microgrid generation scheduling model
using the intelligent meta-heuristic algorithm. Here a modified cuckoo search algo-
rithm is developed and utilized for EMS in microgrids. The results of the studies
conducted for
• Reducing operating costs with and without DR participation.
• Wind and PV resources uncertainties.
The results show that the operating costs are reduced by adaptive demand response
programs.
Various heuristic techniques like Particle Swarm Optimization (PSO), Firefly
Algorithm (FA), and Artificial Bee Colony (ABC) are applied in [17] for optimal
consumption of renewable energy resources (RES). To incorporate increased energy
demand at reduced cost batteries are used. Comparing these techniques ABC turns
out to be very effective.
In paper [18], a combined solution using a genetic algorithm for residential loads
for both economic dispatch and DSM in a MG is presented. The evaluations show
that the given approach can
• Reduce the cost of operation for both suppliers and consumers in a single and
multiple-facility microgrid.
• Cost generation reduction.
• Reduction in shifting of loads inconvenience.
The objective function of [19] containing discontinuous functions uses a Genetic
Algorithm (GA) based solution for savings for the customer with DSM. For Real-time
Microgrid Power Management problem two computational intelligence methods,
particle swarm optimization (PSO) and Ant Colony Optimization (ACO) were intro-
duced in [20]. A mathematical framework or multi-objective and multi-constraint
optimization was presented and the advantages of intelligent methods over traditional
computational techniques for optimization were discussed.
22 H. Vaikund and S. G. Srivani

3.3 Artificial Intelligence Methods

A home automation economic model that has integration with the residential
DERs, dynamic tariffs and enables actions to DR based on the Internet of Things
was proposed in [21]. Decision making is based on the application of Artificial
Intelligence (Fuzzy Logic) for automatic load management.
In paper [22] two Genetic Algorithm (GA) executed alternately, one for deter-
mining the microgrid scheduling and the fuzzy rules and others to tune the member-
ship functions for optimizing microgrid operation. By using a hybridized Fuzzy
and GA algorithm a MG generation schedule, day ahead wind generation electricity
prices and based on load demand allocation of storage power is developed.
In [23, 24] a novel DSM method using dynamic game theory is proposed which
can aid in
• Effective use of DERs by analyzing and coordinating the interactions among users
[23].
• Reduce the total energy cost [23].
• Be modeled for variations in wind power using the Markov chain [23].
• Saves the cost of generation using DSM [24].
• Reduces the Peak-to-Average ratio [24].
A DSM framework integrating smart fuzzy load controller and DR in MG is
proposed in [25]. The Fuzzy load controller for load shifting and load usage uses the
data procured by the smart load monitoring system to make decisions. This method
can
• lessen peak demand
• to minimize energy loss
• increase the efficiency
• effective cost saving.
In [26] a load management system for highly stochastic loads (treated as Markov
models) is proposed. Markov decision process is used to
• reduce the overall cost linked with DR control action.
• increase the certainty of fulfilling the DR.
• reduce the load by modeling the load using each household as Markov chains.
• cost saving compared with the industry used model.
The key focus of an agent-based EMS proposed in [27] is to
• aid trading of power among microgrids with distributed storage and DR.
• make use of energy accessibility from the DERs and diversity in load usage
patterns of the customers.
• DR in reducing the peak demand.
• to reduce electricity costs.
Trends in Energy Management System for Smart … 23

In paper [28] based on the Lagrangian multiplier method, Lagrange programming

neural network is introduced for
• economic dispatch optimization
• reduction in the objective function
• functional cost reduction and
• increase in RERs power generation.
The optimal solutions for power production resources and ESS can be determined
using this approach. Neural network helps in achieving prediction of renewable
resources and load demand a day ahead.

3.4 Stochastic and Robust Programming Approaches

Stochastic model predictive control scheme energy scheduling for optimal EM, the
supply and demand-side uncertainties are taken into consideration in [29]. In this
solution is found efficiently because
• Uncertainties can be handled.
• The energy trading between MG and main grid is found to be in a assigned
trajectory.
• References tracking including uncertainties have considerable improvement than
the traditional scheduling scheme.
In [30] microgrid EMS framework based on agent-based modeling by introducing
Robust Optimization (RO) is proposed. Uncertainties can be handled using the exten-
sion of the framework. Evaluation of each uncertainty impact on the cost accuracy
and revenue function is made possible in this proposed optimization framework. An
increase in the reliability indicators and reduction in energy shortage is evident from
the reliability analysis.

3.5 Model Predictive Control

For MG optimization neural network load forecasting and model predictive control
(MPC) are implemented in the paper [31]. The algorithm updates the optimal course
at each stage for
• energy balance
• operational cost reduction
• reducing load forecasting errors.
24 H. Vaikund and S. G. Srivani

A 2 stage Real-time demand-side management (RDSM) method using MPC based

optimization model for Response Executers (RE) under an environment with uncer-
tainties consist of the ESS and the Plug-in electric vehicles (PEV) is proposed in
[32]. With the proposed framework MG can achieve
• economic benefits
• better dealing with uncertainty
• improve net load characters
• enhancing energy balance.
The analysis of MG DSMs based on different approaches, their contributions and
limitation are given in Table 2. Further, this review will pave the way for developing a
Demand-Side Management System considering the issues caused due to conventional
generators, the privacy of consumers, incorporation of demand response, reliability,
battery status and losses in the system.

4 Conclusions and Scope

Due to various challenges and opportunities, Microgrid penetration is presently rising

across the globe. This paper attempted to provide a review of the various microgrid
demand-side energy management approaches. The review consists of an introduction
to microgrids, their components and associated benefits, limitations, which further
followed by a discussion on Energy Management systems, its architecture, advan-
tages, and constraints. The process of transforming conventional MG into the green
system is facilitated by a powerful tool called Demand-side management (DSM).
These MG DSMs mainly focuses to reduce the energy cost, utilization of renew-
able energy resources, reduction of MG system losses and energy trading with the
main grid. Even though many efforts are taken for developing DSM algorithms
for Intelligent Energy Management, the following challenges have to be met in the
microgrid system: integration of demand response, the robustness of power manage-
ment, losses cost of MGs and system reliability. The choice of approaches which
is discussed in this paper to be implemented for optimization depends on various
factors like uncertainties caused due to renewable energy resources, integration of
demand response, computational time complexity, operational cost of battery and
depth of discharge of battery of in energy storage systems and many more. Artificial
Intelligence promises to be more effective and standalone approach to address the
issues related to demand-side management while considering the different factors
mentioned above. The benefits of Artificial intelligence on Microgrid are worth for
future efforts.
Trends in Energy Management System for Smart … 25

Table 2 Contributions and limitation of MG DSM based on different approaches

References Approach Contributions Limitations
[12] Artificial neural network Losses reduction, optimal Demand response was
based linear use of resources on not considered which
programming demands, energy balance required external control
flexibility techniques
[13] Linear programming Reduction in demand-side Integration of 2 LP
(LP) scheduling cost and problems into one model
improve the load factor not considered
while supply-side
scheduling
[14] Mixed integer linear Operational cost reduction, Advanced forecast,
programming able to operate on different testing and optimization
modes not implemented
[33] Artificial immune Cost improvement by The integration of
systems and particle using the DR technique for renewable sources is
swarm optimization both utility and consumer ignored
[16] Modified cuckoo search Wind and PV resources Small scale power
algorithm uncertainty are considered network considered
and the objective function
is analyzed. The operating
costs reduced by adaptive
demand response programs
[17] Artificial bee colony Optimal use of RERs, Solar power not
(ABC), particle swarm reducing the cost of considered
optimization (PSO), and production and
firefly algorithm (FA) market-clearing price
[18] Genetic algorithm Reduction in energy cost Constraint-capturing
mechanisms and the
larger MG level not
considered
[19] Genetic algorithm Generation cost is more MG for a large area not
without DSM compared to considered
with DSM
[21] Fuzzy logic Residential DER operation Computational
cost-benefit ratio complexity not discussed
improvement
[22] Fuzzy logic Management of power in Non-consideration of loss
the battery and scheduling of power and time
of MG complexity for higher
computation
[23] Game theory Entire energy cost Uncertainties not
reduction in the isolated considered
microgrid system
(continued)
26 H. Vaikund and S. G. Srivani

Table 2 (continued)
References Approach Contributions Limitations
[24] Game theory Benefitting the grid by Hybrid micro-grid is not
reducing the ratio of peak considered for the study
to average and reduction in
load profile are
smoothened which are
caused by supply
constraints
[25] Fuzzy logic Reduction in cost of More renewable energy
energy, loss of power, peak sources not considered
demand and inefficiency of
the grid
[26] Markovian model Reaching demand response Computational
targets as well as reduction complexity not discussed
in cost of operation
[27] Multi-agent system This EMS increases levels Microturbine emission
of thermal comfort and cost is ignored.
consumer’s electrical and Non-consideration of the
reduces the operational operational cost of the
cost of MG battery
[28] Neural network Minimization of the overall Non-consideration of
cost of MG, a neural battery operation cost and
network used for complexity of the time of
forecasting and the results computation
are compared with PSO
[29] Stochastic programming Minimization of CG Discussion of complexity
battery cost of operation, due to time in
trading of energy cost and computation is not
cost of degradation with covered
main grid
[30] Robust programming Realization of performance DR not included in a
of MG (imbalance cost), residential district.
reliability of MG (loss of Complex
expected energy) and load
expectation loss
[31] Model predictive control Static control issues Compensating residual
addressing and elimination forecast errors not
of load forecast errors and discusses
reduction in operations
costs
[32] Model predictive control Power balance Computational
maintenance and the complexity not discussed
operation cost reduction
Trends in Energy Management System for Smart … 27

References

1. Khan B, Singh P (2017) Selecting a meta-heuristic technique for smart micro-grid optimization
problem: a comprehensive analysis. IEEE Access 5
2. Bayindir R et al (2014) A comprehensive study on microgrid technology. Int J Renew Energy
Res 4(4)
3. Zia MF, Elbouchikhi E, Benbouzid M (2018) Microgrids energy management systems: a critical
review on methods, solutions and prospects. Appl Energy
4. Parhizi S et al (2015) State of the art in research on microgrids: a review. IEEE Access 3
5. Borase PB, Akolkar SM (2017) Energy management system for microgrid with power quality
improvement. In: International conference on microelectronic devices, circuits and systems
(ICMDCS). IEEE
6. Kowalczyk A et al (2016) Microgrid energy management system. In: 21st international
conference on methods and models in automation and robotics (MMAR). IEEE
7. Kumar Nunna HSVS et al (2016) Multi-agent based demand response management system for
combined operation of smart microgrids. Sustain Energy Grids Netw 6:25–34
8. Khan AA et al (2016) A compendium of optimization objectives, constraints, tools and
algorithms for energy management in microgrids. Renew Sustain Energy Rev 58:1664–1683
9. Ma J, Ma X (2018) A review of forecasting algorithms and energy management strategies for
microgrids. Syst Sci Control Eng 6(1):237–248
10. Li D, Chiu WY, Sun H (2017) Demand side management in microgrid control systems. In:
Microgrid advanced control methods and renewable energy system integration, pp 203–230
11. Saad W et al (2012) Game theoretic methods for the smart grid. IEEE Signal Process Mag
12. Loganathan N, Lakshmi K (2014) Demand side energy management system using ANN based
linear programming approach. In: IEEE international conference on computational intelligence
and computing research
13. Tan W-N et al (2016) Optimization models for demand-side and supply-side scheduling in
smart grids. In: IEEE 16th international conference on environment and electrical engineering
(EEEIC). IEEE
14. Luna AC et al (2016) Online energy management systems for microgrids: experimental
validation and assessment framework. IEEE Trans Power Electron
15. Rat C et al (2018) Microgrid supply management. Procedia Soc Behav Sci 238:46–52
16. De M et al (2018) A reliable energy management and generation scheduling model in micro-
grids using modified cuckoo search algorithm. In: Emerging trends in electronic devices and
computational techniques (EDCT). IEEE
17. Paliwal NK et al (2016) Demand side energy management in hybrid microgrid system using
heuristic techniques. In: IEEE international conference on industrial technology (ICIT). IEEE
18. Arif A et al (2014) Integrating renewable economic dispatch with demand side management
in micro-grids: a genetic algorithm-based approach. Energy Effic 7:271–284
19. Jayadev V, Shanti Swarup K (2013) Optimization of microgrid with demand side management
using genetic algorithm. In: IET conference on power in unity: a whole system approach. IET
20. Nehrir MH et al (2010) Towards real-time microgrid power management using computational
intelligence methods. In: IEEE PES general meeting
21. Souza AT et al (2018) Development of a demand response system integrated to photovoltaic
microgeneration and energy storage using IoT and artificial intelligence. In: 15th international
conference on the European energy market (EEM). IEEE
22. Fossati JP et al (2015) Optimal scheduling of a microgrid with a fuzzy logic controlled storage
system. Electr Power Energy Syst 68:61–70
23. Wu C et al (2011) Demand side management for wind power integration in microgrid using
dynamic potential game theory. In: IEEE international workshop on smart grid communications
and networks
24. Noor S et al (2018) Energy demand side management with supply constraints: game theoretic
approach. Energy Procedia 145:368–373
28 H. Vaikund and S. G. Srivani

25. Zunnurain I et al (2018) Implementation of advanced demand side management for microgrid
incorporating demand response and home energy management system. Infrastructures 3:50
26. Thomas E et al (2019) Towards demand side management control using household specific
Markovian models. Automatica 101:450–457
27. Kumar Nunna HSVS et al (2013) Energy management in microgrids using demand response
and distributed storage—a multiagent approach. IEEE Trans Power Deliv
28. Wang T et al (2017) Neural networks for power management optimal strategy in hybrid
microgrid. Neural Comput Appl
29. Kou P et al (2016) Stochastic energy scheduling in microgrids considering the uncertainties in
both supply and demand. IEEE Syst J
30. Kuznetsova E et al (2014) An integrated framework of agent-based modeling and robust
optimization for microgrid energy management. Appl Energy 129:70–88
31. Jaboulay P-A et al (2017) Real-time energy management optimization using model predictive
control on a microgrid demonstrator. In: IEEE international conference on energy internet
(ICEI). IEEE
32. Yang X et al (2017) Real-time demand side management for a microgrid considering
uncertainties. IEEE Trans Smart Grid
33. Herath P et al (2018) Computational intelligence based demand response management in a
microgrid. IEEE Trans Ind Appl Comput
IRIS: An Application for the Visually
Impaired Using Google Cloud API

Kajal Naik, Nikita Sawant, Gauravi Kamat, Siddhi Kandolkar,

and Niyan Marchon

Abstract The visually impaired struggle to walk safely without having any human
guidance. They face several problems in their lives which include identifying the
hindrances when they are walking. Traveling from one place to another is a tedious
task for them requiring the help of other people. White canes and guide dogs are
usable but are not fully reliable. Due to this disability, they have to go through a lot
of hardships to conquer over the problems in navigation. Thus there is a need for a
smart solution which would overcome the challenges faced. Focus is on designing
one kind of a visual aid which is cost-effective and efficient. In this context, we
propose a smart stick (IRIS) to help the user in obstacle detection and navigation.
A conventional PVC pipe forms the stick on which ultrasonic sensors, camera, GPS
module, GSM Module which are all interfaced with Raspberry Pi are mounted at
different positions to detect and identify obstacles in the path of the user. Google
Cloud Vision API recognizes the objects in the surroundings and delivers an audio
signal about the same via earphones. The sense of sight of the user is now achieved
by the sense of hearing. It also uses a GSM Module and a GPS module to contact
the user’s relatives in emergency situation by sending them the user’s location. The
whole device’s design ensures independence to the user.

Keywords Ultrasonic sensors · GPS module · GSM module · Raspberry Pi ·

Google Cloud

K. Naik · N. Sawant (B) · G. Kamat · S. Kandolkar · N. Marchon

Padre Conceicao College of Engineering, Verna, Goa, India
e-mail: [email protected]
K. Naik
e-mail: [email protected]
G. Kamat
e-mail: [email protected]
S. Kandolkar
e-mail: [email protected]
N. Marchon
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2021 29

S. N. Merchant et al. (eds.), Advances in Signal and Data Processing,
Lecture Notes in Electrical Engineering 703,
https://doi.org/10.1007/978-981-15-8391-9_3
30 K. Naik et al.

1 Introduction

Blindness is the inadequacy of a person to perceive the things around his/her

surrounding with the eyes. According to the World Health Organization, it is globally
estimated that about 285 million people live with partial or complete blindness [9].
Currently the visually impaired use a white cane or a guide dog to find out if any
obstacles are present in front of them. But these are inefficient in various aspects
and the person using it has to face several problems. In an unknown environment,
the visually impaired might get confused and this will restrict their movement. It
becomes difficult for them to do their routine activities. The proposed smart stick
can help them in moving freely and allowing them to carry out their work with ease.
Using existing technologies we deliver a better result to overcome the disadvantages
of the earlier system. A cost-effective system is necessary which can be used by
them to work smoothly and comfortably. Technology progress should be reflected in
such cases, providing a smart solution solving cost and efficiency issues. Billions of
visually impaired across the World suffer from daily problems that could risk their
lives. This smart stick is a perfect and secure solution. Our focus is to bring out the
best of technology and to help the visually impaired to live independently. In this
fast growing technology globe, where people strive to live independently, this project
proposes a smart stick (IRIS) to support them and help them to navigate with ease.
Efficiency is the main priority, hence one can make use of it very easily.

2 Literature Review

This section is a study on the researches that have been carried out in the assistive
technologies sector. The paper [1] puts forth a system which consists of two cameras.
These cameras are fixed on the glasses of the visually impaired. The proposed project
uses a blind stick onto which the sensors are attached. Infrared sensors are used which
use infrared waves to detect objects in front of the user and an audio signal is sent to the
user. The information about the obstacles must be trained into the system. This paper
[2] proposes a system which consists of Raspberry Pi, camera and ultrasonic sensors
to detect and identify the obstacles. It uses digital image processing for obstacle
identification. This system [3] gives the data of different types of objects along with
their size and distance from the user. The signal is processed using MATLAB. Apart
from audio signal, it also gives out a vibration when an object gets detected. This is
done using ultrasonic sensor. Whenever an obstacle is detected by ultrasonic sensor,
the motor vibrates. This paper [4] proposes a smart stick using machine learning
techniques, Google Assistant and ultrasonic sensor to detect the obstacles and alert
the user via vibration. Also the user can find out his/her location through an audio
output. The project [5] proposes a system using Android Smartphone and image
processing libraries such as OpenCV and Google Cloud Vision API to detect an
IRIS: An Application for the Visually Impaired Using Google … 31

image using the camera of the Smartphone. An audio signal is given out which alerts
the user of an obstacle that is present in front.
There are various other systems that are present in the market. These are Laser
Cane, Sonic Torch, Sonic Path Finder, Mowat Sensor, Meldog, Navbelt, etc. These
systems are called Electronic Travel Aid (ETA) which are developed for the visually
impaired aiding in their mobility. Majority of these systems just detect the obstacle
but does not identify it. The visually impaired do not know what kind of obstacle
is present. Also, these are either bulky or quite expensive and not very accurate.
According to the National Center for Biotechnology Information [6], most of these
commercially available ETAs cost between 300$ and 3000$. Although there are some
cheaper options available in India at a price range of 40$–50$, it only uses sensors
such as ultrasonic sensor or laser to detect an obstacle but does not identify it.

3 Proposed Device and Method

3.1 Construction

The skeleton of the device (IRIS) is made of PVC material on which different compo-
nents are mounted. Raspberry Pi is the vital part of the device. All the components
are interfaced with it for the efficient working of the device. Some of the components
like GSM module (USB dongle), GPS module, power bank (power source for the
Raspberry Pi), and wireless adapter (for Internet connectivity) along with the Pi are
encased in a box. The box is then positioned at a convenient distance near the upper
end. The rest of the components lie outside the box and are positioned at various
levels on the stick. Two ultrasonic sensors are placed at different levels on the stick
to provide obstacle detection above and below the knee level. The first Ultrasonic
sensor is implemented on the lower side of the stick to avoid small obstacles below
the knee level. A buzzer is used which buzzes when an object is discovered by this
sensor. The second ultrasonic sensor is implemented to detect the obstacle above
the knee level. This sensor is set to a particular range to trigger the camera mounted
on the stick to capture the image of the obstacle detected by it for the identification
of the obstacle. The information regarding the type of the obstacle and its distance
from the user is conveyed via headphones or earphones connected to the audio jack
of the Raspberry Pi using text-to-speech convertor, espeak. Two push buttons are
implemented on the stick. One of the push buttons is used for shutting the system off
when not in use and is called the power button. A switch is placed on the box which
is used to cut off the power supply to the Raspberry Pi as soon as the power button
is pressed. The second push button can be pressed by the visually impaired when
they feel they have lost their path or reached an unknown place and need some help
and this button is called the SOS button. The GPS and the GSM module come into
picture as soon as this button is pressed. The location of the user is traced by using
the GPS Module. The device also sends an emergency message (I am in trouble)
Other documents randomly have
different content
think of John Wesley’s doctrine of perfection?”
“Oh, it’s absolutely sound and proven,” admitted Elmer,
wondering what the devil Mr. Wesley’s doctrine of perfection might
be.
It is possible that the presence of the elder Benhams, preventing
too close a communion with Cleo, kept Elmer from understanding
what it meant that he should not greatly have longed to embrace
her. He translated his lack of urgency into virtue; and went about
assuring himself that he was indeed a reformed and perfected
character . . . and so went home and hung about the kitchen,
chattering with little Jane Clark in pastoral jokiness.
Even when he was alone with Cleo, when she drove him in the
proud Benham motor for calls in the country, even while he was
volubly telling himself how handsome she was, he was never quite
natural with her.

II
He called on an evening of late November, and both her parents
were out, attending Eastern Star. She looked dreary and red-eyed.
He crowed benevolently while they stood at the parlor door, “Why,
Sister Cleo, what’s the matter? You look kind of sad.”
“Oh, it’s nothing—”
“Come on now! Tell me! I’ll pray for you, or beat somebody up,
whichever you prefer!”
“Oh, I don’t think you ought to joke about—— Anyway, it’s really
nothing.”
She was staring at the floor. He felt buoyant and dominating, so
delightfully stronger than she. He lifted her chin with his forefinger,
demanding, “Look up at me now!”
In her naked eyes there was such shameful, shameless longing
for him that he was drawn. He could not but slip his arm around her,
and she dropped her head on his shoulder, weeping, all her pride
gone from her. He was so exalted by the realization of his own
power that he took it for passion, and suddenly he was kissing her,
conscious of the pale fineness of her skin, her flattering yielding to
him; suddenly he was blurting, “I’ve loved you, oh, terrible, ever
since the first second I saw you!”
As she sat on his knee, as she drooped against him unresisting,
he was certain that she was very beautiful, altogether desirable.
The Benhams came home—Mrs. Benham to cry happily over the
engagement, and Mr. Benham to indulge in a deal of cordial back-
slapping, and such jests as, “Well, by golly, now I’m going to have a
real live preacher in the family, guess I’ll have to be so doggone
honest that the store won’t hardly pay!”

III
His mother came on from Kansas for the wedding, in January.
Her happiness in seeing him in his pulpit, in seeing the beauty and
purity of Cleo—and the prosperity of Cleo’s father—was such that
she forgot her long dragging sorrow in his many disloyalties to the
God she had given him, in his having deserted the Baptist sanctuary
for the dubious, the almost agnostic liberalisms of the Methodists.
With his mother present, with Cleo going about roused to a rosy
excitement, with Mrs. Benham mothering everybody and frantically
cooking, with Mr. Benham taking him out to the back-porch and
presenting him with a check for five thousand dollars, Elmer had the
feeling of possessing a family, of being rooted and solid and secure.
For the wedding there were scores of cocoanut cakes and
hundreds of orange blossoms, roses from a real city florist in Sparta,
new photographs for the family album, a tub of strictly temperance
punch and beautiful but modest lingerie for Cleo. It was tremendous.
But Elmer was a little saddened by the fact that there was no one
whom he wanted for best man; no one who had been his friend
since Jim Lefferts.
He asked Ray Faucett, butter-maker at the creamery and choir-
singer in the church, and the village was flattered that out of the
hundreds of intimates Elmer must have in the great world outside,
he should have chosen one of their own boys.
They were married, during a half blizzard, by the district
superintendent. They took the train for Zenith, to stop overnight on
their way to Chicago.
Not till he was on the train, the shouting and the rice-showers
over, did Elmer gasp to himself, looking at Cleo’s rather unchanging
smile, “Oh, good God, I’ve gone and tied myself up, and I never can
have any fun again!”
But he was very manly, gentlemanly in fact; he concealed his
distaste for her and entertained her with an account of the beauties
of Longfellow.

IV
Cleo looked tired, and toward the end of the journey, in the
winter evening, with the gale desolate, she seemed scarce to be
listening to his observations on graded Sunday School lessons, the
treatment of corns, his triumphs at Sister Falconer’s meetings, and
the inferiority of the Reverend Clyde Tippey.
“Well, you might pay a little attention to me, anyway!” he
snarled.
“Oh, I’m sorry! I really was paying attention. I’m just tired—all
the preparations for the wedding and everything.” She looked at him
beseechingly. “Oh, Elmer, you must take care of me! I’m giving
myself to you entirely—oh, completely.”
“Huh! So you look at it as a sacrifice to marry me, do you!”
“Oh, no, I didn’t mean it that way—”
“And I suppose you think I don’t intend to take care of you! Sure!
Prob’ly I stay out late nights and play cards and gamble and drink
and run around after women! Of course! I’m not a minister of the
gospel—I’m a saloon-keeper!”
“Oh, dear, dear, dear, oh, my dearest, I didn’t mean to hurt you! I
just meant—— You’re so strong and big, and I’m—oh, of course I’m
not a tiny little thing, but I haven’t got your strength.”
He enjoyed feeling injured, but he was warning himself, “Shut up,
you chump! You’ll never educate her to make love if you go bawling
her out.”
He magnanimously comforted her: “Oh, I know. Of course, you
poor dear. Fool thing anyway, your mother having this big wedding,
and all the eats and the relatives coming in and everything.”
And with all this, she still seemed distressed.
But he patted her hand, and talked about the cottage they were
going to furnish in Banjo Crossing; and as he thought of the
approaching Zenith, of their room at the O’Hearn House (there was
no necessity for a whole suite, as formerly, when he had had to
impress his Prosperity pupils), he became more ardent, whispered to
her that she was beautiful, stroked her arm till she trembled.

V
The bell-boy had scarcely closed the door of their room, with its
double bed, when he had seized her, torn off her overcoat, with its
snow-wet collar, and hurled it on the floor. He kissed her throat.
When he had loosened his clasp, she retreated, the back of her hand
fearfully at her lips, her voice terrified as she begged, “Oh, don’t!
Not now! I’m afraid!”
“That’s damned nonsense!” he raged, stalking her as she backed
away.
“Oh, no, please!”
“Say, what the devil do you think marriage is?”
“Oh, I’ve never heard you curse before!”
“My God, I wouldn’t, if you didn’t act so’s it’d try the patience of
a saint on a monument!” He controlled himself. “Now, now, now! I’m
sorry! Guess I’m kind of tired, too. There, there, little girl. Didn’t
mean to scare you. Excuse me. Just showed I was crazy in love with
you, don’t you see?”
 To his broad and apostolic smirk she responded with a weak
smile, and he seized her again, laid his thick hand on her breast.
Between his long embraces, though his anger at her limpness was
growing, he sought to encourage her by shouting, “Come on now,
Clee, show some spunk!”
She did not forbid him again; she was merely a pale
acquiescence—pale save when she flushed unhappily as he made
fun of the old-fashioned, long-sleeved nightgown which she timidly
put on in the indifferent privacy of the bathroom.
“Gee, you might as well wear a gunny-sack!” he roared, holding
out his arms. She tried to look confident as she slowly moved toward
him. She did not succeed.
“Fellow ought to be brutal, for her own sake,” he told himself,
and seized her shoulders.
When he awoke beside her and found her crying, he really did
have to speak up to her.
“You look here now! The fact you’re a preacher’s wife doesn’t
keep you from being human! You’re a fine one to teach brats in
Sunday School!” he said, and many other strong spirited things,
while she wept, her hair disordered round her meek face, which he
hated.

VI
The discovery that Cleo would never be a lively lover threw him
the more into ambition when they had returned to Banjo Crossing.
Cleo, though she was unceasingly bewildered by his furies, found
something of happiness in furnishing their small house, arranging his
books, admiring his pulpit eloquence, and in receiving, as the
Pastor’s Wife, homage even from her old friends. He was able to
forget her, and all his thought went to his holy climbing. He was
eager for the Annual Conference, in spring; he had to get on, to a
larger town, a larger church.
 He was bored by Banjo Crossing. The life of a small-town
preacher, prevented from engaging even in the bucolic pleasures, is
rather duller than that of a watchman at a railroad-crossing.
Elmer hadn’t, actually, enough to do. Though later, in
“institutional churches,” he was to be as hustling as any other
business man, now he had not over twenty hours a week of real
activity. There were four meetings every Sunday, if he attended
Sunday School and Epworth League as well as church; there was
prayer-meeting on Wednesday evening, choir practise on Friday, the
Ladies’ Aid and the Missionary Society every fortnight or so, and
perhaps once a fortnight a wedding, a funeral. Pastoral calls took not
over six hours a week. With the aid of his reference books, he could
prepare his two sermons in five hours—and on weeks when he felt
lazy, or the fishing was good, that was three hours more than he
actually took.
In the austerities of the library Elmer was indolent, but he did like
to rush about, meet people, make a show of accomplishment. It
wasn’t possible to accomplish much in Banjo. The good villagers
were content with Sunday and Wednesday-evening piety.
But he did begin to write advertisements for his weekly services
—the inception of that salesmanship of salvation which was to make
him known and respected in every advertising club and forward-
looking church in the country. The readers of notices to the effect
that services would be held, as usual, in the Banjo Valley Pioneer
were startled to find among the Presbyterian Church, the Disciples
Church, the United Brethren Church, the Baptist Church, this
advertisement:
WAKE UP, MR. DEVIL!

If old Satan were as lazy as some would-be Christians in

this burg, we’d all be safe. But he isn’t! Come out next
Sunday, 10:30 a. m. and hear a red-blooded sermon by Rev.
Gantry on

WOULD JESUS PLAY POKER?

 M. E. Church
He improved his typewriting, and that was a fine thing to do. The
Reverend Elmer Gantry’s powerful nature had been cramped by the
slow use of a pen; it needed the gallop of the keys; and from his
typewriter were increasingly to come floods of new moral and social
gospels.
In February he held two weeks of intensive evangelistic
meetings. He had in a traveling missioner, who wept, and his wife,
who sang. Neither of them, Elmer chuckled privily, could compare
with himself, who had worked with Sharon Falconer. But they were
new to Banjo Crossing, and he saw to it that it was himself who at
the climax of hysteria charged down into the frightened mob and
warned them that unless they came up and knelt in subjection, they
might be snatched to hell before breakfast.
There were twelve additions to the church, and five renewals of
faith on the part of backsliders, and Elmer was able to have
published in the Western Christian Advocate a note which carried his
credit through all the circles of the saints:

The church at Banjo Crossing has had a remarkable and

stirring revival under Brother T. R. Feesels and Sister Feesels,
the singing evangelist, assisted by the local pastor, Reverend
Gantry, who was himself formerly in evangelistic work as
assistant to the late Sharon Falconer. A great outpouring of
the spirit and far-reaching results are announced, with many
uniting with the church.

He also, after letting the town know how much it added to his
burdens, revived and every week for two weeks personally
supervised a Junior Epworth League—the juvenile department of
that admirable association of young people whose purpose is, it has
itself announced, to “take the wreck out of recreation and make it
re-creation.”
He had a note from Bishop Toomis hinting that the bishop had
most gratifying reports from the district superintendent about
Elmer’s “diligent and genuinely creative efforts” and hinting that at
the coming Annual Conference, Elmer would be shifted to a
considerably larger church.
“Fine!” glowed Elmer. “Gosh, I’ll be glad to get away. These rubes
here get about as much out of high-class religion, like I give them,
as a fleet of mules!”

VII
Ishuah Rogers was dead, and they were holding his funeral at
the Methodist Church. As farmer, as store-keeper, as postmaster, he
had lived all his seventy-nine years in Banjo Crossing.
Old J. F. Whittlesey was shaken by Ishuah’s death. They had
been boys together, young men together, neighbors on the farm,
and in his last years, when Ishuah was nearly blind and living with
his daughter Jenny, J. F. Whittlesey had come into town every day to
spend hours sitting with him on the porch, wrangling over Blaine and
Grover Cleveland. Whittlesey hadn’t another friend left alive. To drive
past Jenny’s now and not see old Ishuah made the world empty.
He was in the front row at the church; he could see his friend’s
face in the open coffin. All of Ishuah’s meanness and fussiness and
care was wiped out; there was only the dumb nobility with which he
had faced blizzard and August heat, labor and sorrow; only the
heroic thing Whittlesey had loved in him.
And he would not see Ishuah again, ever.
He listened to Elmer, who, his eyes almost filled at the drama of a
church full of people mourning their old friend, lulled them with
Revelation’s triumphant song:

These are they that come out of the great tribulation, and
they washed their robes, and made them white in the blood
of the Lamb. Therefore are they before the throne of God;
and they serve him day and night in his temple; and he that
sitteth on the throne shall spread his tabernacle over them.
They shall hunger no more, neither thirst any more; neither
 shall the sun strike upon them, nor any heat: for the Lamb
that is in the midst of the throne shall be their shepherd, and
shall guide them unto fountains of waters of life: and God
shall wipe away every tear from their eyes.

They sang “O God, Our Help in Ages Past,” and Elmer led the
singing, while old Whittlesey tried to pipe up with them.
They filed past the coffin. When Whittlesey had this last
moment’s glimpse of Ishuah’s sunken face, his dry eyes were blind,
and he staggered.
Elmer caught him with his great arms, and whispered, “He has
gone to his glory, to his great reward! Don’t let’s sorrow for him!”
In Elmer’s confident strength old Whittlesey found reassurance.
He clung to him, muttering, “God bless you, Brother,” before he
hobbled out.

VIII
“You were wonderful at the funeral today! I’ve never seen you so
sure of immortality,” worshiped Cleo, as they walked home.
“Yuh, but they don’t appreciate it—not even when I said about
how this old fellow was a sure-enough hero. We got to get on to
some burg where I’ll have a chance.”
“Don’t you think God’s in Banjo Crossing as much as in a city?”
“Oh, now, Cleo, don’t go and get religious on me! You simply
can’t understand how it takes it out of a fellow to do a funeral right
and send ’em all home solaced. You may find God here, but you
don’t find the salaries!”
He was not angry with Cleo now, nor bullying. In these two
months he had become indifferent to her; indifferent enough to stop
hating her and to admire her conduct of the Sunday School, her
tactful handling of the good sisters of the church when they came
snooping to the parsonage.
 “I think I’ll take a little walk,” he muttered when they reached
home.
He came to the Widow Clark’s house, where he had lived as
bachelor.
Jane was out in the yard, the March breeze molding her skirt
about her; rosy face darker and eyes more soft as she saw the
pastor hailing her, magnificently raising his hat.
She fluttered toward him.
“You folks ever miss me? Guess you’re glad to get rid of the poor
old preacher that was always cluttering up the house!”
“We miss you awfully!”
He felt his whole body yearning toward her. Hurriedly he left her,
and wished he hadn’t left her, and hastened to get himself far from
the danger to his respectability. He hated Cleo again now, in an
injured, puzzled way.
“I think I’ll sneak up to Sparta this week,” he fumed, then: “No!
Conference coming in ten days; can’t take any chances till after
that.”

IX
The Annual Conference, held in Sparta, late in March. The high
time of the year, when the Methodist preachers of half a dozen
districts met together for prayer and rejoicing, to hear of the
progress of the Kingdom and incidentally to learn whether they were
to have better jobs this coming year.
The bishop presiding—Wesley R. Toomis, himself—with his
district superintendents, grave and bustling.
The preachers, trying to look as though prospective higher
salaries were unworthy their attention.
Between meetings they milled about in the large auditorium of
the Preston Memorial Methodist Church: visiting laymen and nearly
three hundred ministers.
 Veteran country parsons, whiskered and spectacled, rusty-coated
and stooped, still serving two country churches, or three or four;
driving their fifty miles a week; content for reading with the
Scriptures and the weekly Advocate.
New-fledged country preachers, their large hands still calloused
from plow-handle and reins, content for learning with two years of
high school, content with the Old Testament for history and geology.
The preachers of the larger towns; most of them hard to
recognize as clerics, in their neat business suits and modest four-in-
hands; frightfully cordial one to another; perhaps a quarter of them
known as modernists and given to reading popular manuals of
biology and psychology; the other three-quarters still devoted to
banging the pulpit apropos of Genesis.
But moving through these masses, easily noticeable, the
inevitable successes: the district superintendents, the pastors of
large city congregations, the conceivable candidates for college
presidencies, mission-boards, boards of publication, bishoprics.
They were not all of them leonine and actor-like, these staff
officers. No few were gaunt, or small, wiry, spectacled, and earnest;
but they were all admirable politicians, long in memory of names,
quick to find flattering answers. They believed that the Lord rules
everything, but that it was only friendly to help him out; and that the
enrollment of political allies helped almost as much as prayer in
becoming known as suitable material for lucrative pastorates.
Among these leaders were the Savonarolas, gloomy fellows,
viewing the progress of machine civilization with biliousness; capable
of drawing thousands of auditors by their spicy but chaste
denunciations of burglary, dancing, and show-windows filled with
lingerie.
Then the renowned liberals, preachers who filled city tabernacles
or churches in university towns by showing that skipping whatever
seemed unreasonable in the Bible did not interfere with considering
it all divinely inspired, and that there are large moral lessons in the
paintings of Landseer and Rosa Bonheur.
 Most notable among the aristocrats were a certain number of
large, suave, deep-voiced, inescapably cordial clerical gentlemen
who would have looked well in Shakespearean productions or as
floor-walkers. And with them was presently to be found the
Reverend Elmer Gantry.
He was a new-comer, he was merely hoping to have the
Conference recognize his credentials and accept him as a member,
and he had only a tiny church, yet from somewhere crept the rumor
that he was a man to be watched, to be enrolled in one’s own
political machine; and he was called “Brother” by a pastor whose
sacred rating was said to be not less than ten thousand a year. They
observed him; they conversed with him not only on the sacraments
but on automobiles and the use of pledge-envelopes; and as they
felt the warmth of his handshake, as they heard the amiable bim-
bom of his voice, saw his manly eyes, untroubled by doubts or
scruples, and noted that he wore his morning clothes as well as any
spiritual magnate among them, they greeted him and sought him
out and recognized him as a future captain of the hosts of the
Almighty.
Cleo’s graciousness added to his prestige.
For three whole days before bringing her up to the Conference,
Elmer had gone out of his way to soothe her, flatter her, assure her
that whatever misunderstandings they might have had, all was now
a warm snugness of domestic bliss, so that she was eager, gently
deferential to the wives of older pastors as she met them at
receptions at hotels.
Her obvious admiration of Elmer convinced the better clerical
politicians of his domestic safeness.
And they knew that he had been sent for by the bishop—oh, they
knew it! Nothing that the bishop did in these critical days was not
known. There were many among the middle-aged ministers who had
become worried over prolonged stays in small towns, and who
wanted to whisper to the bishop how well they would suit larger
opportunities. (The list of appointments had already been made out
by the bishop and his council, yet surely it could be changed a little
—just the least bit.) But they could not get near him. Most of the
time the bishop was kept hidden from them at the house of the
president of Winnemac Wesleyan University.
But he sent for Elmer, and even called him by his first name.
“You see, Brother Elmer, I was right! The Methodist Church just
suits you,” said the bishop, his eyes bright under his formidable
brows. “I am able to give you a larger church already. It wouldn’t be
cricket, as the English say—ah, England! how you will enjoy going
there some time; you will find such a fruitful source of the broader
type of sermons in travel; I know that you and your lovely bride—
I’ve had the pleasure of having her pointed out to me—you will both
know the joy and romance of travel one of these days. But as I was
saying: I can give you a rather larger town this time, though it
wouldn’t be proper to tell you which one till I read the list of
appointments to the Conference. And in the near future, if you
continue as you have in your studies and attention to the needs of
your flock and in your excellence of daily living, which the district
superintendent has noted, why, you’ll be due for a much larger field
of service. God bless you!”

X
Elmer was examined by the Conference and readily admitted to
membership.
Among the questions, from the Discipline, which he was able to
answer with a hearty “yes” were these:
Are you going on to perfection?
Do you expect to be made perfect in love in this life?
Are you earnestly striving after it?
Are you resolved to devote yourself wholly to God and his work?
Have you considered the Rules for a Preacher, especially those
relating to Diligence, to Punctuality, and to Doing the Work to which
you are assigned?
 Will you recommend fasting or diligence, both by precept and
example?

* * * * * * *

It was, the Conference members said, one to another, a pleasure

to examine a candidate who could answer the questions with such
ringing certainty.

* * * * * * *

Celebrating his renunciation of all fleshy devices and pleasures by

wolfing a steak, fried onions, fried potatoes, corn, three cups of
coffee, and two slices of apple pie with ice cream, Elmer
condescended to Cleo, “I went through a-whooping! Liked to of seen
any of those poor boobs I was with in the seminary answer up like I
did!”

XI
They listened to reports on collections for missions, on the
creation of new schools and churches; they heard ever so many
prayers; they were polite during what were known as “inspirational
addresses” by the bishop and the Rev. Dr. S. Palmer Shootz. But they
were waiting for the moment when the bishop should read the list of
appointments.
They looked as blank as they could, but their nails creased their
palms as the bishop rose. They tried to be loyal to their army, but
this lean parson thought of the boy who was going to college, this
worried-faced youngster thought of the operation for his wife, this
aged campaigner whose voice had been failing wondered whether
he would be kept on in his well-padded church.
The bishop’s snappy voice popped:
Sparta District:
 Albee Center, W. A. Vance
Ardmore, Abraham Mundon—
And Elmer listened with them, suddenly terrified.
What did the bishop mean by a “rather larger town”? Some
horrible hole with twelve hundred people?
Then he startled and glowed, and his fellow priests nodded to
him in congratulation, as the bishop read out “Rudd Center, Elmer
Gantry.”
For there were forty-one hundred people in Rudd Center; it was
noted for good works and a large pop factory; and he was on his
way to greatness, to inspiring the world and becoming a bishop.
 CHAPTER XXII

I
a year he spent in Rudd Center, three years in Vulcan, and two
years in Sparta. As there were 4,100 people in Rudd Center, 47,000
in Vulcan, and 129,000 in Sparta, it may be seen that the Reverend
Elmer Gantry was climbing swiftly in Christian influence and
character.
In Rudd Center he passed his Mizpah final examinations and
received his Bachelor of Divinity degree from the seminary; in Rudd
Center he discovered the art of joining, which was later to enable
him to meet the more enterprising and solid men of affairs—oculists
and editors and manufacturers of bathtubs—and enlist their practical
genius in his crusades for spirituality.
He joined the Masons, the Odd Fellows, and the Maccabees. He
made the Memorial Day address to the G. A. R., and he made the
speech welcoming the local representative home from Congress
after having won the poker championship of the House.
Vulcan was marked, aside from his labors for perfection, by the
birth of his two children—Nat, in 1916, and Bernice, whom they
called Bunny, in 1917—and by his ceasing to educate his wife in his
ideals of amour.
It all blew up a month after the birth of Bunny.
Elmer had, that evening, been addressing the Rod and Gun Club
dinner. He had pointed out that our Lord must have been in favor of
Rods and Guns for, he said, “I want you boys to notice that the
Master, when he picked out his first disciples, didn’t select a couple
of stoop-shouldered, pigeon-toed mollycoddles but a pair of first-
class fishermen!”
He was excited to intoxication by their laughter.
 Since Bunny’s birth he had been sleeping in the guest-room, but
now, walking airily, he tiptoed into Cleo’s room at eleven, with that
look of self-conscious innocence which passionless wives instantly
catch and dread.
“Well, you sweet thing, it sure went off great! They all liked my
spiel. Why, you poor lonely girl, shame you have to sleep all alone
here, poor baby!” he said, stroking her shoulder as she sat propped
against the pillows. “Guess I’ll have to come sleep here tonight.”
She breathed hard, tried to look resolute. “Please! Not yet!”
“What do you mean?”
“Please! I’m tired tonight. Just kiss me good-night, and let me
pop off to sleep.”
“Meaning my attentions aren’t welcome to Your Majesty!” He
paced the floor. “Young woman, it’s about time for a showdown! I’ve
hinted at this before, but I’ve been as charitable and long-suffering
as I could, but, by God, you’ve gotten away with too much, and then
you try to pretend—— ‘Just kiss me good-night!’ Sure! I’m to be a
monk! I’m to be one of these milk-and-water husbands that’s
perfectly content to hang around the house and not give one little
yip if his wife don’t care for his method of hugging! Well, believe me,
young woman, you got another guess coming, and if you think that
just because I’m a preacher I’m a Willie-boy—— You don’t even
make the slightest smallest effort to learn some passion, but just act
like you had hard work putting up with me! Believe me, there’s other
women a lot better and prettier—yet, and more religious!—that
haven’t thought I was such a damn’ pest to have around! I’m not
going to stand—— Never even making the slightest effort—”
“Oh, Elmer, I have! Honestly I have! If you’d only been more
tender and patient with me at the very first, I might have learned—”
“Rats! All damned nonsense! Trouble with you is, you always
were afraid to face hard facts! Well, I’m sick of it, young woman.
You can go to the devil! This is the last time, believe me!”
He banged the door; he had satisfaction in hearing her sob that
night; and he kept his vow about staying away from her, for almost a
month. Presently he was keeping it altogether; it was a settled thing
that they had separate bedrooms.
And all the while he was almost as confused, as wistful, as she
was; and whenever he found a woman parishioner who was willing
to comfort him, or whenever he was called on important but never
explained affairs to Sparta, he had no bold swagger of satisfaction,
but a guilt, an uneasiness of sin, which displayed itself in
increasingly furious condemnation of the same sin from his pulpit.
“O God, if I could only have gone on with Sharon, I might have
been a decent fellow,” he mourned, in his sorrow sympathetic with
all the world. But the day after, in the sanctuary, he would be salving
that sorrow by raging, “And these dance-hall proprietors, these
tempters of lovely innocent girls, whose doors open to the pit of
death and horror, they shall have reward—they shall burn in
uttermost hell—burn literally—BURN!—and for their suffering we
shall have but joy that the Lord’s justice has been resolutely done!”

II
Something like statewide fame began to cling about the
Reverend Elmer Gantry during his two years in Sparta—1918 to
1920. In the spring of ’18 he was one of the most courageous
defenders of the Midwest against the imminent invasion of the
Germans. He was a Four-Minute Man. He said violent things about
atrocities, and sold Liberty Bonds hugely. He threatened to leave
Sparta to its wickedness while he went out to “take care of our poor
boys” as a chaplain, and he might have done so had the war lasted
another year.
In Sparta, too, he crept from timidly sensational church
advertisements to such blasts as must have shaken the Devil
himself. Anyway, they brought six hundred delighted sinners to
church every Sunday evening, and after one sermon on the horrors
of booze, a saloon-keeper, slightly intoxicated, remarked “Whoop!”
and put a fifty-dollar bill in the plate.
 Not to this day, with all the advance in intellectual advertising,
has there been seen a more arousing effort to sell salvation than
Elmer’s prose poem in the Sparta World-Chronicle on a Saturday in
December, 1919:
WOULD YOU LIKE YOUR MOTHER TO GO
BATHING WITHOUT STOCKINGS?

Do you believe in old-fashioned womanhood, that can love

and laugh and still be the symbols of God’s own
righteousness, bringing a tear to the eye as one remembers
their brooding tenderness? Would you like to see your own
dear mammy indulging in mixed bathing or dancing that
Hell’s own fool monkeyshine, the one-step?

Reverend Elmer Gantry

will answer these questions and others next Sunday

morning. Gantry shoots straight from the shoulder.

POPLAR AVENUE METHODIST CHURCH

Follow the crowd to the beautiful times

At the beautiful church with the beautiful chimes.

III
While he was in Sparta, national prohibition arrived, with its high-
colored opportunities for pulpit-orators, and in Sparta he was
inspired to his greatest political campaign.
The obviously respectable candidate for mayor of Sparta was a
Christian Business Man, a Presbyterian who was a manufacturer of
rubber overshoes. It is true that he was accused of owning the
buildings in which were several of the worst brothels and blind tigers
in the city, but it had amply been explained that the unfortunate
gentleman had not been able to kick out his tenants, and that he
gave practically all his receipts from the property to missionary work
in China.
His opponent was a man in every way objectionable to Elmer’s
principles: a Jew, a radical who criticized the churches for not paying
taxes, a sensational and publicity-seeking lawyer who took the cases
of labor unions and negroes without fee. When he consulted them,
Elmer’s Official Board agreed that the Presbyterian was the only man
to support. They pointed out that the trouble with the radical Jew
was that he was not only a radical but a Jew.
Yet Elmer was not satisfied. He had, possibly, less objection to
houses of ill fame than one would have judged from his pulpit
utterances, and he certainly approved the Presbyterian’s position
that “we must not try dangerous experiments in government but
adhere courageously to the proven merits and economies of the
present administration.” But talking with members of his
congregation, Elmer found that the Plain People—and the plain, the
very plain, people did make up such a large percentage of his flock—
hated the Presbyterian and had a surprised admiration for the Jew.
“He’s awful’ kind to poor folks,” said they.
Elmer had what he called a “hunch.”
“All the swells are going to support this guy McGarry, but darned
if I don’t think the Yid’ll win, and anybody that roots for him’ll stand
ace-high after the election,” he reasoned.
He came out boisterously for the Jew. The newspapers squealed
and the Presbyterians bellowed and the rabbis softly chuckled.
Not only from his pulpit but in scattered halls Elmer campaigned
and thundered. He was smeared once with rotten eggs in a hall near
the red-light district, and once an illicit booze-dealer tried to punch
his nose, and that was a very happy time for Elmer.
The booze-dealer, a bulbous angry man, climbed up on the stage
of the hall and swayed toward Elmer, weaving with his fists,
rumbling, “You damn’ lying gospel-shark, I’ll show you—”
 The forgotten star of the Terwillinger team leaped into life. He
was calm as in a scrimmage. He strode over, calculatingly regarded
the point of the bootlegger’s jaw, and caught him on it, exact. He
saw the man slumping down, but he did not stand looking; he
swung back to the reading-stand and went on speaking. The whole
audience rose, clamorous with applause, and Elmer Gantry had for a
second become the most famous man in town.
The newspapers admitted that he was affecting the campaign,
and one of them swung to his support. He was so strong on virtue
and the purity of womanhood and the evils of liquor that to oppose
him was to admit one’s self a debauchee.
At the business meeting of his church there was a stirring
squabble over his activities. When the leading trustee, a friend of the
Presbyterian candidate, declared that he was going to resign unless
Elmer stopped, an aged janitor shrieked, “And all the rest of us will
resign unless the Reverend keeps it up!” There was gleeful and
unseemly applause, and Elmer beamed.
The campaign grew so bellicose that reporters came up from the
Zenith newspapers; one of them the renowned Bill Kingdom of the
Zenith Advocate-Times. Elmer loved reporters. They quoted him on
everything from the Bible in the schools to the Armenian mandate.
He was careful not to call them “boys” but “gentlemen,” not to slap
them too often on the back; he kept excellent cigars for them; and
he always said, “I’m afraid I can’t talk to you as a preacher. I get too
much of that on Sunday. I’m just speaking as an ordinary citizen
who longs to have a clean city in which to bring up his kiddies.”
Bill Kingdom almost liked him, and the story about “the crusading
parson” which he sent up to the Zenith Advocate-Times—the
Thunderer of the whole state of Winnemac—was run on the third
page, with a photograph of Elmer thrusting out his fist as if to crush
all the sensualists and malefactors in the world.
Sparta papers reprinted the story and spoke of it with reverence.
The Jew won the campaign.
 And immediately after this—six months before the Annual
Conference of 1920—Bishop Toomis sent for Elmer.

IV
“At first I was afraid,” said the bishop, “you were making a great
mistake in soiling yourself in this Sparta campaign. After all, it’s our
mission to preach the pure gospel and the saving blood of Jesus,
and not to monkey with politics. But you’ve been so successful that I
can forgive you, and the time has come—— At the next Conference
I shall be able to offer you at last a church here in Zenith, and a very
large one, but with problems that call for heroic energy. It’s the old
Wellspring Church, down here on Stanley Avenue, corner of
Dodsworth, in what we call ‘Old Town.’ It used to be the most
fashionable and useful Methodist church in town, but the section has
run down, and the membership has declined from something like
fourteen hundred to about eight hundred, and under the present
pastor—you know him—old Seriere, fine noble Christian gentleman,
great soul, but a pretty rotten speaker—I don’t guess they have
more than a hundred or so at morning service. Shame, Elmer,
wicked shame to see this great institution, meant for the quickening
of such vast multitudes of souls, declining and, by thunder, not
hardly giving a cent for missions! I wonder if you could revive it? Go
look it over, and the neighborhood, and let me know what you think.
Or whether you’d rather stay on in Sparta. You’ll get less salary at
Wellspring than you’re getting in Sparta—four thousand, isn’t it?—
but if you build up the church, guess the Official Board will properly
remunerate your labors.”
A church in Zenith! Elmer would—almost—have taken it with no
salary whatever. He could see his Doctor of Divinity degree at hand,
his bishopric or college presidency or fabulous pulpit in New York.
He found the Wellspring M. E. Church a hideous graystone hulk
with gravy-colored windows, and a tall spire ornamented with tin
gargoyles and alternate layers of tiles in distressing red and green.
The neighborhood had been smart, but the brick mansions, once
leisurely among lawns and gardens, were scabrous and slovenly,
turned into boarding-houses with delicatessen shops in the
basements.
“Gosh, this section never will come back. Too many of the
doggone hoi polloi. Bunch of Wops. Nobody for ten blocks that
would put more’n ten cents in the collection. Nothing doing! I’m not
going to run a soup-kitchen and tell a bunch of dirty bums to come
to Jesus. Not on your life!”
But he saw, a block from the church, a new apartment-house,
and near it an excavation.
“Hm. Might come back, in apartments, at that. Mustn’t jump too
quick. Besides, these folks need the gospel just as much as the
swell-headed plutes out on Royal Ridge,” reflected the Reverend Mr.
Gantry.
Through his old acquaintance, Gil O’Hearn of the O’Hearn House,
Elmer met a responsible contractor and inquired into the fruitfulness
of the Wellspring vineyard.
“Yes, they’re dead certain to build a bunch of apartment-houses,
and pretty good ones, in that neighborhood these next few years. Be
a big residential boom in Old Town. It’s near enough in to be handy
to the business section, and far enough from the Union Station so’s
they haven’t got any warehouses or wholesalers. Good buy,
Reverend.”
“Oh, I’m not buying—I’m just selling—selling the gospel!” said
the Reverend, and he went to inform Bishop Toomis that after prayer
and meditation he had been led to accept the pastorate of the
Wellspring Church.
So, at thirty-nine, Cæsar came to Rome, and Rome heard about
it immediately.
 CHAPTER XXIII

I
he did not stand by the altar now, uplifted in a vow that he would
be good and reverent. He was like the new general manager of a
factory as he bustled for the first time through the Wellspring
Methodist Church, Zenith, and his first comment was “The plant’s
run down—have to buck it up.”
He was accompanied on his inspection by his staff: Miss Bundle,
church secretary and personal secretary to himself, a decayed and
plaintive lady distressingly free of seductiveness; Miss Weezeger, the
deaconess, given to fat and good works; and A. F. Cherry, organist
and musical director, engaged only on part time.
He was disappointed that the church could not give him a
pastoral assistant or a director of religious education. He’d have
them, soon enough—and boss them! Great!
He found an auditorium which would hold sixteen hundred
people but which was offensively gloomy in its streaky windows, its
brown plaster walls, its cast-iron pillars. The rear wall of the chancel
was painted a lugubrious blue scattered with stars which had ceased
to twinkle; and the pulpit was of dark oak, crowned with a foolish,
tasseled, faded green velvet cushion. The whole auditorium was
heavy and forbidding; the stretch of empty brown-grained pews
stared at him dolorously.
“Certainly must have been a swell bunch of cheerful Christians
that made this layout! I’ll have a new church here in five years—one
with some pep to it, and Gothic fixin’s and an up-to-date educational
and entertainment plant,” reflected the new priest.
The Sunday School rooms were spacious enough, but dingy,
scattered with torn hymn books; the kitchen in the basement, for
church suppers, had a rusty ancient stove and piles of chipped
dishes. Elmer’s own study and office was airless, and looked out on
the flivver-crowded yard of a garage. And Mr. Cherry said the organ
was rather more than wheezy.
“Oh, well,” Elmer conferred with himself afterward, “what do I
care! Anyway, there’s plenty of room for the crowds, and, believe
me, I’m the boy can drag ’em in! . . . God, what a frump that Bundle
woman is! One of these days I’ll have a smart girl secretary—a
good-looker. Well, hurray, ready for the big work! I’ll show this town
what high-class preaching is!”
Not for three days did he chance to think that Cleo might also
like to see the church.

II
Though there were nearly four hundred thousand people in
Zenith and only nine hundred in Banjo Crossing, Elmer’s reception in
the Zenith church basement was remarkably like his reception in the
Banjo basement. There were the same rugged, hard-handed
brothers, the same ample sisters renowned for making doughnuts,
the same brisk little men given to giggling and pious jests. There
were the same homemade ice cream and homemade oratory. But
there were five times as many people as at the Banjo reception, and
Elmer was ever a lover of quantity. And among the transplanted
rustics were several prosperous professional men, several well-
gowned women, and some pretty girls who looked as though they
went to dancing school, Discipline or not.
He felt cheerful and loving toward them—his, as he pointed out
to them, “fellow crusaders marching on resolutely to achievement of
the Kingdom of God on earth.”
It was easy to discover which of the members present from the
Official Board of the church were most worth his attentions. Mr.
Ernest Apfelmus, one of the stewards, was the owner of the Gem of
the Ocean Pie and Cake Corporation. He looked like a puffy and
bewildered urchin suddenly blown up to vast size; he was very rich,
Miss Bundle whispered; and he did not know how to spend his
money except on his wife’s diamonds and the cause of the Lord.
Elmer paid court to Mr. Apfelmus and his wife, who spoke quite a
little English.
Not so rich but even more important, Elmer guessed, was T. J.
Rigg, the famous criminal lawyer, a trustee of Wellspring Church.
Mr. Rigg was small, deep-wrinkled, with amused and knowing
eyes. He would be, Elmer felt instantly, a good man with whom to
drink. His wife’s face was that of a girl, round and smooth and blue-
eyed, though she was fifty and more, and her laughter was lively.
“Those are folks I can shoot straight with,” decided Elmer, and he
kept near them.
Rigg hinted, “Say, Reverend, why don’t you and your good lady
come up to my house after this, and we can loosen up and have a
good laugh and get over this sewing-circle business.”
“I’d certainly like to.” As he spoke Elmer was considering that if
he was really to loosen up, he could not have Cleo about. “Only, I’m
afraid my wife has a headache, poor girl. We’ll just send her along
home and I’ll come with you.”
“After you shake hands a few thousand more times!”
“Exactly!”
Elmer was edified to find that Mr. Rigg had a limousine with a
chauffeur—one of the few in which Elmer had yet ridden. He did like
to have his Christian brethren well heeled. But the sight of the
limousine made him less chummy with the Riggses, more respectful
and unctuous, and when they had dropped Cleo at the hotel, Elmer
leaned gracefully back on the velvet seat, waved his large hand
poetically, and breathed, “Such a welcome the dear people gave me!
I am so grateful! What a real outpouring of the spirit!”
“Look here,” sniffed Rigg, “you don’t have to be pious with us!
Ma and I are a couple of old dragoons. We like religion; like the
good old hymns—takes us back to the hick town we came from; and
we believe religion is a fine thing to keep people in order—they think
of higher things instead of all these strikes and big wages and the
kind of hell-raising that’s throwing the industrial system all out of
kilter. And I like a fine upstanding preacher that can give a good
show. So I’m willing to be a trustee. But we ain’t pious. And any
time you want to let down—and I reckon there must be times when
a big cuss like you must get pretty sick of listening to the sniveling
sisterhood!—you just come to us, and if you want to smoke or even
throw in a little jolt of liquor, as I’ve been known to do, why we’ll
understand. How about it, Ma?”
“You bet!” said Mrs. Rigg. “And I’ll go down to the kitchen, if
cook isn’t there, and fry you up a couple of eggs, and if you don’t
tell the rest of the brethren, there’s always a couple of bottles of
beer on the ice. Like one?”
“Would I!” cheered Elmer. “You bet I would! Only—I cut out
drinking and smoking quite a few years ago. Oh, I had my share
before that! But I stopped, absolute, and I’d hate to break my
record. But you go right ahead. And I want to say that it’ll be a
mighty big relief to have some folks in the church that I can talk to
without shocking ’em half to death. Some of these holier-than-thou
birds—— Lord, they won’t let a preacher be a human being!”
The Rigg house was large, rather faded, full of books which had
been read—history, biography, travels. The smaller sitting-room, with
its log fire and large padded chairs, looked comfortable, but Mrs.
Rigg shouted, “Oh, let’s go out to the kitchen and shake up a welsh
rabbit! I love to cook, and I don’t dast till after the servants go to
bed.”
So his first conference with T. J. Rigg, who became the only
authentic friend Elmer had known since Jim Lefferts, was held at the
shiny white-enamel-topped table in the huge kitchen, with Mrs. Rigg
stalking about, bringing them welsh rabbit, with celery, cold chicken,
whatever she found in the ice box.
“I want your advice, Brother Rigg,” said Elmer. “I want to make
my first sermon here something sen—well, something that’ll make
’em sit up and listen. I don’t have to get the subject in for the
church ads till tomorrow. Now what do you think of some pacifism?”
 “Eh?”
“I know what you think. Of course during the war I was just as
patriotic as anybody—Four-Minute Man, and in another month I’d of
been in uniform. But honest, some of the churches are getting a lot
of kick out of hollering pacifism now the war’s all safely over—some
of the biggest preachers in the country. But far’s I’ve heard,
nobody’s started it here in Zenith yet, and it might make a big
sensation.”
“Yes, that’s so, and course it’s perfectly all right to adopt pacifism
as long as there’s no chance for another war.”
“Or do you think—you know the congregation here—do you think
a more dignified and kind of you might say poetic expository sermon
would impress ’em more? Or what about a good, vigorous, right-out-
from-the-shoulder attack on vice? You know, booze and immorality—
like short skirts—by golly, girls’ skirts getting shorter every year!”
“Now that’s what I’d vote for,” said Rigg. “That’s what gets ’em.
Nothing like a good juicy vice sermon to bring in the crowds. Yes,
sir! Fearless attack on all this drinking and this awful sex immorality
that’s getting so prevalent.” Mr. Rigg meditatively mixed a highball,
keeping it light because next morning in court he had to defend a
lady accused of running a badger game. “You bet. Some folks say
sermons like that are just sensational, but I always tell ’em: once the
preacher gets the folks into the church that way—and mighty few
appreciate how hard it is to do a good vice sermon; jolt ’em enough
and yet not make it too dirty—once you get in the folks, then you
can give ’em some good, solid, old-time religion and show ’em
salvation and teach ’em to observe the laws and do an honest day’s
work for an honest day’s pay, ’stead of clock-watching like my
doggone clerks do! Yep, if you ask me, try the vice. . . . Oh, say, Ma,
do you think the Reverend would be shocked by that story about the
chambermaid and the traveling man that Mark was telling us?”
Elmer was not shocked. In fact he had another droll tale himself.
He went home at one.
 “I’ll have a good time with those folks,” he reflected, in the luxury
of a taxicab. “Only, better be careful with old Rigg. He’s a shrewd
bird, and he’s onto me. . . . Now what do you mean?” indignantly.
“What do you mean by ‘onto me’? There’s nothing to be onto! I
refused a drink and a cigar, didn’t I? I never cuss except when I lose
my temper, do I? I’m leading an absolutely Christian life. And I’m
bringing a whale of a lot more souls into churches than any of these
pussy-footing tin saints that’re afraid to laugh and jolly people. ‘Onto
me’ nothing!”

III
On Saturday morning, on the page of religious advertisements in
the Zenith newspapers, Elmer’s first sermon was announced in a
two-column spread as dealing with the promising problem: “Can
Strangers Find Haunts of Vice in Zenith?”
They could, and with gratifying ease, said Elmer in his sermon.
He said it before at least four hundred people, as against the
hundred who had normally been attending.
He himself was a stranger in Zenith, and he had gone forth and
he had been “appalled—aghast—bowed in shocked horror” at the
amount of vice, and such interesting and attractive vice. He had
investigated Braun’s Island, a rackety beach and dance-floor and
restaurant at South Zenith, and he had found mixed bathing. He
described the ladies’ legs; he described the two amiable young
women who had picked him up. He told of the waiter who, though
he denied that Braun’s Restaurant itself sold liquor, had been willing
to let him know where to get it, and where to find an all-night game
of poker—“and, mind you, playing poker for keeps, you understand,”
Elmer explained.
On Washington Avenue, North, he had found two movies in
which “the dreadful painted purveyors of putrescent vice”—he meant
the movie actors—had on the screen danced “suggestive steps
which would bring the blush of shame to the cheeks of any decent
woman,” and in which the same purveyors had taken drinks which
he assumed to be the deadly cocktails. On his way to his hotel after
these movies three ladies of the night had accosted him, right under
the White Way of lights. Street-corner loafers—he had apparently
been very chummy with them—had told him of blind pigs, of dope-
peddlers, of strange lecheries.
“That,” he shouted, “is what one stranger was able to find in your
city—now my city, and well beloved! But could he find virtue so
easily, could he, could he? Or just a lot of easy-going churches,
lollygagging along, while the just God threatens this city with the fire
and devouring brimstone that destroyed proud Sodom and
Gomorrah in their abominations! Listen! With the help of God
Almighty, let us raise here in this church a standard of virtue that no
stranger can help seeing! We’re lazy. We’re not burning with a fever
of righteousness. On your knees, you slothful, and pray God to
forgive you and to aid you and me to form a brotherhood of helpful,
joyous, fiercely righteous followers of every commandment of the
Lord Our God!”
The newspapers carried almost all of it. . . . It had just happened
that there were reporters present—it had just happened that Elmer
had been calling up the Advocate-Times on Saturday—it had just
happened that he remembered he had met Bill Kingdom, the
Advocate reporter, in Sparta—it had just happened that to help out
good old Bill he had let him know there would be something stirring
in the church, come Sunday.
The next Saturday Elmer advertised “Is There a Real Devil
Sneaking Around with Horns and Hoofs?” On Sunday there were
seven hundred present. Within two months Elmer was preaching,
ever more confidently and dramatically, to larger crowds than were
drawn by any other church in Zenith except four or five.
But, “Oh, he’s just a new sensation—he can’t last out—hasn’t got
the learning and staying-power. Besides, Old Town is shot to pieces,”
said Elmer’s fellow vinters—particularly his annoyed fellow
Methodists.