IET COMPUTING SERIES 53

Nature-Inspired
Optimization Algorithms
and Soft Computing
Other volumes in this series:
Volume 1 Knowledge Discovery and Data Mining M.A. Bramer (Editor)
Volume 3 Troubled IT Projects: Prevention and turnaround J.M. Smith
Volume 4 UML for Systems Engineering: Watching the wheels, 2nd Edition J. Holt
Volume 5 Intelligent Distributed Video Surveillance Systems S.A. Velastin and P.
Remagnino (Editors)
Volume 6 Trusted Computing C. Mitchell (Editor)
Volume 7 SysML for Systems Engineering J. Holt and S. Perry
Volume 8 Modelling Enterprise Architectures J. Holt and S. Perry
Volume 9 Model-Based Requirements Engineering J. Holt, S. Perry, and M. Bownsword
Volume 13 Trusted Platform Modules: Why, when and how to use them A. Segall
Volume 14 Foundations for Model-Based Systems Engineering: From patterns to
models J. Holt, S. Perry, and M. Bownsword
Volume 15 Big Data and Software Defined Networks J. Taheri (Editor)
Volume 18 Modeling and Simulation of Complex Communication M.A. Niazi (Editor)
Volume 20 SysML for Systems Engineering: A model-based approach, 3rd Edition J.
Holt and S. Perry
Volume 22 Virtual Reality and Light Field Immersive Video Technologies for Real-
World Applications G. Lafruit and M. Tehrani
Volume 23 Data as Infrastructure for Smart Cities L. Suzuki and A. Finkelstein
Volume 24 Ultrascale Computing Systems J. Carretero, E. Jeannot, and A. Zomaya
Volume 25 Big Data-Enabled Internet of Things M. Khan, S. Khan, and A. Zomaya
(Editors)
Volume 26 Handbook of Mathematical Models for Languages and Computation A.
Meduna, P. Horáček, and M. Tomko
Volume 29 Blockchains for Network Security: Principles, technologies and
applications H. Huang, L. Wang, Y. Wu, and K.R. Choo (Editors)
Volume 30 Trustworthy Autonomic Computing T. Eza
Volume 32 Network Classification for Traffic Management: Anomaly detection,
feature selection, clustering and classification Z. Tari, A. Fahad, A. Almalawi,
and X. Yi
Volume 33 Edge Computing: Models, technologies and applications J. Taheri and S.
Deng (Editors)
Volume 34 AI for Emerging Verticals: Human-robot computing, sensing and
networking M.Z. Shakir and N. Ramzan (Editors)
Volume 35 Big Data Recommender Systems Vols 1 & 2 O. Khalid, S.U. Khan and A.Y.
Zomaya (Editors)
Volume 37 Handbook of Big Data Analytics Vols 1 & 2 V. Ravi and A.K. Cherukuri (Editors)
Volume 39 ReRAM-Based Machine Learning H. Yu, L. Ni and S.M.P. Dinakarrao
Volume 40 E-learning Methodologies: Fundamentals, technologies and applications
M. Goyal, R. Krishnamurthi, and D. Yadav (Editors)
Volume 44 Streaming Analytics: Concepts, architectures, platforms, use cases and
applications P. Raj, C. Surianarayanan, K. Seerangan, and G. Ghinea (Editors)
Volume 44 Streaming Analytics: Concepts, architectures, platforms, use cases and
applications P. Raj, A. Kumar, V. Garcı́a Dı́az, and N. Muthuraman (Editors)
Volume 46 Graphical Programming Using LabVIEWTM: Fundamentals and advanced
techniques J.C. Rodrı́guez-Quiñonez and O. Real-Moreno
Volume 54 Intelligent Network Design Driven by Big Data Analytics, IoT, AI and Cloud
Computing S. Kumar, G. Mapp, and K. Cergiz (Editors)
Volume 56 Earth Observation Data Analytics Using Machine and Deep Learning:
Modern tools, applications and challenges S. Garg, S. Jain, N. Dube and N.
Varghese (Editors)
Volume 57 AIoT Technologies and Applications for Smart Environments. M. Alazab, M.
Gupta, and S. Ahmed (Editors)
Volume 60 Intelligent Multimedia Technologies for Financial Risk Management:
Trends, tools and applications K. Sood, S. Grima, B. Rawal, B. Balusamy,
E. Özen, and G.G.G. Gan (Editors)
Volume 63 Personal Knowledge Graphs (PKGs): Methodology, tools and applications
S. Tiwari, F. Scharffe, F. Ortiz-Rodrı́guez, and M. Gaur (Editors)
Volume 115 Ground Penetrating Radar: Improving sensing and imaging through
numerical modelling X.L. Travassos, M.F. Pantoja, and N. Ida
Nature-Inspired
Optimization Algorithms
and Soft Computing
Methods, technology and applications
for IoTs, smart cities, healthcare and
industrial automation

Edited by
Rajeev Arya, Sangeeta Singh, Maheshwari P. Singh,
Brijesh R. Iyer and Venkat N. Gudivada

The Institution of Engineering and Technology

Published by The Institution of Engineering and Technology, London, United Kingdom
The Institution of Engineering and Technology is registered as a Charity in England &
Wales (no. 211014) and Scotland (no. SC038698).
† The Institution of Engineering and Technology 2023
First published 2023

This publication is copyright under the Berne Convention and the Universal Copyright
Convention. All rights reserved. Apart from any fair dealing for the purposes of research
or private study, or criticism or review, as permitted under the Copyright, Designs and
Patents Act 1988, this publication may be reproduced, stored or transmitted, in any
form or by any means, only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction in accordance with the terms of licences issued
by the Copyright Licensing Agency. Enquiries concerning reproduction outside those
terms should be sent to the publisher at the undermentioned address:

The Institution of Engineering and Technology

Futures Place
Kings Way, Stevenage
Hertfordshire SG1 2UA, United Kingdom
www.theiet.org

While the authors and publisher believe that the information and guidance given in this
work are correct, all parties must rely upon their own skill and judgement when making
use of them. Neither the author nor publisher assumes any liability to anyone for any
loss or damage caused by any error or omission in the work, whether such an error or
omission is the result of negligence or any other cause. Any and all such liability is
disclaimed.
The moral rights of the author to be identified as author of this work have been
asserted by him in accordance with the Copyright, Designs and Patents Act 1988.

British Library Cataloguing in Publication Data

A catalogue record for this product is available from the British Library

ISBN 978-1-83953-516-1 (hardback)

ISBN 978-1-83953-517-8 (PDF)

Typeset in India by MPS Limited

Printed in the UK by CPI Group (UK) Ltd, Eastbourne
Cover credit: Starling murmuration/PHOTOSTOCK-ISRAEL/SCIENCE PHOTO LIBRARY/
Science Photo Library
 Contents

About the editors xi

Foreword xiii
Preface xv

1 Introduction to various optimization techniques 1

Trishna Saikia, Koushlendra Kumar Singh, Sergey Vityazev
and Akbar Sheikh Akbar
1.1 Introduction 1
1.2 Optimization 1
1.3 Search for optimality 2
1.4 Needs for optimization 3
1.5 A brief history of metaheuristics optimization 4
1.6 Difference between metaheuristics optimization and heuristic
optimization 5
1.7 Implications of metaheuristic optimization 5
1.8 Heuristic optimization algorithms 6
1.8.1 Constructive heuristic optimization algorithms 7
1.9 Metaheuristics optimization algorithms 7
1.9.1 Trajectory-based metaheuristic algorithms 8
1.9.2 Population-based metaheuristic algorithms 9
1.10 Theoretical analysis 11
1.11 Systematic approach for the selection of optimization algorithms 11
References 12

2 Nature-inspired optimization algorithm: an in-depth view 15

Ankit Gambhir, Ajit Kumar Verma, Ashish Payal and Rajeev Kumar Arya
2.1 Introduction to nature-inspired algorithm 15
2.2 Search for an ideal algorithm 16
2.3 Extensive review of nature-inspired algorithm 16
2.4 Analysis of nature-inspired algorithms 17
2.5 Classes of optimization algorithm 18
2.6 General classification of nature-inspired algorithms 18
2.7 Evolutionary Algorithms 19
2.7.1 Genetic algorithm 19
2.7.2 Differential evolution 20
2.8 Bio-inspired algorithms 20
vi Nature-inspired optimization algorithms and soft computing

2.8.1 Swarm intelligence-based bio-inspired algorithms 20

2.8.2 Variants of swarm algorithms 29
2.8.3 Bio-inspired but not swarm intelligence based 29
2.9 Physics- and chemistry-based algorithm 31
2.9.1 WCA 31
2.10 Application of nature-inspired optimization algorithm on
constraints engineering problem 31
2.10.1 Nature-inspired optimization algorithm (NIOA)-based
clustering routing protocols 31
2.10.2 Implementation of WCA on leach routing protocol 36
2.10.3 Nature-inspired optimization algorithm applied
in solid-state wielding 37
2.11 Conclusion 45
References 45

3 Application aspects of nature-inspired optimization algorithms 53

Abhinav Kumar, Subodh Srivastava, Vinay Kumar
and Niharika Kulshrestha
3.1 Introduction 53
3.2 Application domains of nature-inspired optimization algorithms 57
3.2.1 Optimization in image denoising 58
3.2.2 Optimization in image enhancement 59
3.2.3 Optimization in image segmentation 59
3.2.4 Optimization in image feature extraction and selection 60
3.2.5 Optimization in image classification 60
3.3 Implementation 61
3.3.1 Parameter tuning 63
3.3.2 Manual tuning 64
3.3.3 Grid search 64
3.3.4 Random search 64
3.3.5 Metaheuristic optimization 64
3.4 Constrained and unconstrained optimization 65
3.4.1 Constrained optimization 65
3.4.2 Unconstrained optimization 66
3.5 How to deal with constraints 66
3.5.1 Penalty functions 67
3.5.2 Linear penalty function 67
3.5.3 Quadratic penalty function 68
3.5.4 Constraint handling techniques 68
3.5.5 Hybrid constraint handling techniques 69
3.6 Feature selection 70
3.6.1 Feature selection based on GA 70
3.6.2 Feature selection based on PSO 71
3.6.3 Feature selection based on ACO 71
3.6.4 Feature selection based on ABC 71
 Contents vii

3.6.5 Feature selection based on CS 71

3.6.6 Feature selection based on FF 73
3.7 Practical engineering applications 73
3.8 Conclusion 75
References 77

4 Particle swarm optimization applications and implications 85

Sapna Katiyar and Aakansha Garg
4.1 Introduction to PSO 85
4.1.1 PSO elements 86
4.1.2 PSO algorithm 86
4.1.3 Standard pseudo code 88
4.1.4 PSO advantages and disadvantages 88
4.1.5 PSO applications 89
4.2 Outline of swarm intelligence 89
4.2.1 General swarm principles 95
4.3 PSO for single-objective problem 98
4.4 PSO for multi-objective problem 99
4.5 Different approaches of multi-objective PSO 99
4.5.1 Objective function aggregation approach 100
4.5.2 Objective function ordering approach 100
4.5.3 Non-Pareto, vector-evaluated approach 100
4.5.4 Algorithms based on Pareto dominance 101
4.6 Variants of PSO algorithm 103
4.6.1 Discrete PSO 103
4.6.2 Binary PSO 104
4.6.3 Adaptive PSO 104
4.6.4 Hybrid PSO 104
4.6.5 Neighborhood-guaranteed convergence PSO 104
4.6.6 Neighborhood search strategies PSO 104
4.6.7 Immunity-enhanced PSO 104
4.7 PSO in hybrid environment 105
4.8 Computational experiments 106
4.9 Convergence 106
4.10 PSO implications on image processing problems 107
4.11 PSO implications on optimum route-finding problems 108
4.12 Implementation and results 109
References 109

5 Advanced optimization by nature-inspired algorithm 119

Altaf Q.H. Badar
5.1 Introduction 119
5.2 List of nature-inspired algorithms 120
viii Nature-inspired optimization algorithms and soft computing

5.3 Optimization techniques 125

5.3.1 Anarchic society optimization (ASO) 125
5.3.2 Antlion optimizer (ALO) 126
5.3.3 Cat swarm optimization 128
5.3.4 Crow search algorithm 129
5.3.5 Cuckoo search 129
5.3.6 Mine blast algorithm 130
5.3.7 Water cycle algorithm 131
5.4 Conclusion 132
References 132

6 Application and challenges of optimization in Internet of

Things (IoT) 147
Hemlata Sharma, Shilpa Srivastava and Varuna Gupta
6.1 Introduction 147
6.2 Application of optimization in the IoT 148
6.2.1 Challenges of optimization in IoT 149
6.3 Network optimization in IoT 149
6.3.1 Types of network optimization in IOT 150
6.3.2 Algorithms of network optimization in IoT 150
6.3.3 Advantages of network optimization in IoT 152
6.3.4 Disadvantages of network optimization in IoT 152
6.4 Nature-inspired optimization in IoT 153
6.4.1 Algorithms of nature-inspired optimization in IoT 153
6.4.2 Role of nature-inspired algorithms in IoT 155
6.4.3 Advantages of nature-inspired optimization in IoT 156
6.4.4 Disadvantages of nature-inspired optimization in IoT 156
6.5 Evolutionary algorithms in IoT 157
6.5.1 Algorithms of evolutionary optimization in the IoT 158
6.5.2 Role of evolutionary algorithms in IoT 159
6.5.3 Advantages of evolutionary optimization in IoT 159
6.5.4 Disadvantages of evolutionary optimization in IoT 160
6.6 Bio-inspired heuristic algorithms in IoT 160
6.6.1 Types of bio-inspired heuristic algorithm in IoT 161
6.6.2 Role of bio-inspired heuristic algorithm in IoT 162
6.6.3 Advantages of bio-inspired heuristic algorithm in IoT 163
6.6.4 Limitations of bio-inspired heuristic algorithm in IoT 163
6.7 Load optimization in cognitive IoT 164
6.7.1 Uses of load optimization in cognitive IoT 165
6.7.2 Advantages of load optimization in cognitive IoT 165
6.7.3 Disadvantages of load optimization in cognitive IoT 166
6.8 Comparative analysis 166
References 168
 Contents ix

7 Optimization applications and implications in biomedicines and

healthcare 173
Niharika Kulshrestha, Abhinav Kumar, Vinay Kumar
and Subodh Srivastava
7.1 Introduction 173
7.2 Role of optimization algorithms in healthcare systems 175
7.3 Optimization algorithms in medical diagnosis 177
7.4 Optimization algorithms in biomedical informatics 178
7.5 Optimization algorithms in biomedical image processing 179
7.6 Optimization algorithms for ECG classification 179
7.7 Feature extraction and classification 180
7.7.1 Case study 1: feature extraction in mammography 181
7.7.2 Features are extracted by discrete wavelet transform 182
7.7.3 Features are extracted by the Gabor filter 182
7.7.4 Case study II: feature extraction in speech and pattern
recognition 183
7.7.5 Feature selection 184
7.7.6 Classification 185
7.8 Optimization algorithm-based intelligent detection of
heart disorders 186
7.9 Using predictive analytics in healthcare 187
7.10 Optimization algorithm for smart healthcare: innovations and
technologies 188
7.11 Issues and challenges in using optimization algorithms for smart
healthcare and wearables 189
References 191

8 Applications and challenges of optimization in industrial

automation 201
Devarani Devi Ningombam
8.1 Factory digitalization 201
8.1.1 Birth of factory automation 202
8.2 Product flow monitoring 203
8.2.1 Creating applications with monitoring in mind from
the start 204
8.2.2 Organize products into several categories 204
8.2.3 Include real-time tracking technologies 205
8.3 Inventory management 205
8.3.1 Time efficiency 207
8.3.2 Scalability 208
8.3.3 Accuracy 208
8.3.4 Synchronization 208
8.3.5 Quality of delivery (QoD) 209
8.4 Safety and security 209
8.5 Quality control 211
x Nature-inspired optimization algorithms and soft computing

8.6 Packaging optimization 213

8.7 Logistics and supply chain optimization 215
References 216

9 Expectations from modern evolutionary approaches for image

processing 219
R. Srikanth and V. Rajagopal
9.1 Application domains of nature-inspired optimization algorithms 222
9.1.1 Implementation 225
9.1.2 Finding optimized threshold level using harmony search
optimization algorithm 225
9.2 Results 234
9.3 Parameter tanning 239
9.4 Constrained and unconstrained optimization 241
9.5 How to deal with constraints 243
9.6 Feature selection 245
9.7 Advantages of using optimization techniques in engineering
applications 246
9.8 Conclusion 248
References 248

10 Conclusion 251
Aakansha Garg, Varuna Gupta, Vandana Mehndiratta, Yash Thakur
and Dolly Sharma
10.1 Concluding remarks 251
10.2 Challenges and potentials of bio-inspired optimization
algorithms for IoT applications 255
10.2.1 Challenges 255
10.2.2 Potentials 256
10.3 Challenges and opportunities of bio-inspired optimization
algorithms for biomedical applications 256
10.4 Recent trends in smart cities planning based on nature-inspired
computing 257
10.5 Future perspectives of nature-inspired computing 258
10.6 Bio-inspired heuristic algorithms 259
10.7 Probable future directions 260
References 261

Index 267
 About the editors

Rajeev Arya is an assistant professor in the Department of Electronics &

Communication Engineering at the National Institute of Technology, Patna, India.
His research interests cover the fields of wireless communication and security
issues, device-to-device communication based IoT System, UAV communications
in 5G and beyond networks, soft computing techniques and applications. He has
published in international journals and conferences. He is a member of IEI, IEEE
ISRD-, and IAENG. He received his Ph.D. degree in Communication Engineering
from the Indian Institute of Technology (IIT Roorkee), India.

Sangeeta Singh is an assistant professor in the Department of Electronics and

Communication Engineering at the National Institute of Technology, Patna, India.
Her research interests include soft computing techniques and applications and
beyond CMOS Devices Green Electronics steep switching transistors. She has
actively participated in technical courses, workshops, and seminars at the NITs.
She is a member of IEEE and IEEE EDS Society. She received her Ph.D. degree in
Electronics and Communication Engineering from PDPM-IIITDM Jabalpur, India.

Maheshwari P. Singh is a professor in the Department of Computer Science and

Engineering at NIT Patna, India. He has written and edited several research books.
His research interests include machine learning and wireless sensor networks, fuzzy
set, social media and security. He has actively participated in technical courses,
workshops and seminars at the IITs and NITs. He is a member of IE (Fellow), ACM
(Senior), IEEE (Senior), and ISTE (Life Member). He received his Ph.D. degree in
Computer Science and Engineering from MNNIT Allahabad, India.

Brijesh Iyer holds a Ph.D. degree in Electronics and Telecommunication

Engineering from Indian Institute of Technology, Roorkee, India. He is presently a
senior faculty member in the Department of E&TC Engineering, Dr Babasaheb
Ambedkar Technological University, India (a state technological University of
Maharashtra-India). His research interests include RF front end design for 5G and
beyond, IoT, biomedical imaging and signal processing. He has two patents to his
credit, has authored five books on cutting edge technologies, and published over 40
research papers in peer-reviewed journals and conference proceedings. He has
served as a program committee member of various international conferences. He is
a member of IEEE MTTS, ISTE, IEANG, and IETE.
xii Nature-inspired optimization algorithms and soft computing

Venkat N. Gudivada is the chairperson of and a professor in the Computer Science

Department at East Carolina University, USA. His research has been funded by
NSF, NASA, U.S. Department of Energy, US Department of Navy, US Army
Research Office, Marshall University Foundation, and West Virginia Division of
Science and Research. His research interests include cognitive computing, com-
putational linguistics/NLP, information retrieval, automated question generation,
data management and NoSQL systems, and personalization of learning. He has
published over 110 articles in peer reviewed journals, book chapters, and con-
ference proceedings. He is an IEEE Senior Member, and member of IARIA) and
the Honor Society of Phi Kappa Phi. He received his PhD degree in Computer
Science from the University of Louisiana – Lafayette, Louisiana, The USA.
 Foreword

The objective of this book series is to enlighten researchers with the current trends
in the field of blockchain technology and its integration with distributed computing
and systems. Blockchains show riveting features such as decentralization, verifia-
bility, fault tolerance, transparency, and accountability. The technology enables
more secure, transparent, and traceable scenarios to assist in distributed applica-
tions. We plan to bring forward the technological aspects of blockchain technology
for Internet of Things (IoTs), cloud computing, edge computing, fog computing,
wireless sensor networks (WSNs), peer-to-peer (P2P) networks, mobile edge
computing, and other distributed network paradigms for real-world enhanced and
evolving future applications of distributed computing and systems.
We are honoured to write the foreword of this first book in our book series.
This valuable contribution to the field of blockchain technology and social media
computing will serve as a useful reference for those seeking to deepen their
understanding of this rapidly evolving field and inspire further research and inno-
vation. The authors have provided insightful and thought-provoking perspectives
on the current state and future potential of blockchain technology in social media
computing, and we are confident that this book will be a valuable resource for years
to come.
As the world becomes more interconnected and reliant on technology, social
media has emerged as a ubiquitous tool for communication, entertainment, and
commerce. However, the centralized nature of many social media platforms has led
to concerns about data privacy, security, and content moderation. With the rise of
fake news, privacy breaches, and centralized control, the need for a more secure
and decentralized social media platform has never been greater. In recent years,
blockchain technology has emerged as a potential solution to many of these chal-
lenges, offering a decentralized and secure platform for social media computing.
The book “Blockchain Technology for Secure Social Media Computing”
provides a comprehensive overview of blockchain technology’s use in social media
computing. With contributions from experts in the field, the book covers a wide
range of topics, from the basics of blockchain technology to its applications in
social media computing and emerging trends in the field. In addition, this book
provides a comprehensive overview of the latest technologies, and applications of
blockchain in online social media. From blockchain-based security for social media
computing to emerging trends in social networking design, models, and handling,
the authors cover a broad range of topics to address the challenges and opportu-
nities of blockchain technology in the social media sphere.
xiv Nature-inspired optimization algorithms and soft computing

The chapters are written by leading researchers and academics who share their
experience and knowledge to provide readers with a practical understanding of how
blockchain technology can be leveraged to build a more secure, transparent, and
decentralized social media platform. It is an essential guide for researchers, prac-
titioners, and students who are interested in learning more about blockchain tech-
nology and its potential to transform social media computing.

Dr. Brij B. Gupta, Director, CCRI & Professor, CSIE, Asia University, Taiwan

Professor Gregorio Martinez Perez, Department of Information and

Communication Engineering, University of Murcia, Spain

Dr. Tu N. Nguyen, Department of Computer Science, Kennesaw State University,

Georgia, USA
 Preface

Recently, the field of bio-inspired computing applications for practical industrial

uses has seen a surge in research activity. Most of the time, it is impractical to solve
issues with large data sets using an exhaustive data search. Powerful tools are
available thanks to optimization methods for handling data analysis and learning
problems. Several techniques are used to design and put optimization algorithms
into practice, and they all work better. However, the search area expands expo-
nentially with problem size in a number of applications. Sets of more adaptable and
flexible algorithms are required to overcome constraints and handle larger scale
combinatorial and highly nonlinear optimization problems efficiently. Research at
the nexus of biomedical imaging, data analysis, and system development is moti-
vated by new optimization research trends that are primarily founded on machine
learning and artificial intelligence. The goal of nature-inspired computing is to
transfer the exceptional information-processing skills of the natural world to the
computational sphere. Due to its efficacy and distinctiveness, it forges close ties
with computational biology and other computing models that are motivated by
biology. Numerous fields, including engineering, medicine, industry, education,
and the military, can handle optimization in high-dimensional real-world issues
thanks to some metaheuristic search algorithms with population-based frameworks.
A significant area of computational intelligence, soft computing, and optimi-
zation in general is the study of nature-inspired optimization methods. Since
algorithms do not necessitate the fulfilment of any mathematical conditions, they
offer effective tools for issues that cannot be solved using conventional and clas-
sical mathematical methods. The Internet of Things (IoTs), which uses smart
devices, smart sensors, and machine learning, and cloud computing are other
popular subjects. They encompass all kinds of systems, including business, finance,
industry, manufacturing, management, and the environment, in addition to infor-
mation and communication technology. IoT creates a significant quantity of data
and links the physical world to the Internet. Big data processing is made easier by a
cloud computing environment, which also uses machine learning and big data
analysis to make wise choices. Additionally, the fusion of intellect and technology
enables the reduction of human effort while increasing the precision and efficacy of
the latter. By connecting innovative thinking with innovative action and innovative
execution in nearly every aspect of human life, smart devices, and systems are
revolutionizing the world. A system becomes intelligent when it has the ability to
collect data, process it, analyse it, and make decisions. Sensors are typically used
for data capture and gathering, and processing converts raw data into usable
xvi Nature-inspired optimization algorithms and soft computing

information. The task of making decisions is ultimately assisted by the analysis of

data using clever algorithms.
This edited book’s general goal is to compile the most recent chapter reviews
on the interrelated fields of biomedical imaging, soft computing, and data analysis.
This book is dedicated to the analysis, modeling, simulation, and applications in the
interdisciplinary fields of these cutting-edge technologies, from challenges and
current issues to foundations, latest research and development, and opportunities.
We will attempt to cover every facet of the interplay between metaheuristics and
complex systems, including evolutionary algorithms, swarm intelligence, etc. We
intend to offer a thorough platform for researchers to present novel methods for
handling various types of uncertainties in practical issues using cutting-edge
machine learning algorithms. We will examine the potential future work in IoTs
and its novel approach to bridging the divide between theory and day-to-day
practical problems. Additionally, we will present a unified manifesto for applied
scientists to use machine learning methods in many different engineering fields,
such as IoTs and healthcare, as well as the difficulties they will encounter in the
process. This book will investigate new learning domains and include case studies
based on both synthetic and actual data from various domains. We anticipate that it
will act as a one-stop resource for field researchers, scholars, scientists, and
industrialists.
Chapter 1 discusses various optimization techniques, search for optimality,
need for optimization, brief history of metaheuristics optimization, difference
between metaheuristics and heuristics optimization. It represents a confounded
arrangement of prerequisites to understand the implications for metaheuristic
optimization, theoretical analyses, and systematic approaches for the selection of
optimization algorithms. It is difficult to choose an optimization algorithm that will
find the best solution to the issue while considering all the limitations. There is not
a single algorithm that can fix every optimization issue.
Chapter 2 represents an in-depth view of the nature-inspired optimization
algorithms, search for an ideal algorithm, evolutionary algorithms, chemistry-based
algorithms, physics-based algorithms, and various other algorithms. This chapter
shows the path to general strategies and frameworks for efficiency improvement of
nature-inspired algorithms, applications on single objective and multi objective
problems and finally some concluding remarks.
Chapter 3 attempts to address the application aspects of nature-inspired opti-
mization algorithms. It begins with the definition of nature-inspired optimization
algorithms. In addition to that, it also deals with its implementation, and parameter
tunning. Furthermore, it also presents a study on constrained and unconstrained
optimization. Later, it covers the feature selection and practical engineering
applications of nature-inspired optimization algorithms.
Chapter 4 is dedicated for the particle swarm optimization applications and
implications. It starts with the outline of swarm intelligence, PSO for single
objective problem, and PSO for multi objective problem. It also covers the different
approaches of multi objective PSO, different variants of PSO, PSO in hybrid
environment, and many other computational experiments.
 Preface xvii

Chapter 5 covers the advanced optimization by nature-inspired algorithms. It

will give a comprehensive review of the latest optimization techniques; cat swarm
optimization, cuckoo optimization, mine blast optimization, water cycle algorithm-
based optimization, anarchic society optimization, teaching-learning-based opti-
mization, ant lion optimizer, and crow search algorithm.
Chapter 6 deals with the applications and challenges of optimization in Internet
of Things (IoTs). It starts with the network optimization in IoT and then it covers
the nature-inspired optimization in IoT. It also includes the evolutionary algorithms
in IoT and bio-inspired heuristic algorithms in IoT.
In Chapter 7, the role of optimization in healthcare systems, medical diagnosis,
biomedical informatics, biomedical image processing, ECG classification, feature
extraction and classification, and intelligent detection of disordered systems have
been discussed in detail. Different techniques have been used to explore methods
for predictive analytics in healthcare as well as innovations and technologies for
smart healthcare. This chapter provides examples of some of the problems and
difficulties that can arise when using optimization methods for wearable technol-
ogy and smart healthcare.
Chapter 8 focuses on the need of the industrial automation and its optimization
as if the companies want to keep their stock cheap while also being more respon-
sive and cutting delivery times, they must make the best use of their industrial and
logistics capacities, optimize and streamline flows throughout the supply chain, and
get the most out of their resources. It covers the various aspects of the applications
and challenges of optimization in industrial automation, factory digitalization,
product flow monitoring, inventory management, safety and security, quality con-
trol, and packaging optimization.
Chapter 9 deals with the expectations from modern evolutionary approaches
for image processing. This chapter covers how the complex optimization issues in
image processing could be solved using environmentally inspired optimization
algorithms like genetic algorithms, particle swarm optimization, and ant colony
optimization.
Finally, Chapter 10 gives the detailed concluding remarks of the study present
in the earlier chapters. It highlighted the various challenges and potentials of bio-
inspired optimization algorithms for IoT applications, challenges, and opportunities
of bio-inspired optimization algorithms for biomedical applications, recent trends
in smarty cities planning based on nature-inspired computing, and lastly future
perspectives of nature-inspired computing.
This page intentionally left blank
 Chapter 1
Introduction to various optimization techniques
Trishna Saikia1, Koushlendra Kumar Singh1,
Sergey Vityazev2 and Akbar Sheikh Akbar3

1.1 Introduction

Today in every field, there is a desire to get maximum profit with the least
investment. Efficiency and utilization with a minimum investment are key
requirements. And here comes the idea of optimization. Optimization is a process
of choosing the best available options from given alternatives which, as a result,
gives us the best solution. For example, in the design of a bridge, civil engineers
must make many decisions in various stages of construction. So, optimization is
nothing but making the best feasible decision. The goal of such decisions is either
to maximize the profit or to minimize the effort. It is a crucial tool for analyzing
systems. Both maximizing and minimizing are two categories of optimization
problems. Optimization methods are applied to many problems in various fields to
handle practical problems. It is not limited to some fields only; the idea is used
widely in all fields. With the advancement in computing techniques, optimization
has become an important part of problem solving. Many optimization techniques
have been proposed in the last few years by researchers. However, despite many
optimization algorithms, no method is suitable for all optimization problems. The
most appropriate method is selected based on the specific problem.

1.2 Optimization
Optimization is finding the condition that maximizes or minimizes the value of a
function. In mathematical terms, an optimization problem is a problem of finding
the best optimum solution from the given set of solutions. Classification of opti-
mization problems is based on several factors, such as it is based on the existence of
constraints, also based on the nature of variables, and depending on the structure of
the problem. Operations research is a field of mathematics that deals with the

1
Machine Vision and Intelligence Lab, National Institute of Technology Jamshedpur, India
2
Ryazan State Radio Engineering University, Russia
3
Leeds Beckett University, UK
2 Nature-inspired optimization algorithms and soft computing

application of different methods and techniques to solve problems by bringing the

best or optimal solutions. Mathematical optimization techniques such as linear and
nonlinear programming and network flow theory were developed during World
War II. Mathematical optimization is useful in many different fields. Some of the
fields are engineering, manufacturing, policy modeling, transportation, economics,
genetics, etc.
An optimization algorithm is used to find the best optimal solution. Many
optimization algorithms can be used based on the aim and characteristics of the
problem. If the aim is to find the derivative of a function, optimization can be
classified as gradient-based optimization and if the function requires only a random
value that gets maximized or minimized, it is classified as gradient-free optimiza-
tion. Optimization algorithms can also be classified into trajectory-based and
population-based. A trajectory-based algorithm uses one solution at a time to find
the path with continuous iteration. And population-based algorithms use multiple
solutions which, in combination, find multiple paths [1].
Looking from another perspective, optimization algorithms are also classified
as deterministic algorithms and stochastic algorithms. These are the two types of
optimization algorithms that are mostly used in most fields to perform optimiza-
tion. Deterministic algorithms use specific rules and have predefined outputs.
Stochastic algorithms use random conditions to find the optimal solution.
Algorithms with stochastic behavior are also known as metaheuristics. In the past,
stochastics was referred to as heuristics. The performance of metaheuristics is
considered better than simple heuristics. So, no algorithm is a perfect match for a
given problem. A combination of one algorithm with another algorithm is used to
design a more efficient algorithm.
The first step in the optimization process is to make a model. A model consists
of the objective function, variables, and constraints. An objective function is a
function that is maximized or minimized. It is the output. Variables are those
components that act as the parameter of a model. Variables are the input to the
model. Constraints are used to put limits on the variables. They describe the rela-
tionship between various inputs. And the second step is to determine the category
of the model. Some problems have constraints, and some have not. Problems can
also be static or dynamic.

1.3 Search for optimality

Optimization is all about finding the best solution from the given set of feasible
solutions. For example, selecting a site for an industry, finding the optimum design
of a system, or planning the best strategy to get maximum profit. Reaching
optimality is the key reason for performing optimization techniques on various
problems. Optimality is the conditions that a function must satisfy to get an opti-
mum or best solution. Local optimum is a point that gives an optimal, either
maximal or minimal solution to the optimization problem. We can find the opti-
mum point for a given problem in two separate ways. First, by optimizing the
 Introduction to various optimization techniques 3

parameters and constants of the system. And the second method is by optimizing all
variables or selected variables of the system.
There are several methods used for solving optimization problems. For
example, the hill climbing method starts from a basic approach and then moves to
improve solutions. This method solves global problems by using the information on
local optimum solutions. The optimum point is dependent on nearby points as
defined by the method.

1.4 Needs for optimization

There are several activities taking place every day around the world; these activities
can be considered as a system (whether it can be theoretical entities or physical
entities). The performance of these systems depends on the working principles of
various indices of the system, which means the efficient operation of various
indices makes the system efficient. This is only possible when the optimization of
various indices has been performed. While representing a system in a mathematical
model, algebraic variables are used to represent these indices. These variables
depend upon many factors [2]. After representing the system in a mathematical
model, some optimization techniques are applied. As a result of this, the values of
those variables are found, which either maximize or minimize the system. In the
case of maximization, it must maximize the profit or gain of the system and in the
case of minimization, it must minimize the loss or waste of the system.
Optimization is needed in many fields, such as in electrical engineering to mini-
mize total harmonic distortion in the multilevel inverter; in civil engineering,
optimization techniques are used in every step of a project life cycle; in mechanical
engineering also, optimization is used for mechanical design, similarly, in com-
puter science, there are many areas where optimization being used, etc. [3].
Let us take an example for a better understanding of the importance of opti-
mization. These days cloud computing has become the latest computing model for
a variety of applications. It can allow access to shared resources immediately after
receiving a request from clients and also discharged with a little administration. In
cloud computing, all the applications keep running on a virtual platform and all the
resources are distributed among the virtual machines [4]. Due to these kinds of
user-friendly applications, the number of clients using cloud services increased
drastically. The main goal of a cloud service provider is to create an illusion among
clients that they have an infinite number of resources but in reality, they do not
have one. So, creating this kind of illusion is possible only with the help of efficient
scheduling. So if we can develop such a scheduling algorithm that can utilize the
resources which are being provided by the service provider in a better manner or an
optimal manner then it would immensely end in a better performance by the service
provider. A good task scheduler should adapt its scheduling strategy to the chan-
ging environment and the type of tasks. Therefore, a dynamic task scheduling
algorithm is appropriate for cloud environments such as ant colony optimization,
particle swarm optimization, etc.
4 Nature-inspired optimization algorithms and soft computing

1.5 A brief history of metaheuristics optimization

People knowingly or unknowingly used metaheuristic techniques from ancient days
to solve enormous backbreaking optimization problems. Nevertheless, huge pro-
gress has been made since the establishment of the first metaheuristics concepts.
Though in the third period of 1980s, the word “metaheuristic” was first thought up,
people were applying heuristic (even metaheuristic) techniques before 1940, but
they did not formally study this. For humans, heuristics are so natural, but for
human minds having only the heuristic strategy to solve a diversity of problems
without having any prior experience is doubtful. The human mind must solve those
problems by using the derived heuristic form, which is nothing but the metaheur-
istic strategies. In the period from 1940 to 1960, several applications used heuristic
strategies. In 1958, to solve “ill-structured” problems, problems that cannot be
formulated explicitly or solved by known and feasible computational techniques,
Simon and Newell used a heuristics strategy. In 1958, their prediction slightly
emerged optimistically. From that, we can surely accept that problem solving
becomes more flexible with heuristic strategy than with exact methods. In 1963, the
evolutionary strategy was developed by Ingo Rechenberg and Hans Paul Schwefel
at the Technical University of Berlin. Moreover, in 1966, evolutionary program-
ming was developed by L.J. Fogel et al. In evolutionary programming, solutions
are represented as finite state machines, but the concept of a population and
crossover was missing there. In 1975, the real start of this field was started by John
Holland; he was the first who acknowledged the importance of both population and
crossover through his seminal book on genetic algorithms [5].
The development of evolutionary strategies as well as evolutionary program-
ming became the first turning point in the field of metaheuristics. The period from
1980 to 1990 was considered the most electrifying time for the development of
metaheuristics algorithms. In 1983, the development of simulated annealing was
the key step in the development of metaheuristics. It is an optimization technique
that was developed by S. Kirkpatrick et al. In 1986, an artificial immune system
was developed by Farmer et al., which was also considered a major step in the
development of metaheuristics. In 1980, Glover constituted the use of memory in
metaheuristics, which was published later in 1997 in his seminal book on “Tabu
Search” (Glover and Laguna). In 1992, Marco Dorigo described the innovative
work on ant colony optimization, which is a swarm intelligence-inspired search
technique, in his Ph.D. thesis on optimization and natural algorithms. In 1992, after
publishing a book on genetic programming by John R. Koza, a new field of
machine learning began.
The development of particle swarm optimization is also significant progress in
the development of metaheuristics, which was developed by James Kennedy and
Russell C. Eberhart later in 1995. The vector-based evolutionary algorithm called
differential evolution proved to be more efficient than genetic algorithms in many
applications. This vector-based evolutionary algorithm was developed by R. Storn
and K. Price later in 1997. In 2001, harmony search, a music-inspired algorithm,
was developed by Zong Woo Geem et al. Around 2002, Passino developed a
 Introduction to various optimization techniques 5

bacteria foraging algorithm, and, in 2004, the honey bee algorithm was developed
by S. Nakrani and C. Tovey. Based on this algorithm, in 2005, novel bee algorithm
was developed by D.T. Pham et al., and the artificial bee colony was also devel-
oped by D. Karaboga. An efficient cuckoo search algorithm was proposed by Yang
and Deb during the period from 2009 to 2010 [6]. Nevertheless, so many meta-
heuristics algorithms are now being developed to use for optimization.

1.6 Difference between metaheuristics optimization and

heuristic optimization

There are so many problems in the world. Among those, some problems are very
hard to solve, which we consider as NP-problems. For these problems, finding the
exact solution is very difficult by using conventional methods. Either it fails to
solve the problems or it becomes too time consuming. So, to solve these kinds of
problems, researchers used approximate methods to find approximate solutions.
Both metaheuristics and heuristics are approximate approaches that are used to
solve different optimization problems [3]. Simply, we can define a heuristic
approach as a method that uses local information to find the solution to a given
problem and a reverse metaheuristics approach as a method that uses global
information to solve any given problem. In addition to this, both these optimization
techniques differ from each other in many ways.
Naturally, heuristic methods are deterministic. On the other hand, metaheuristics
methods are an extended form of heuristic approach along with randomization.
Most of the heuristics methods are algorithmic or iterative types, but most of
the metaheuristics methods are nature-inspired; they are also iterative types. In both
methods, the nature of the solution is inexact and gives a near-optimal solution.
Metaheuristics methods are also called guided random search techniques, as it
does a random search to explore the entire search space. It is an advantage over
heuristic methods because metaheuristics methods avoid getting trapped in local
optima and always drive the search toward global optima. In metaheuristics
methods, there is a starting point for the search process, which must be specified, as
well as a stopping condition must also be set to stop the search process. Both these
methods are generally easy to implement, and usually, they can be assimilated
efficiently.

1.7 Implications of metaheuristic optimization

The application of metaheuristic optimization algorithms is growing with time, as

they are quite attractive algorithms, which are having some distinct advantages
over conventional optimization algorithms. Metaheuristic optimization algorithms
can solve complex real-world problems by finding high-quality solutions [7]. There
are some problems regarding parameter adaptation in metaheuristic optimization.
Researchers are trying to reduce this problem. After trying so many approaches,
they proposed a modification of these methods by using fuzzy logic to obtain better
6 Nature-inspired optimization algorithms and soft computing

results than the original methods [8]. Many researchers used these metaheuristic
optimization algorithms with some modifications to find more efficient solutions
throughout their research work. There is an efficient ant colony optimization
algorithm, which is also a metaheuristic optimization algorithm that can take all
types of optimization problems like linear, non-linear, and mixed integer [9]. A
good balance is needed for all metaheuristics algorithms. Every metaheuristic
algorithm follows different strategies to balance mainly two factors: one is to
identify the high-quality search space for finding better results and another factor is
not wasting unnecessary time in the already explored search space or the search
space, which does not give a promising solution.
Metaheuristic optimization methods can mainly be classified into two parts:
one is single-solution-based metaheuristics and the other is population-based
metaheuristics. Single-solution-based algorithms are also called trajectory-based
algorithms. Those algorithms work on a single solution at any point in time, while
population-based metaheuristic algorithms work on the whole population or a
whole bundle of solutions at a time. Metaheuristic algorithms are mainly inspired
by nature. For example, evolutionary algorithms are inspired by biology, particle
swarm optimization, and ant colony optimization, these swarm intelligence-based
algorithms are inspired by ant or bee colonies, and the simulated annealing algo-
rithm is inspired by physics, etc. Metaheuristic optimization algorithms can also be
classified into two categories: one is deterministic and the other is stochastic. A
deterministic optimization algorithm follows a deterministic decision approach to
solve a given optimization problem, e.g., tabu search and local search. While sto-
chastic optimization algorithms follow random rules to solve a given optimization
problem, e.g., simulated annealing and evolutionary algorithms. There are different
criteria based on which metaheuristic optimization algorithms can be classified, but
here we have discussed a few.

1.8 Heuristic optimization algorithms

Heuristic methods involve using iterative computational procedures to improve a
candidate solution toward an optimal solution with respect to a given quality
measure [10]. These methods can explore vast spaces of candidate solutions to find
near-optimal solutions at a reasonable computational cost without making any
significant assumptions about the problem being optimized. Despite their practi-
cality, heuristic algorithms cannot guarantee feasibility or optimality and, in some
cases, may need help to determine how close a feasible solution is to optimality.
Over the years, various heuristic tools have been developed to solve optimization
problems that were previously challenging or even impossible to solve. The pri-
mary advantage of these tools is that they offer quick solutions that are easy to
comprehend and implement [11].
Heuristic algorithms are useful in providing fast and practical short-term
solutions to planning and scheduling issues. However, the main drawback of
heuristic approaches is their inability to guarantee optimal solutions to problems.
 Introduction to various optimization techniques 7

While heuristic methods may offer a quick fix to specific planning or scheduling
challenges, they are only sometimes effective in delivering the best possible results.
Moreover, heuristic approaches need more flexibility to create lasting, optimal
solutions that enhance productivity and profitability.
This section provides an overview of various heuristic optimization techni-
ques, including fundamental constructive algorithms such as evolutionary pro-
gramming and greedy strategies and local search algorithms like hill climbing.

1.8.1 Constructive heuristic optimization algorithms

Constructive heuristic algorithms consist of three phases:
1. Index development, where jobs are sorted based on a specific attribute of the
problem instance [12].
2. Solution construction, where a solution is built recursively by trying
unscheduled jobs at different positions in a partial schedule until the schedule
is complete [12].
3. Improvement, where the existing solution is modified to enhance its quality [12].
Many well-known problems, such as the traveling salesman problem, are
tackled using these approaches. Construction heuristics create a complete tour by
iteratively extending a connected partial tour. Another method builds several tour
fragments and combines them to form a complete tour. One example of a con-
struction heuristic is the greedy heuristic. In this approach, all edges in Graph G are
sorted by weight, and then each edge is examined sequentially, starting from the
edge with the least weight. If the edge does not result in a cycle of less than n edges
or vertices of a degree greater than two, it is added to the current partial candidate
solution. The greedy heuristic has several variations that use different criteria to
select the edge to add. One of these is the Quick-Borůvka Heuristic. This approach
is based on Borůvka’s minimum spanning tree algorithm and sorts the vertices in G
arbitrarily. It then processes each vertex in the specified order and adds the mini-
mum weight edge that does not result in a cycle of less than n edges or vertices of a
degree greater than two. The savings heuristic, on the other hand, builds multiple
partial tours by selecting a base vertex and cyclic paths. It then combines two cyclic
paths at each construction step by removing an edge incident to the base vertex in
both paths and connecting the resulting paths into a new cyclic path. This process
continues until a single cyclic path is left. The edges to remove are chosen to
minimize the total cost. Greedy tours are at most (1 + log n)/2 times longer than
optimal tours, while the length of a savings tour is at most a factor of (1 + log n)
above the optimal length. No worst-case bounds on solution quality are known for
Quick-Borůvka tours [13].

1.9 Metaheuristics optimization algorithms

In the realm of computer science and mathematics, a metaheuristic is a heuristic or
procedure that operates on a higher level than a partial search algorithm. Its goal is
8 Nature-inspired optimization algorithms and soft computing

to find, generate, tune, or select an algorithm that can provide a satisfactory solu-
tion to optimization problems or machine learning problems. Metaheuristics
explore a subset of solutions that would otherwise be too vast to be completely
enumerated or explored. In the realm of computer science and mathematics, a
metaheuristic is a heuristic or procedure that operates on a higher level than a
partial search algorithm. These methods do not rely on many assumptions about the
problem at hand, which makes them applicable to various problems. However,
unlike iterative methods and optimization algorithms, metaheuristics cannot guar-
antee that a globally optimal solution can be found for some problems. Although
most literature on metaheuristics is experimental in nature, providing empirical
results from computer experiments with the algorithms, some formal theoretical
results are also available, typically concerning convergence and the potential to
find the global optimum.
The field of metaheuristics has produced many algorithms that claim novelty
and practical efficacy. While high-quality research can be found in the field, many
publications suffer from flaws such as vagueness, lack of conceptual elaboration,
poor experiments, and disregard for previous literature. Metaheuristic algorithms
belong to the category of computational intelligence paradigms, mainly used for
solving complex optimization problems. This section aims to provide an overview
of metaheuristic optimization algorithms. There are two main categories of meta-
heuristic algorithms: trajectory-based and population-based.

1.9.1 Trajectory-based metaheuristic algorithms

A trajectory-based metaheuristic algorithm employs a single agent or solution to
move through the search space in a piecewise manner. These algorithms accept
better solutions while also allowing for the possibility of accepting suboptimal
solutions with a certain probability. By tracing a trajectory through the search
space, these algorithms may reach the global optimum.
One example of a trajectory-based metaheuristic algorithm is simulated
annealing, which is a probabilistic optimization method that efficiently finds
solutions close to the global maximum. Simulated annealing accepts neighboring
states probabilistically, even if they result in worse outcomes, which sets it apart
from search algorithms like hill climbing and beam search.
Other trajectory-based metaheuristic algorithms include tabu search (TS),
iterated local search (ILS), and guided local search (GLS). TS is an iterative
neighborhood search algorithm that dynamically changes the neighborhood to
enhance local search [14]. By actively avoiding points in the search space already
visited, loops in search trajectories can be avoided, and local optima can be
escaped. TS combines local search (LS) and memory structures with the main
feature being the use of explicit memory. Memory is used to achieve two goals: to
prevent the search from revisiting previously visited solutions and to explore
unvisited areas of the solution space.
The ILS metaheuristic employs a LS strategy that utilizes a single solution
throughout the iterative process [15]. It is a computationally efficient method that
 Introduction to various optimization techniques 9

improves the quality of local optima, which is particularly beneficial in high-

dimensional problems. The ILS method typically starts with a randomly generated
initial solution (in binary encoding) within the predefined search space, and the
stopping criterion is often a specified number of iterations. The fundamental con-
cept of ILS revolves around the steps of local search, perturbation, and the stopping
criterion.
GLS is a metaheuristic algorithm that accumulates penalties during a search to
assist LS algorithms in escaping local minima and plateaus [16]. If the LS algo-
rithm settles in a local optimum, GLS modifies the objective function using a
specific approach. The LS then operates using an augmented objective function,
which is tailored to bring the search out of the local optimum.

1.9.2 Population-based metaheuristic algorithms

In optimization problems with a high-dimensional search space, exact optimiza-
tion algorithms are inadequate in providing a suitable solution. As the problem
size increases, the search space grows exponentially, making an exhaustive
search impractical. Furthermore, classical approximate optimization methods,
such as greedy-based algorithms, rely on several assumptions to solve problems.
However, validating these assumptions can be difficult in some cases. As a result,
meta-heuristic algorithms have been extensively developed to solve optimization
problems. These algorithms make few or no assumptions about a problem and can
explore very large spaces of candidate solutions. Population-based meta-heuristic
algorithms, such as genetic algorithm (GA), ant colony optimization (ACO),
particle swarm optimization (PSO), and artificial bee colony (ABC), are parti-
cularly useful for global searches due to their ability to explore worldwide and
exploit locally.
GA is an evolutionary algorithm that takes inspiration from Darwin’s theory of
natural selection [17]. GA broadly involves the following steps:

Initialization: The algorithm starts with a population of potential solutions to the

problem.
Selection: A fitness function is applied to evaluate the fitness of everyone in the
population, and the fittest individuals are selected for reproduction.
Crossover: The selected individuals are used to create new solutions by
exchanging genetic information through different crossover points and mating
techniques.
Mutation: To maintain diversity in the population, some individuals undergo
small mutations after crossover.

The above process continues until convergence, i.e., the population becomes
highly like its previous state. The converged population is used as the solution to
the given problem. Despite criticisms of the evolutionary theory, GA can provide
good enough solutions for complex problems, including NP-complete problems
such as the Knapsack Problem, Bin Packing, and Graph Coloring.
10 Nature-inspired optimization algorithms and soft computing

ACO is a technique used to solve computational problems, particularly those

that involve finding good paths through graphs [18]. It is based on the behavior of
real ants and uses pheromone-based communication, where ants leave pheromone
trails along paths to find food. The ACO algorithm can be described as follows:
● Every ant starts from a random vertex or city.
● For each iteration, each ant visits a city along the edge of the graph and keeps
the route in memory.
● The ants avoid previously visited cities but travel to every city in the graph by
probabilistically choosing the edge, depending on the amount of pheromone
associated with the path. Higher pheromone values increase the chance of the
ant choosing the path.
● The pheromone value is decreased by a percentage for each iteration.
● A solution is generated for every ant upon completing its visit, and pher-
omones are assigned to routes depending on the quality of the solution.
● The process is repeated until a solution converges, or the termination condition
is met.
ACO algorithms are commonly used in various fields, especially in vehicle
routing, scheduling, and set problems. The combination of artificial ants and local
search algorithms has proven effective for numerous optimization tasks.
PSO is a bio-inspired optimization algorithm that aims to find optimal solu-
tions in a solution space [19]. Unlike other optimization algorithms, PSO does not
depend on the gradient or any differential form of the objective; only the objective
function is required. PSO works by initializing a swarm of candidate solutions,
referred to as particles, and moving them around in the search space based on
simple formulae. Each particle’s movement is guided by its own best-known
position in the search space, as well as the entire swarm’s best-known position. As
improved positions are discovered, they guide the swarm’s movements in sub-
sequent iterations. This process is repeated until a satisfactory solution is found,
although it is not guaranteed to converge to an optimal solution. PSO is a
straightforward algorithm that is widely used in various fields, such as engineering,
economics, and finance.
The ABC optimization technique is a swarm intelligence algorithm that mimics
the foraging process of honeybee swarms. It is a stochastic search method that uses
the quality of nectar as a fitness evaluator for candidate solutions represented as food
sources in the search space [20]. The algorithm involves three types of bees:
employed bees, onlookers, and scouts. The number of employed bees is equal to the
number of food sources around the hive. Each employed bee goes to its assigned food
source and returns to the hive to dance on the location. If an employed bee’s food
source is abandoned, it becomes a scout and searches for a new food source.
In the ABC algorithm, the position of a food source represents a possible
solution to the optimization problem, and the nectar amount of a food source cor-
responds to the quality (fitness) of the associated solution. The initial population
(food source positions) is randomly distributed. The population then undergoes
cycles of search processes by employed, onlooker, and scout bees. Employed bees
 Introduction to various optimization techniques 11

modify the position of their food source and discover new positions. They keep the
new position if their nectar amount is higher than the previous one. Onlooker bees
evaluate the nectar information from all employed bees and choose a food source to
modify. If the new nectar amount is higher, the onlooker keeps the new position.
The abandoned sources are replaced with new sources by artificial scouts.
The ABC algorithm is a powerful optimization technique that has been suc-
cessfully used in various domains, such as image processing, classification, and
feature selection.

1.10 Theoretical analysis

The objective of optimization is to minimize the system or to maximize the system.
An optimization algorithm is used to find an optimal solution by choosing from
various solutions. Based on the behavior of problems, an appropriate algorithm is
selected. There are various criteria used to select an algorithm. These are the
computational time of the algorithm, capability to find near optimum solutions,
ease of use, etc. [21]. The property of the algorithm is also an important criterion
for selection. Some of the characteristics are the need for computational resources,
level of complexity, performance, robustness, etc. The two main stochastic meth-
ods used to solve nonconvex problems are heuristics and metaheuristics algorithms.
Both algorithms speed up the process of finding the optimal solution [22]. While in
some cases where finding an optimal solution is extremely hard, these methods
work fine. Heuristics are problem-dependent methods, whereas metaheuristics are
problem-independent methods. Heuristics find good solutions for larger problems.
And they also provide acceptable solutions for a wide range of problems. But they
do not guarantee the solution. Metaheuristic algorithms are just more powerful
mechanisms to solve optimization problems with better solutions. Most of the
metaheuristic methods are nature-inspired and iterative. But the heuristics method
is algorithmic or iterative.

1.11 Systematic approach for the selection of

optimization algorithms
Many optimization algorithms can be used based on the aim and characteristics of
the problem. But it is important to select the appropriate algorithm from the
available algorithms. The general criteria used for selecting an algorithm are based
on the problem characteristics. If the problem is convex, then we can use a simplex
algorithm, interior-point methods, and various gradient-based methods. Newton’s
algorithm is a classic example of a gradient-based optimization algorithm. It is an
unconstrained algorithm. Fletcher–Reeves and the Broyde Fletcher–Goldfarb–
Shanno (BFGS) methods are popularly used for unconstrained problems.
The Fletcher–Reeves method works well for quadratic functions. The BFGS
method uses previous information from iterations to find the optimum. The
sequential unconstrained minimization techniques (SUMT) approach can be used
12 Nature-inspired optimization algorithms and soft computing

for unconstrained problems [23]. If the characteristic of the problem is unknown,

then a direct search method is used. This type of method is based on metaheuristics
and heuristics. Algorithms with stochastic behavior are known as metaheuristics.
Various methods can be used to find the unconstrained maxima and minima of
a function. Problems having equality constraints can be optimized by using the
Lagrange multiplier method, and similarly, the problems with inequality con-
straints can be optimized by using Kuhn–Tucker conditions. Most of the methods
use numerical techniques where the approximate solutions are found by following
an iterative approach starting from a basic solution. Therefore, selecting an opti-
mization algorithm to find the best solution to the problem while satisfying all
constraints is a challenging task. There is no single algorithm available that solves
all optimization problems.

References
[1] https://en.wikipedia.org/wiki/Mathematical_optimization#Classification_
of_critical_points_and_extrema
[2] Foulds, L.R., Optimization Techniques: An Introduction, New York,
Heidelberg, Berlin: Springer-Verlag, 1981.
[3] C. Bastien and T. Marco. An Introduction to Metaheuristics for
Optimization, Springer, 2018.
[4] Arunarani, D.M. and V. Sugumaran. Task scheduling techniques in cloud
computing: a literature survey. Future Generation Computer Systems,
91407, 415, 2019.
[5] Kenneth, S., S. Marc, and G. Fred. A history of metaheuristics. cs.AI, 4 Apr
2017.
[6] Yang, X.-S. Metaheuristic optimization. Scholarpedia, 6(8), 11472, 2011.
[7] Eden, M.R., I. Marianthi, and G.P. Towler (eds.). Proceedings of the 13th
International Symposium on Process Systems Engineering – PSE, 2018,
https://doi.org/10.1016/B978-0-444-64241-7.50129-4
[8] Fevrier, V., M. Patricia, and C. Oscar. A survey on nature-inspired optimi-
zation algorithms with fuzzy logic for dynamic parameter adaptation. Expert
Systems with Applications, 41, 6459–6466, 2014, https://doi.org/10.1016/j.
eswa.2014.04.015
[9] Omid, B.-H., S. Mohammad, and H.A. Loáiciga. Meta-Heuristic and
Evolutionary Algorithms for Engineering Optimization, WILEY, 2017.
[10] Zanakis, S.H. and R.E. James. Heuristic “optimization”: why, when, and
how to use it. Interfaces, 11(5), 84–91, 1981.
[11] Lee, K.Y. and A. El-Sharkawi Mohamed (eds.). Modern Heuristic
Optimization Techniques: Theory and Applications to Power Systems,
vol. 39, New York, NY: John Wiley & Sons, 2008.
[12] Framinan, J.M., N.D.G. Jatinder, and R. Leisten. A review and classification
of heuristics for permutation flow-shop scheduling with makespan objective.
Journal of the Operational Research Society, 55(12), 1243–1255, 2004.
 Introduction to various optimization techniques 13

[13] Christofides, N. and S. Eilon. Algorithms for large-scale traveling salesman

problems. Journal of the Operational Research Society, 23(4), 511–518,
1972.
[14] Lourenço, H.R., O.C. Martin, and T. Stützle. Iterated Local Search, New
York, NY: Springer US, 2003.
[15] Voudouris, C., P.K.T. Edward, and A. Alsheddy. Guided local search. In
Handbook of Metaheuristics, Boston, MA: Springer, pp. 321–361, 2010.
[16] Glover, F. Tabu search: a tutorial. Interfaces, 20(4), 74–94, 1990.
[17] Mirjalili, S. Genetic algorithm. In Evolutionary Algorithms and Neural
Networks: Theory and Applications, Cham: Springer, pp. 43–55, 2019.
[18] Dorigo, M., M. Birattari, and T. Stutzle. Ant colony optimization. IEEE
Computational Intelligence Magazine, 1(4), 28–39, 2006.
[19] Kennedy, J. and R. Eberhart. Particle swarm optimization. In Proceedings of
ICNN’95 – International Conference on Neural Networks, vol. 4, pp. 1942–
1948, IEEE, 1995.
[20] Karaboga, D. and B. Akay. A comparative study of artificial bee colony
algorithm. Applied Mathematics and Computation, 214(1), 108–132, 2009.
[21] Entner, D., P. Fleck, T. Vosgien, et al. A systematic approach for the
selection of optimization algorithms including end-user requirements
applied to box-type boom crane design. Applied System Innovation, 2, 20,
2019. 10.3390/asi2030020.
[22] Stojanović, I., I. Brajević, P.S. Stanimirović, L.A. Kazakovtsev, and Z.
Zdravev. Application of heuristic and metaheuristic algorithms in solving
constrained weber problem with feasible region bounded by arcs.
Mathematical Problems in Engineering, 2017, Article ID 8306732, 13
pages, 2017. https://doi.org/10.1155/2017/8306732
[23] Venter, G. Review of Optimization Techniques, 2010. 10.1002/9780470686652.
eae495.
This page intentionally left blank
 Chapter 2
Nature-inspired optimization algorithm:
an in-depth view
Ankit Gambhir1, Ajit Kumar Verma2, Ashish Payal3 and
Rajeev Kumar Arya4

This chapter deals in an in-depth view of nature-inspired optimization algorithms.

It provides a comprehensive overview of optimization methods influenced by
nature. It also reflects the ideal search for the optimum nature-inspired algorithm.
The general categorization of such algorithms is demonstrated in this chapter.
Analysis and various classes of nature-inspired algorithms are revealed in detail.
Various optimization algorithms such as evolutionary algorithm and genetic algo-
rithm; bio-inspired algorithm such as particle swarm optimization (PSO), ant col-
ony optimization (ACO), bat algorithm, and crow search algorithm; bio-inspired
algorithm but not swarm intelligence such as water cycle algorithm (WCA)
and mine blast algorithm; chemistry-based algorithm; physics-based algorithm; and
many other algorithms are extensively surveyed in the general framework and
strategies for efficiency improvement are reviewed in detail in this chapter.
In addition, the distinctions between single-objective and multi-objective problems
are discussed in detail. Lastly, the application of such algorithms and concluding
remarks are presented.

2.1 Introduction to nature-inspired algorithm

Numerous scientists, academics, and academicians have been inspired by nature in

a variety of ways, making it an abundant foundation of motivation. Currently, many
novel algorithms are inspired by nature, as these were produced by emulating
nature’s inspiration [1]. Numerous applications in the real world involve the opti-
mization of diverse objectives, such as the minimizing consumption of energy and
expenses and maximizing performance and efficiency [2].

1
Delhi Technical Campus, India
2
Western Norway University of Applied Sciences, Norway
3
Guru Gobind Singh Indraprastha University, India
4
National Institute of Technology – Patna, India
16 Nature-inspired optimization algorithms and soft computing

The optimization algorithms that imitate the behavior of natural and biological
systems are referred to as nature-inspired algorithms. These algorithms are also known
as intelligent optimization algorithms or nature-inspired metaheuristic algorithms.
Heuristic algorithms generate innovative solutions through trial and error, but meta-
heuristic algorithms employ memory and other forms of knowledge and strategy. These
algorithms are well-known, efficient methods for handling a variety of challenging
optimization issues [2]. Consequently, the vast majority of natural algorithms are bio-
inspired. Over a hundred distinct algorithms and variants are existing today. There are
many classification levels for highlighting the source of inspiration. For convenience,
we employ the most evolved sources, such as biology, physics, and chemistry [3].

2.2 Search for an ideal algorithm

Real-world optimization issues are extremely exigent to address, and numerous

systems need to cope with them. For addressing such issues various optimization
tools must be employed, despite the fact that there is no assurance that an optimal
solution can be obtained. Optimization is an emerging field of study that provides
optimal solutions to complex real problems. Despite much research and advance-
ments, there are still a number of difficulties and limitations with nature-inspired
algorithms [4–6]. A few of these concerns are highlighted below:
1. Absence of a coherent mathematical framework and a lack of an in-depth
understanding of how and at what rate such algorithms may converge.
2. There are numerous nature-inspired algorithms, and their comparative analyses
are primarily based on numerical experiments or simulation models on a cer-
tain set of performance criteria; hence, it is difficult to demonstrate that such
comparisons are always objective.
3. Since the majority of research focuses on small-scale problems, it is uncertain
if these approaches may scale up to large-scale optimization problems.
4. It is also unclear what the prerequisites are for the development of swarming
and intelligent behavior.

2.3 Extensive review of nature-inspired algorithm

Some of the most significant nature-inspired algorithms are PSO based on swarm
intelligence, genetic algorithms (GA) [7,8], and differential evolution [9] based on
evolution, PSO based on flocks of a bird moving in search of food [10,11], ACO
[12] based on intelligent behavior of ants, firefly algorithm (FFA) [13] based on
intelligent behavior of firefly, cuckoo search algorithm (CSA) [14] inspired by
search intelligence of cuckoo, harmony search [15] based on the pursuit of the ideal
harmony is the goal of music, artificial bee colony (ABC) [16] chicken swarm
optimization [17] based on the foraging actions of a chicken swarm that had been
partitioned into several different subgroups based on the intelligence of bee,
glowworm algorithm (GWA) [18] mimics the behavior of glow worms. Bat
 Nature-inspired optimization algorithm 17

Table 2.1 Chronology of optimization

algorithm

Year Optimization algorithm

1975 GA
1992 ACO
1995 PSO
1996 Differential evolution
2001 Harmony search
2005 ABC
2007 FFA
2009 CSA
2010 BA
2012 WCA
2012 FPA
2013 Mine blast algorithm
2015 Water wave optimization

algorithm (BA) [19] based on behavior of bats, flower pollination algorithm (FPA)
[20] based on biological evolution of plants, WCA [21] inspired by the nature’s
process of evaporation, condensation, and precipitation. Mine blast algorithm [22]
based on the mine bomb explosion concept, water wave optimization [23] inspired
by the shallow water wave theory, to find the optimal solution these algorithms are
significantly used. Initially, the search space is a randomly initialized population.
For every round, present population is swapped by a newly created population.
Table 2.1 reflects the chronological timeline of various nature-inspired algorithms.

2.4 Analysis of nature-inspired algorithms

These algorithms possess three major features, as shown in Table 2.2 [3]:
(a) Investigation (population initiation and global/local search)
(b) Adaption and adjustment (randomness)
(c) Key operators (selection, mutation, crossover, and algorithm formulas)

Table 2.2 Analysis of nature-inspired algorithms

S. Component Role
no.
1 Population Population contain samples, performs both local and
worldwide searches
2 Randomness Adjustment as per global and local search. Also get
far from local minima
3 Selection Vital constituents for convergence
4 Mutation, crossover, and Create new solutions and their assessment iteratively
algorithmic formulas
18 Nature-inspired optimization algorithms and soft computing

2.5 Classes of optimization algorithm

Nature-inspired algorithms apply to a broader range of problems without the need
for specialized knowledge. Nature-inspired algorithms, unlike classical algorithms,
are typically designed for global search. Heuristic algorithms generate innovative
solutions through trial and error, but metaheuristic algorithms employ memory and
other forms of knowledge and strategy. These algorithms are well-known, effective
techniques for dealing with a variety of complex optimization problems [5,6].
A heuristic is a rational strategy for logically moving through the solution
space to generate an estimated solution in a stipulated time. However, on the other
hand, the prime objective of a metaheuristic is to unearth a global optimum. These
algorithms are mainly inspired by nature.
Heuristic algorithms are rather informed search technique, which analytically
investigates a state space; on the other hand metaheuristic comprises two or more
orders, such that the overall seeking performance keeps fluctuating while surveying
the state space. As opposed to the problem-independent approach used by meta-
heuristics, the heuristic approach relies on the experience of previous solutions to
arrive at the best possible one [1].

2.6 General classification of nature-inspired algorithms

Classifications of nature depend on the actual perspective and motivations. Nature-
inspired algorithms can be broken down into several different categories according
to their primary point of inspiration from the natural world (Figure 2.1).
● Evolutionary algorithms
● Bio-inspired algorithms
● Bio-inspired but not swarm intelligence-based algorithms
● Algorithms based on chemistry and physics
● Additional algorithms

Bio-inspired Other
but not SI algorithms
based

SI

Physics and Evolutionary

chemistry algorithms
based

Figure 2.1 Classification of nature-inspired algorithms

 Nature-inspired optimization algorithm 19

It is pertinent to mention that the classification presented here is not exclusive as

some algorithms could also be classified into distinct categories at the same time as
classification mainly depends on emphasis and perspective.

2.7 Evolutionary algorithms

Evolutionary algorithm (EA) replicates Darwin’s idea of evolution. They are founded on
the natural law of evolution. In each generation, the fittest individual is chosen. The
population of individuals and their representation, the fitness function, the parent selec-
tion mechanism, crossover and mutation operators, the survivor selection mechanism
(replacement), and the initialization and termination condition procedures are their
constituent parts. Evolutionary algorithms select the fittest members of the current
population to continue to the next generation. In the sections that follow, we’ll break
down some of the most notable algorithms that use natural evolution as their foundation.

2.7.1 Genetic algorithm

John Holland invented genetic algorithm (GA) in 1975 using binary representations
of individuals [7]. It mimics the process of natural selection, in which only those
organisms that can adapt to changes in their ecosystem survive and reproduce. In
straightforward terms, they assume “survival of the fittest” amid successive gen-
erations. Each generation consists of a population of people, and each person
symbolizes a point in the search space and a potential answer [8].
GA is based on the chromosomal performance and genetic structure similarity
among populations. Following are the GA foundation:
● Individuals within a population compete for resources and mates
● Individuals who are prosperous (healthiest) mate to produce more progeny
than others
● The “fittest” parent’s genes proliferate across the generation
● Thus, each succeeding generation is better adapted to its ecosystem
Each individual is assigned a fitness score that reflects his or her ability to
“compete.” The individual with the highest (or nearly ideal) fitness score is
obtained.
The GA maintains the population (chromosomes/solutions) and fitness scores.
Individuals with higher fitness scores have greater reproductive opportunities than
those with lower fitness scores. There must be room created for incoming residents
because the population size is predetermined. Therefore, once all available mating
chances in the existing population have been used up, some people pass away and
are replaced by newcomers, giving rise to a new generation [8].
Each next generation has more “superior genes” than its predecessor.
Consequently, each innovative generation has better “partial solutions” than its
predecessors. The population is converged when progenies produced have no sig-
nificant differences from progenies produced by preceding populations. The algo-
rithm is eventually transformed into a set of solutions for the problem.
20 Nature-inspired optimization algorithms and soft computing

2.7.2 Differential evolution

Storn and Price introduced the algorithm in 1996 [9]. For nonlinear optimization
issues, it is a population-based stochastic optimization approach. It only makes a
few, if any, assumptions about the core optimization issue and is capable of quickly
examining enormous strategy spaces. One of the most flexible and trustworthy
population-based search algorithms, differential evolution (DE) demonstrates the
robustness of multimodal problems.
The main benefit of standard DE is that only three control settings need to be
changed. The trial vector generation method and the control settings used have a
significant impact on the performance of DE in a specific optimization problem.
One must first select a trial vector generation approach and then adjust the control
parameters of the optimization problem in order to achieve desirable optimization
results. Finding the ideal values for the control parameters is often difficult, time-
consuming, and complex, especially for particular challenging circumstances [9]

2.8 Bio-inspired algorithms

These algorithms are based on certain admirable characteristics of the biological

system (Figure 2.2). These algorithms are further divided into two categories: bio-
inspired algorithms based on swarm intelligence and bio-inspired algorithms that
are not based on swarm Intelligence [10] (Figure 2.3).

2.8.1 Swarm intelligence-based bio-inspired algorithms

Swarm algorithms have been magnetizing great interest in the last two decades and
got enormous recognition. There are many reasons for such acknowledgment and
attention. Two major reasons are: (a) these algorithms are adaptable and

CLASSES OF NATURE-INSPIRED ALGORITHMS

Evolutionary
Algorithm
Other Algorithms
4%
15%

Bio-Inspired
24%

Physics & Chemistry

based 25%

Bio-Inspired but not

swarm intelligence
32%

Figure 2.2 Classes of bio-inspired algorithms

 Nature-inspired optimization algorithm 21

Bio-Inspired
Algorithms

Swarm Not based on

Intelligence Swarn
based Intelligence

Figure 2.3 Classes of bio-inspired algorithms

multipurpose; (b) they are very competent in solving nonlinear design problems.
This section discusses the mathematical context of the algorithms, without covering
mathematical proofs of these algorithms. Just to aid the reader to comprehend a
particular optimization algorithm. Further, techniques have been explored to
improve these algorithms for better results.
Swarm intelligence relates to the cooperative, emerging performance of sev-
eral, cooperating agents who track some simple guidelines. By the inspiration from
swarm behavior in nature, various swarm algorithms have been developing in the
last decades. The swarm agents are the core concept of the swarm algorithm.
However, every agent might be treated as obtuse. The comprehensive association
of various agents might display some self-organization pattern. Moreover, this can
act like some kind of supportive intelligence. Self-organization and disseminated
control are astonishing elements of swarm systems in nature, leading to an emit
behavior. Arise from native communication among the system agents is known as
emit behavior and it is unlikely to be accomplished by any of the agents of the
system performing unaided.
The collective behavior of natural agents, like particles, pollens of flower’s,
bees, and fish have been used to design the multi-agent system. Since 1991, swarm-
based algorithms have been proposed by adopting distinctive swarm behaviors of
ants, particles, flies, bees, cats, etc. These algorithms are formulated by mimicking
the intelligent behavior of the organisms like fishes, fireflies, ants, bees, birds, and
bats. A huge number of algorithms come under this group. Such kinds of algo-
rithms are mostly population-based stimulated by the cooperative behavior of
social insects as well as animal societies some algorithms such as PSO, FFA, and
CSA [10–19]. This section explores various types of multi-agent systems known as
metaheuristics and focuses on swarm-based algorithms only. The standard PSO
[11] adopts the collective behavior of birds, while the FFA [13] focuses on the
flashing behavior of fireflies. However, BA [19] uses the echolocation of foraging
bats. Swarm algorithms are the widely held and widely used algorithms for various
optimization problems. There are several reasons for such attractiveness: firstly,
swarm algorithms commonly share information between multiple agents; secondly,
several agents can be parallelized naturally so that complex optimization turns out
to be more practical as per the implementation scenario. Several well-known
22 Nature-inspired optimization algorithms and soft computing

swarm algorithms have been explained in this subsection to understand their brief
mathematical background and list out the well-known variant of every algorithm.
These algorithms have produced better results for most of the problems and hence
grabbed the attention of investigators, while some other algorithms have had enquiries
raised on their presence due to insignificant results or less visibility to a majority of
authors. A few of them are elaborated in next section. Most of the researched swarm
algorithms are listed below regardless of their recognition. Twenty-five algorithms are
listed in Table 2.3 and shown in Figure 2.4, mentioning name of founder/author/
pioneer, the year of development, behavior pattern of swarm, and the number of
citations from Google Scholar. However, a higher number of citations of an algorithm
do not directly translate to a better capability of that algorithm.

Table 2.3 Year-wise citation of bio-inspired swarm-based algorithms

S. no. Algorithm Author Year Citations

1 ACO Colorni et al. [12] 1991 3,989
2 PSO Eberhart and Kennedy [11] 1995 12,958
3 HS Geem et al. [15] 2001 4,017
4 ABC Karaboga [16] 2005 4,731
5 FFA Yang [13] 2008 3,035
6 CSA Yang [14] 2009 3,034
7 BA Yang [19] 2009 2,086
8 Chicken swarm optimization Xianbing Meng et al. [17] 2014 563

NUMBER OF CITATIONS

18,562
20,000
12,958
15,000

10,000
3,989 4,017 4,731 3,034 3,036
5,000 2,086
563

Particle Swarm Optimization PSO Ant Colony Optimization ACO

Differential Evolution DE Harmony Search HS
Artificial Bee Colony ABC Firefly Algorithm FFA
Cuckoo Search Algorithm CSO Bat Algorithm BA
Chicken Swarm Optimization

Figure 2.4 Number of citations

 Nature-inspired optimization algorithm 23

2.8.1.1 PSO
This algorithm is inspired by nature whose basis is swarm intelligence. Observing
the behavior of a swarm of birds seeking food leads to the manifestation of PSO
[11]. The primary objective of the method is to determine the particle locations
which will yield the optimal solution of the specified cost function. The PSO
algorithm is based on the examination of a group of birds, in which a bird that has
located food will guide the others to the food source [24].
The initialization is arbitrary, and multiple iterations are then performed, with
the particles position and velocity getting updated at the completion of each itera-
tion [25].
Algorithm for PSO

1. Initialize a S particle with arbitrarily chosen CHs

Xij(0) = (x i,j(0) , y i,j(0))
The placement of the sensor nodes
2. Determine each particle cost function:
(a) 8ki, i = 1,2, . . . ,N
Calculate the distance d (ki,CHp,q) between nodes ki and nodes CHp, q..
Assign node kii to CHp,q:
d(ki,CHp, q) = min q 1,2,..., {d(ki,CHp,q) }
(b) Calculate the cost function:
cost function = .C1 + (1-ɑ).C2
C1 = max q = 1,2,...,q { Sd(ki,CHp, q)/CHp,q }
C2=i=1NE(ki)/ q=1QE(CHp,q)
3. Find the local and global optimal conditions for each particle
4. Alter the position and velocity of the particle by
Vid (t) = W.Vid (t) + L1. H1 (Pbestid - Xid (t)+ L2.H2(Gbest-Xid(t))
Xid(t)=Xid(t-1)+Vid (t)
Or
Vid (t+1) = W.Vid (t-1) + L1.H1 (Pbestid - Xid (t-1)+ L2.H2 (Gbest-Xid(t-1))
Xid(t+1)=Xid(t)+Vid(t+1)

Where V and X represent the velocity and position of the particle, respec-
tively, t represents time, L1 and L2 stand for learning factors or acceleration
coefficients, 1 and 2 are arbitrary values between 0 and 1, W is the inertia weight
such that (0<W<1), and Pbestid and Gbest are particle’s best and global
position.
Until a satisfactory value for the global position best is reached, the procedure
of updating V and X is repeated (Gbest). The particle then uses the below equations
to recalculate Pbestid and Gbest based on the cost function.
Pbestid = Pid if cost function of Pi<cost function of Pbesti
Pbestid else
Gbest = Pid if cost function of Pi<cost function of Gbest
Gbest else
24 Nature-inspired optimization algorithms and soft computing

5. The closest (x,y) coordinates are used to re-map the new locations.
Repeat steps 2–5 until the maximum number of iterations has been reached.

2.8.1.2 ABC optimization

An example of a bio-inspired algorithm is the ABC optimization algorithm. Honey
bee behavior is used as a basis for the model [16]. Workers, observers, and scouts
are the three types of bees. Half of the colony initially consists of worker bees,
while the other half consists of observer bees. Bees with a job are tasked with
finding new food sources in the area and spreading the word to their fellow bees.
An observer bee follows a new arrival from the area where food is plentiful. The
term “Scout bee” refers to a certain species of bee that searches for food in a
random fashion. When the nectar supply from a new source is greater than that of
the previous source, the worker bees will switch to the new source. Observer bees,
after exhausting all other options, settle on the food source with the highest like-
lihood and relocate it. The contest between the new and old cost functions is
maintained, with the best solution being saved until the maximum number of cycles
has been reached [24,26]. The procedure for the algorithm is as follows.
1. Generate the starting population Yi, i = 1...M
2. Determine the population
3. Set the cycle’s value equal to 1; repeat
(a) For each employed bee in the workforce,
(i) Create new solutions Ki by
Kij= yij + aij (yij-ykj)
(ii) Where a is an arbitrary number between [1,1], Ki is candidate
solution, yi is present solution, and yk is neighbor solution
j = {1,2 . . . N}
(iii) Examine the price function
(iv) Execute the greedy selection process
(b) For every onlooker bee
(i) Select a solution yi on the basis of Gi
(ii) Gi = CSi / k = 1MCSk
(iii) Where CSi is the cost function of solution i, M = number of food
source = number of employed bee
(iv) Find new solutions Ki by
(v) Kij = yij + aij (yij - ykj)
(vi) Determine the cost function and apply the greedy selection pro-
cedure. If a solution exists, we can replace it with the scout’s
finding, i.e.,
(vii) yij = yj min + arb(0,1) (yj max  yj min)
j = {1,2 . . . N}
where, N is the dimension of the solution vector.
4. Construct the most effective solution until.
Cycle = cycle + 1, cycle = maximum cycle number.
 Nature-inspired optimization algorithm 25

2.8.1.3 Chicken swarm optimization

It is also a bio-inspired swarm-based algorithm that was developed by studying the
foraging actions of a chicken swarm that had been partitioned into several different
subgroups, as shown in Figure 2.2. Every single individual in each subgroup
worked their way closer and closer to the ideal one [17]. Rooster involves the
uppermost position in the hierarchical order, as they possess the finest food-
searching capacity. The roosters will likewise battle with different chickens who
attack their region. Hens possess the subsequent position and they obey their
gathering mate roosters in the food-hunting procedure [27]. There is likewise
rivalry among the chickens in the food-searching process. Chicks possess the most
reduced position and they follow their mom while searching for food. In this
way, the chickens arrange themselves in the food-hunting procedure. This natural
conduct of chickens is related to the competence to optimize the problem. To
encourage the depiction of CSO, we have to idealize the conduct of the group [28].
1. Few gatherings of chickens in the swarm of chickens. Each gathering of
chickens incorporates a rooster, various hens, and various chickens.
2. As per the fitness value of the chicken, chicken is affirmed as a rooster, a hen,
and a chicken.
8
< 1 If fm  fk
m2 ¼ fk  fm
: jfmþ 2 j
e else
We expect some portion of the finest fitness value as a rooster, i.e., leader of a
gathering, and the piece of the most noticeably worst fitness as the chicken,
and the different as a hen, hen arbitrarily joining a gathering, chicken arbi-
trarily discover a hen as a mother.
3. The chicken follows a gathering of partners to discover food, they may need to
avert other chicken to have their possessed food, and chicken are additionally
liable to eat food, which has not been found by them but other chicken [29].
The algorithm depends on the following presumptions:
● The quantity of hens is most elevated in the gathering.
● All the hens do not hatch eggs together. Only some hens are mother hens.
● The mother hens are chosen arbitrarily from the arrangement of hens.
● There is a difference between the number of chicks and hens.
Mathematical modeling is shown below.
Rooster position update:
  2
m;n ¼ X m;n  1 þ randn 0; m
X tþ1 t

Where k [ [1, Number of rooster node], k 6¼ m and Xmn denotes the position of
rooster m in nth dimension during t and t + 1 itertion, randnð0; m2 Þ is Gaussian
random variable with variance m2 (which is a small constant) and mean value is 0.
fm is the fitness value for the subsequent rooster m. [ is a low value constant.
26 Nature-inspired optimization algorithms and soft computing

Hens’ position update:

   
tþ1
Xm;n ¼ Xm;n
t
þ K1randn Xp1;n
t
 Xm;n
t
þ K2randn Xp2;n
t
 Xm;n
t

Where
fmfp1
K1 ¼ ejfmþ2j

K2 ¼ efp2fm

p1 is the index of a rooster, and p2 is a chicken from the group that can be a hen or
rooster and a uniform random number between 0 and 1 generated by randn.
Chicks’ position update:
 
tþ1
Xm;n ¼ Xm;n
t
þ AB Xq;nt
 Xm;n
t

t
Xm;n denotes position of chicken’s mother, AB is arbitrary number between 0
and 2 to obtain the optimum position.

2.8.1.4 ACO
Dorigo first suggested ACO in his PhD thesis in 1992 [12]. The swarming nature
established by ACO is modeled after ant colonies. It provides a sufficient solution
to the engineering optimization issue. Ants typically leave behind pheromone trails
that lead one another to food sources as they scout the area. The stimulated or
computer-generated “ants” also keep track of their locations and the qualities of
their resolves, enabling additional ants in later simulation cycles to find better
solutions. Artificial “ants” navigate within a constrained location while displaying
all potential resolutions to find the best possible resolutions [30]. It demonstrates
the ACO pseudo code, which calls for two-step calculations.
Create ants solutions as shown in equation below.
To understand the mathematics of ACO, let us consider probability pkab of an
ant to move from a state to b state.
 a  b 
Mab Nab
pkab ¼ P   b 
M a Nac
ZE ac

Update pheromones:

X
j
new
Mab old
E:Mab þ 4Mab
k

k¼1

where Mab is the amount of pheromone deposited for a state transition ab, E is the
pheromone evaporation coefficient.
DMabk is the amount of pheromone deposited by kth ant.
 Nature-inspired optimization algorithm 27

Pseudo code of ACO

Objective function (a), a = (a1, ... ... an)T
Describe pheromone rate evaporation rate g
while (Max_number_of_iteration)
{
forloop_over_all_n_dimensions
{
Generate novel solutions using Equation
Calculate the newly generated solutions
Revise pheromone using Equation (3.2)
}
Find out the current best
}
Output the finest results and pheromone dispersal.

2.8.1.5 FFA
The flashing pattern of tropical fireflies serves as the basis for the FFA. It was created
in 2007 by Xin-She Yang [13]. This algorithm applied three fundamental rules. First,
because fireflies have no sexual preference, they are all attracted to each other.
Second, when the distance between fireflies grows, the attraction, which is propor-
tionate to the luminosity, decreases. The less bright firefly will therefore move
toward the more brightly lit one for any pair of flashing fireflies. Last but not the
least, a firefly’s brightness is determined by how the objective function is perceived,
and brightness is correlated with the cost of the objective function [31–33].

Pseudo Code of Firefly Algorithm

Objective function (a), a = (a1, ... ... ad)t
Reset algorithm parameters:
MaxGen: the maximal number of generations
@: the light absorption coefficient
w: the particular distance from the light source
d: the domain space
Generate the initial population of fireflies or ai (i = 1, 2 , . . . , k)
calculate the intensity of light ı́i at ai via f(ai)
while (t < Max_number_of_iteration)
28 Nature-inspired optimization algorithms and soft computing

{
for i = 1 to k
{
for j = 1 to k
{
If (ı́j > ı́i
{
Move firefly i towards j by using Equation
Attractiveness changes as a function of distance w via e@w
2

Assess novel solutions and update the intensity of light

}
}
reorder the fireflies according to intensity of light and find the value current fitness
function;
}
Output the best results.
ati ¼ ati þ b0e@w ðaj  aiÞ þ aei
2

2.8.1.6 CSA
This method is based on the cuckoo parasites’ breeding behavior. These natural
processes are ingrained by the algorithm [14].
In this algorithm, a few following presumptions are made:
Assumption 1: One egg is laid by a cuckoo at a time, and it is then dropped into a
different nest each time.
Assumption 2: The nest that produces the highest caliber eggs will be passed
down to succeeding generations.
Assumption 3: The number of host nests is fixed, and each nest holds only one egg.
Assumption 4: There is a pa chance that the host bird will find a foreign egg.
A cuckoo egg stands for a fresh solution, and each nest represents a solution.
The algorithm’s goal is to employ novel, possibly superior solutions [34,35].
Pseudo Code of CSO:

Objective function (x), x = (x1, . . . . . . xn)T

Generate initial population of n host nests xi
while (t < Max_number_of_iteration)
{
t=t+1
 Nature-inspired optimization algorithm 29

Get a cuckoo that has been randomly created using Levy’s flights as a solution, and
then assess its fitness function (i.e., Fi).
Fi > Fj
Replace j by new solution.
The worst nets are uncontrolled to a certain degree (pa), and new ones are generated.
Maintain the best nests. g*. }
Output the best results g* and xt+1.

2.8.2 Variants of swarm algorithms

Table 2.4 is a presented list of swarm algorithms. Using a variety of methods, such as
parameter tweaking, Gaussian distribution, Levy’s distribution, chaotic distribution, and
others, one can find a better version of any swarm algorithm. New swarm algorithms that
consider changing swarm behaviors have been developed since the original swarm
algorithm was created and are still in use today [36]. Additionally, several aspects of the
above-discussed methods are already present in these algorithms [37]. For instance,
Levy’s flight distribution is used by the FPA and the BA, respectively [38]. Any algo-
rithm that has been hybridized has been done so primarily to increase performance. The
performance of swarm algorithms has been improved using a variety of methods.
The mathematical structure of the algorithm has a significant impact on the
choice of a swarm algorithm and the corresponding improvement technique [39]. For
instance, whereas FPA can be searched both locally and globally, it may become
snared in a local optimum. The use of chaotic distribution can make this better.
Following an examination of swarm algorithms and their variations, the fol-
lowing conclusion has been reached:
1. Only a small number of the many developed swarm algorithms have been the
focus of additional study.
2. Techniques including parameter adjustment, Levy’s distribution, chaotic distribution,
and Gaussian distribution have all been shown to improve algorithm performance.
3. The mathematical structure of the algorithm has a significant impact on the
performance of a swarm algorithm and each version.

2.8.3 Bio-inspired but not swarm intelligence based

2.8.3.1 FPA
In order to improve the ability of optimization algorithms to explore in global space, Yang
employed the pollination concept in optimization problems in 2012 [40]. Pollination is the
process by which plants reproduce and are fertilized. It entails transferring pollen, a sticky
substance made from the flower’s male stamen, to the stigma (i.e., top of female part
called PISTIL). Pollinating agents like the wind, animals (such as insects, birds, and bats),
or the plant itself can spread pollen. Stamen and filament are collectively referred to as
such, and pistil refers to the stigma, style, ovary, and ovules [41]. In nature, flower pol-
lination serves to promote plant reproduction and the survival of the fittest.
Table 2.4 Summary of swarm-based nature-inspired algorithms

S. no. Algorithm Author Year Behavior

1 ACO Colorni et al. [12] 1991 In order to solve discrete optimization issues or problems with
unknown solutions, a colony of artificial ants collaborates or
works together.
2 PSO Eberhart and Kennedy 1995 Interaction of the particles in the swarm.
[10]
3 HS Geem et al. [15] 2001 The pursuit of the ideal harmony is the goal of music.
4 ABC Karaboga [16] 2005 Based on honey bees’ clever foraging techniques.
5 FFA Yang [13] 2008 A firefly’s flash serves as a signaling mechanism to draw in
other fireflies.
6 CSA Yang [14] 2009 Inspired by the eco-location of bats. Inspired by the obligate
brood parasitism of some cuckoo species, which deposit their
eggs in the nests of other host birds (of different species).
7 BA Yang [19] 2009 Inspired by the eco location of bats.

8 Chicken swarm algorithm Xianbing Meng et al. [17] 2014 Dividing the hens into several subgroups and watching the
foraging behavior of the swarm.
9 Glow worm swarm Krishnanand and Ghose 2005 Mimics the behavior of glow worms.
optimization (GWSO) [18]
10 Wasp swarm algorithm Pinto [68] 2007 Wasp colonies establish a hierarchy among themselves through
(WSA) communication among the individuals.
11 Lion optimization algorithm Wang et al. [69] 2012 Lion prides promote the principle of group living and
(LOA) evolution.
12 Spider monkey algorithm Bansal et al. [70] 2014 Stimulates spider monkeys’ foraging behavior.
13 Whale optimization Mirjalili and Lewis [71] 2016 Mimics humpback whales’ social behavior.
algorithm (WOA)
14 Wolf pack algorithm Hu-Sheng et al. [72] 2014 To find prey on the Tibetan Plateau, the wolf pack joins forces
and collaborates closely.
 Nature-inspired optimization algorithm 31

With reference to natural processes, FPA can be idealized as follows: self-pollination

is seen as local pollination, whereas cross pollination is regarded as global pollination.
Both local and global pollination are governed by a switch probability, p. The value of p
is affected by variables like wind and pollinator availability, among others.

2.9 Physics- and chemistry-based algorithm

2.9.1 WCA
Engineering problems have been sorted using conventional search methods for the
past few decades. Numerous promising algorithms have produced promising out-
comes. In 2012, Hadi Eskandar et al. [21] proposed WCA. The fundamental prin-
ciple of WCA is inspired by nature and motivated by the natural water cycle process,
which includes how streams and rivers are formed and water drains to the sea. It
simulates the evaporation, condensation, and precipitation processes that make up the
fundamental hydrological cycle. In the atmosphere, water evaporates to create
clouds, which then further condenses to produce rain and release water back onto the
land. We refer to this as precipitation. If a river does not enter the sea, every time two
streams converge, they create a new, higher order stream. On constrained benchmark
issues and engineering design optimization, WCA is also validated. The idea of
evaporation rate for rivers and streams is also considered in WCA in some articles.
WCA’s comprehensive MATLAB codes were presented in [42]. Additionally, it is
used to solve issues that are confined and uncontrolled. The stated results were highly
encouraging. WCA’s comprehensive review were presented in [43].

2.10 Application of nature-inspired optimization

algorithm on constraints engineering problem
2.10.1 Nature-inspired optimization algorithm (NIOA)-
based clustering routing protocols
Engineering constraint issues were reformulated throughout the past 10 years using
optimization methods that were influenced by nature. Many encouraging computa-
tions have found potential results (Figure 2.5). Numerous bio-inspired algorithms
have been discovered to execute competitively on a variety of optimization problems
in recent study.

CLUSTERING &
OPTIMAL COVERAGE ROUTING

APPLICATION OF NATURE-
INSPIRED OPTIMIZATION
ALGORITHMS

LOCALIZATION DATA AGGREGATION

Figure 2.5 Application of NIOA in WSN

32 Nature-inspired optimization algorithms and soft computing

Tillet et al. [44] have provided the first study based on the application of
evolutionary algorithms on cluster head selection. PSO has been used for cluster
head (CH) detection.
It is an evolutionary algorithm technique in which test answers are allowed to
interact and work together to find the optimal answer to the problem at hand, much like
a natural swarm of birds. The wireless sensor network (WSN) algorithm can be broken
down into three logical parts. The clusters are produced in the first section. Each
cluster’s cluster heads are chosen in the second section. The third section assigns
cluster heads to each of the network’s nodes. Each phase in this situation uses PSO. A
succinct overview of the use of PSO in WSN to handle a variety of difficulties,
including optimal deployment, node localization, clustering, and data aggregation, was
presented by Kulkarni et al. [45]. An energy-efficient CH selection approach based on
PSO has been presented by Srinivasa Rao et al. [46], employing an effective particle
model and fitness function. They have taken into account a number of factors,
including residual energy, intra-cluster distance, and sink distance, for the energy
efficiency of the suggested approach. The technique was tested using several WSN
scenarios, sensor node counts, and CHs. By utilizing the PSO-based methodology
within the cluster as opposed to the base station, Buddha Singh et al. [47] had trans-
formed it into a semi-distributed method. Their suggested method’s main goal is to
locate the head nodes close to the cluster density’s center. On the basis of the optimal
CH position, they have also calculated the estimated distance travelled by packet
transmission from node to BS. Additionally, they have computed the anticipated
number of retransmissions along the path to CH and examined the impact of link
failure. Last but not the least, power computation is performed to assess the energy
savings, and the results are compared with LEACH-C and procedures.
Latiff et al. [48] defined a new cost function in their research titled “Energy-
aware clustering for wireless sensor networks using particle swarm optimization.”
The proposed protocol creates clusters evenly distributed over the whole WSN field
by choosing a high-energy node as the CH.
To determine the maximum quality thresholding potential for cluster generation in
WSNs, Jenn-Long Liu et al. [49] developed a genetic algorithm (GA)-based entirely
adaptive clustering procedure. The results of the simulation show that the most likely
distribution closely reflects the analytical conclusions made by the authors’ updated for-
mulas. Given that it uses the most effective probability, power-efficient clustering algo-
rithm, the suggested LEACH-GA method exceeds LEACH in terms of network lifetime.
Complex combinatorial problems can be solved using swarm intelligence tech-
niques like ACO [50]. It is a cleverly devised approach for resolving challenging
integration issues. To resolve the routing issue in sensor networks, Tiago Camilo,
Carlos Carreto, and others [51] discovered the usefulness of the Ant Colony
Optimization metaheuristic. The energy-efficient ant-based routing (EEABR) pro-
tocol finds paths between nerves and node nodes using “lightweight” ants, which can
be improved by grade and energy levels. The experimental findings demonstrated
that the set of rules produces excellent outcomes in various WSNs
A novel WSN router operating protocol was introduced by Selcuk Okdem et al.
[52]. The protocol is implemented by offering a potent multi-course data transfer
technique and using the ACO algorithm to enhance route paths and obtain
 Nature-inspired optimization algorithm 33

dependable connections in the event of node faults. The suggested approach is con-
trasted with EEABR, an event-based template-based set of fashion ant-based rules.
The results show that their approach significantly reduces the amount of energy
consumed, which is utilized as a performance indicator for unique-sized WSNs.
The improvement of artificial bee colonies is suggested by Karaboga et al.
[53]. Bees are divided into three groups: employed, spectator, and scout. In the
beginning, one type of bee—an employed bee—makes up half of the colony, and
another—an onlooker bee—makes up the other half. Employed bees scout out
potential food sources in the area and advise curious bees of new food sources.
According to the knowledge they have, spectator bees search for a fresh one in the
vicinity of food supplies that are nearby. A scout bee is a bee that searches at
random. Employed bees switch to a new food source if the new food source’s
nectar content is greater than the previous one. After conducting all possible sear-
ches, observer bees select the food source with the highest probability and move it
to a new location. Up until the maximum number of cycles is reached, the cost
function of the new one is compared to the old one, and the best solution is stored.
In their article, “A comprehensive survey: ABC algorithm and applications,”
Dervis Karaboga et al. [54] gave a thorough assessment of the application of ABC
in numerous disciplines.
PSO with time variable acceleration constants, hierarchical PSO with time
varying acceleration constants, PSO with time varying inertia weight, and PSO
with supervisor student mode are the four PSO variations that Guru et al. [55]
presented for energy aware clustering. In wireless sensor networks, four PSO
techniques were used to group sensors into clusters. Based on integration criteria, it
operates. The inertia w weight drops in line during the correction from 0.9 during
the first multiplication to 0.4 during the last repetition. The PSO determines the
inertia gravity in the acceleration times, and the acceleration constants c1 and
c2 change sequentially throughout the multiplication, causing the particles to travel
in huge steps at first but then smaller steps in each repetition.
Various authors have applied PSO in various areas of WSN.
● PSO in node deployment
● PSO in node localization
● PSO in energy aware clustering
● PSO in data aggression
Brief summaries of various well-known nature-inspired algorithms used in clus-
tering routing protocols are shown in Table 2.5.

In Node Deployment

Application of PSO In Node Localization

in WSN
In Energy Aware Clustering

In Data Aggregation

Figure 2.6 Application of PSO in WSN

34 Nature-inspired optimization algorithms and soft computing

Table 2.5 Summary of some of the prominent NIOA applied on clustering routing
protocols

S. no. Author Optimization Available information Year

algorithm
used
1 Tillet et al. [73] PSO Reduces energy consumption 2003
and speeds up data transfer
Use of basic PSO
2 Latiff et al. [74] PSO 1. Established a new cost 2011
function.
2. Clusters are evenly
distributed
3 Guru et al. [55] Variant of PSO Only the physical distance 2005
between the CH and node is
considered for application of
PSO
4 Yubin Xu et al. PSO Improve the LEACH-based 2011
[75] method. Improve clustering and
identify the CH node
5 Thilagavathi PSO 1. Performance evaluation of 2015
et al. [76] LEACH in WSNs
2. Optimal number of clusters
and cluster head
6 Cao et al. [77] PSO Shown an analysis of CH 2008
election for multi-hop
communication
7 Jin et al. [78] GA Clustering teaches about the 2003
sensor network’s communication
distance
8 Hussain et al. GA 1. Energy-efficient clusters for 2009
[79] routing in WSNs are proposed
2. Examined the progressive
loss of energy
9 Norouzi et al. GA A mathematical formula that 2011
[80] increases the protection against
lifetime risk was proposed
10 Luo et al. [81] QGA Quantum GA (QGA)-based QoS 2010
routing protocol, a proposed
technique, is effective with best-
effort traffic
11 Seo et al. [82] 2D GA 1. Using GA, achieved the ideal 2009
cluster formation
2. Developed an algorithm known
as the location-aware two-
dimensional GA (LA2D-GA).
12 Salehpour et al. ACO Presented a useful routing 2008
[83] strategy for large-scale WSNs
based on clusters
13 Camilo et al. ACO Suggested EEABR algorithm to 2006
[51] increase the lifetime of WSNs
(Continues)
 Nature-inspired optimization algorithm 35

Table 2.5 (Continued)

S. no. Author Optimization Available information Year

algorithm
used
14 Mao et al. [84] ACO To achieve network load 2013
balancing, it is advised that high-
density WSNs use a non-uniform
clustering protocol that is energy
cognizant
15 Almshreqi et al. ACO Projected an algorithm for 2012
[85] balanced energy consumption
16 Dervis ABCO The ABCO algorithm’s search 2012
Karaboga et al. characteristics, which are very
[86] effective and optimizing, were
used to enhance CH selection
Compared with the other
protocols based on LEACH
and PSO
17 Kumar et al. Two-step A two-step ABCO algorithm is 2015
[87] ABCO suggested for issues with data
clustering. K-means algorithm is
utilized to construct the initial
food source position for the ABC
algorithm
Comparison of the traditional
ABCO algorithm and its
variations, including ACO, PSO
GA, and K-means algorithms
18 Pawandeep ABCO It was suggested that an ABC- 2016
et. al. [88] based energy efficient adaptive
clustering hierarchy protocol
might effectively increase the
network life and stability of
WSNs
19 Potthuri et al. Differential 1. First, initializing the popula- 2016
[89] evolution (DE) tion using the opposite
and simulated point method
annealing 2. Self-adaptive control para-
algorithm meters for the DE are taken
into consideration
3. Cost function evaluation
4. Distribution of CH
20 Kaur et al. [90] Hybrid PSO & 1. Clustering protocol for WSNs 2017
ACO based on ACOPSO
2. Discovered the quickest route
between CH and BS
3. Compressive sensing is used
to reduce packet size
36 Nature-inspired optimization algorithms and soft computing

For the performance study of appropriate routing protocols and clustering

strategies in sensor networks, a number of methodologies have been put out in the
literature. Although the literature has documented substantial advancements in
wireless sensor network life extension, there are still a number of problems that
need to be resolved. While numerous studies and methodologies have shown strong
agreement for improving both, due to various problems and difficulties of infra-
structure less mobile wireless networks, it is more difficult to provide a
suitable routing protocol and clustering scheme than it is in fixed networks, and this
problem is perpetually open in infrastructure less ad hoc wireless networks.
A thorough study of the literature is followed by the identification of the
research gaps.
Due to node mobility, a major problem in wireless sensor networks is that
communication must be conducted while the network architecture is changing
(dynamically). In a wireless sensor network, each node communicates with its
neighbors directly within its transmission range, but it must use intermediary
nodes to send its messages hop by hop with the aid of a routing protocol in order to
communicate with nodes that are located outside of this range. In order to find
effective and appropriate routing protocols for various sensor network applications,
research is required.
Common network management systems available today were often created to
operate on wireless networks that were static or relatively small. Numerous ad hoc
network applications involve large networks with thousands of nodes, like in the
case of sensor networks and tactical networks, which are exceedingly difficult to
administer. Clustering is a crucial field of research for wireless sensor networks
since it increases network size, lowers overheads, and makes the network more
manageable when there is both extreme mobility and a significant number of
mobile nodes. Therefore, a new clustering technique must be found and created in
order to divide the enormous network into smaller, more manageable networks. In
clustering, the gateway node facilitates inter-cluster communication while the
cluster head oversees and retains recent routing information. The loss of stored
routing information and reduced battery life of nodes, as well as the modification of
the route between two nodes, has all been linked to the frequent switching of the
cluster head and gateway node. This instability also affects the performance of the
routing protocol and the cluster structure. To choose a new CH, communication
overhead in the form of message switching is necessary.
To retain the cluster head and gateway node change as little as possible and to
increase the stability of the cluster structure, it is crucial to prevent the loss of
routing information. The majority of the clustering literature reviews rely on
choosing the cluster head and gateway node independently. Therefore, there is a
huge need to find innovative clustering methods for wireless sensor networks.

2.10.2 Implementation of WCA on leach routing protocol

Another possible interpretation of the consonance is as follows: in this scenario, the
initial population consists of all nodes, and fitness function is determined by their
 Nature-inspired optimization algorithm 37

individual energies [56]. During the startup phase, the LEACH Protocol is used to
elect CH, and the node that has the highest energy is known as the sea or CH. More
energy-efficient nodes are clustered at the cluster’s epicenter (where the rivers
empty into the sea), while less efficient nodes are located further from the center
(streams farthest from the sea). After some iterations, the energy of each node will
be computed, and its new positions will be decided. Rivers are selected from nodes
with the optimal amount of energy, while less energetic nodes are rejected. We
select high-energy nodes as rivers and low-energy ones as streams. Similarly, to
how the locations of rivers and streams are revised at each energy assessment, a
new CH is selected at regular intervals. The same procedure is carried out repeat-
edly till the maximum number of rounds has been reached [57]. In the end, the
WCA reduces the expense of selecting the best node that can serve as the cluster
head in the WSN’s hierarchical routing protocol. The acquired findings showed
that the suggested technique outperforms traditional LEACH in terms of results.

2.10.3 Nature-inspired optimization algorithm applied in

solid-state wielding
This study examines how the mechanical characteristics of the AA 6082-T6 joint
are affected by underwater friction stir welding (UFSW) process parameters and to
simulate this process using a variety of evolutionary optimization strategies. The
impact of differences in these parameters on the ultimate tensile strength (UTS in
MPa), percentage elongation (E in%), and impact strength (IS in J) of the welded
joint was studied using experimental measurements and records. To model this
UFSW process, two evolutionary optimization algorithms were used, namely, PSO
and FFA. In these simulations, the cost function was an artificial neural network
with two layers that resembled a nonlinear function, and it was used to forecast the
values of the response variables, UTS, E, and IS, which had previously been
measured experimentally. These simulations involved multiple experiments with
various randomly chosen data sets, and the accuracy of each simulation was then
compared. Results showed that, while predicting the values of the response vari-
able, the FFA-based simulation outperformed the PSO with the least mean square
error (MSE). For UTS, E, and IS, it was found that the minimum MSE for the FFA-
based simulation was as low as 0.009%, 0.004%, and 0.017%, respectively.
There are now several joining techniques available, but joining aluminum alloys
(AAs) remain difficult due to their high heat conductivity and low melting point [58].
During fusion welding (FW) of AAs, porosity, solidification cracking, residual strains,
oxidation, and other discontinuities are also formed, reducing the aesthetics and prac-
tical weld qualities. Friction stir welding (FSW) is an effective method for combining
similar and dissimilar materials because it does not require melting and recasting the
material [59–61]. This 1990s-era method relies heavily on frictional heating and severe
plastic deformation of the foundation. Because of the thermo-mechanical cycles cre-
ated during the process, welding some AAs with FSW causes the strengthening pre-
cipitates to dissolve or become coarser, which reduces the mechanical performance of
the joints [62]. In earlier investigations, external cooling was used to solve this issue.
38 Nature-inspired optimization algorithms and soft computing

One such procedure that uses water to adjust the temperature in joints is UFSW [63].
Underwater, which may be still or may be continuously flowing across the surface of
the samples to be welded, is where the welding is done. Underwater, which may be still
or may be continuously flowing across the surface of the samples to be welded, is
where the welding is done. Due to the extensive circulation and high heat-capturing
capabilities of the water, the coarsening and dissolving of precipitates are controlled
[60]. Over FSW and FW, UFSW has the capacity to offer better mechanical qualities
and fine-grain structural features. Heat circulation, material movement, and intermix-
ing during UFSW are significantly influenced by the various welding settings as well
as varied joint designs and tool features. This alters the joint’s mechanical character-
istics and produces change in the macro- and microstructural aspects [59–63].
Therefore, choosing the best parameter combination and optimizing it are crucial for
enhancing the mechanical properties of the joints. Sharma et al. [64] investigated
multi-response optimization using the TOPSIS technique for different FSW joints.
They discovered that the rotating speed has a significant impact on micro hardness and
UTS. Palanivel et al. [65] used central composite face-centered factorial during dis-
similar joining of AAs utilizing FSW to establish empirical modeling between the
FSW parameters and the UTS. The literature that is now accessible also shows a dearth
of research into the simulation of the UFSW process for using various evolutionary
optimization techniques. Motivated by these knowledge gaps, the following goals were
systematically designed and accomplished by various authors: (i) to investigate the
impact of three UFSW input parameters on the UTS, E%, and IS of marine grade AA
6082-T6 joints using the Taguchi’s L18 standard orthogonal array and (ii) to compare
the simulation abilities of these optimization algorithms.

2.10.3.1 Experimental details

Butt welding AA 6082-T6 with dimensions of 200  50  3 mm was done using
UFSW. The tool’s material of choice was H 13 tool steel. In this work, tri-flute
profiled tools with a 6-mm pin diameter and 2.7-mm pin height were used. The
welding tools were angled at 20 degrees. An UFSW-specific vertical milling machine
was used to accomplish a single-pass run. The tool shoulder diameter (A), rotational
speed (B), and traverse speed ranges for three UFSW parameters were established
using the trial run results (C). Table 2.6 lists the UFSW process parameters and their
levels. Taguchi’s L18 orthogonal array (OA) was used to construct the 18 runs that
are listed in Table 2.7. Using wire EDM, the tensile and impact parts were cut. A
tensometer was used to measure the UTS and% E at room temperature and at a cross-
head speed of 2 mm/min. The IS was computed using the Charpy Impact test.

Table 2.6 Parameter (designation)

Parameter (designation) Unit Level 1 Level 2 Level 3

A mm 17 20 –
B rpm 710 900 1,120
C mm/min 50 63 80
 Nature-inspired optimization algorithm 39

In order to achieve better results from the chosen prediction mode, the para-
meter values were normalized to a range of (0, 1).
Five independent experiments were carried out for each set of input parameters to
analyze the data using each optimization procedure. The following sections include the
findings of top three runs (experiments) for selected optimization algorithm.
The values of numerous parameters or constraints that must be established in
each method, such as the number of iterations, population or swarm size, and
number of elements or particles taken into account, are described in Table 2.8. This
table’s values were determined through a series of exploratory trials. Table 2.9 also

Table 2.7 Taguchi L18 OA

S. no. A B C
1 17 710 50
2 17 710 63
3 17 710 80
4 17 900 50
5 17 900 63
6 17 900 80
7 17 1,120 50
8 17 1,120 63
9 17 1,120 80
10 20 710 50
11 20 710 63
12 20 710 80
13 20 900 50
14 20 900 63
15 20 900 80
16 20 1,120 50
17 20 1,120 63
18 20 1,120 80

Table 2.8 Additional parameters that are stated for each algorithm

S. no. PSO parameters Firefly parameters

1 Max. no. of iterations 25,000 Max. no. of iterations 5,000
2 Size of the swarm 96, 192,96a Size of the swarm 96, 192,96a
3 Number of particles 32, 36a, 32 Number of particles 32, 36a, 32
(elements) (elements)
3 Correction factor (C1) 1.2 Coefficient of light 1.50
absorption
4 Correction factor (C1) 2.4 Value of attraction 2.30
coefficient
5 Weight damping ratio 0.99 Mutation coefficient 0.25
a
Refers to separate values for UTS, E, and IS response variables.
40 Nature-inspired optimization algorithms and soft computing

Table 2.9 Values of input and the response parameter

Input parameters Response parameters

S. no. A B C UTS (MPa) E (%) IS (J)
1 17 710 50 224 11.12 8
2 17 710 63 228 9.67 7
3 17 710 80 223 8.67 6
4 17 900 50 226 12.11 9
5 17 900 63 236 9.78 7
6 17 900 80 241 9.76 7
7 17 1120 50 222 12.89 8
8 17 1,120 63 231 9.87 8
9 17 1,120 80 235 9.67 7
10 20 710 50 208 12.67 8
11 20 710 63 214 12.33 8
12 20 710 80 219 11.54 7
13 20 900 50 227 12.89 9
14 20 900 63 232 12.87 8
15 20 900 80 232 10.43 9
16 20 1,120 50 218 12.89 9
17 20 1,120 63 221 9.34 8
18 20 1,120 80 228 12.56 7

shows that different parameters were required to simulate UTS, E%, and IS. To
simulate the E% parameter, the model required more parameters than the other
response parameters, particularly for the PSO and FFA algorithms [66,67].

2.10.3.2 Results of the PSO algorithm

It is evident from the results obtained (Table 2.6) that PSO correctly anticipated the
response (UTS) values. For all 18 samples, the total range of highest and lowest
error was found as 0.015–1.575% (Table 2.10). However, the error was just around
2% for the majority of the samples.
The projected values were reasonably similar to the experimental values
anticipated for (%E) as shown in Table 2.11. The difference or error between
forecasted and experiments values for each of the 18 samples range between
0.126% and 8.370%.
The data in Table 2.12 showed that projected values are really quite close to
experimental values. In other instances, the divergence is found as 0.002%. The
total measured range of error was 6.607–0.002%.

2.10.3.3 Results of the FFA

The results of the best three experiments are shown in Tables 2.13–2.15.
Table 2.13 findings demonstrate that the FFA was successful in estimating the
UTS values. For all 18 samples, the total range of maximum and minimum error
was found to be 0.009–0.836%. However, this discrepancy was only around 1% for
the vast majority of the samples.
 Nature-inspired optimization algorithm 41

Table 2.10 Experimental and predicted values from PSO algorithm for the UTS
parameter

S. Experimental Prediction Prediction Prediction Max. Min.

no. value (MPa) 1 (MPa) 2 (MPa) 3 (MPa) error (%) error (%)
1 224.000 223.383 222.219 222.326 0.795 0.275
2 228.000 228.038 229.088 228.280 0.477 0.017
3 223.000 223.319 223.530 223.187 0.238 0.084
4 226.000 228.472 227.938 229.560 1.575 0.857
5 236.000 235.632 237.248 237.304 0.553 0.156
6 241.000 239.596 238.930 240.229 0.859 0.320
7 222.000 221.604 221.878 220.295 0.768 0.055
8 231.000 229.784 228.657 228.840 1.014 0.526
9 235.000 236.359 236.694 234.623 0.721 0.160
10 208.000 208.045 208.132 208.924 0.444 0.021
11 214.000 214.448 215.555 214.523 0.726 0.209
12 219.000 218.602 217.574 218.429 0.651 0.182
13 227.000 227.035 225.704 225.067 0.851 0.015
14 232.000 229.778 231.452 229.147 1.230 0.236
15 232.000 233.908 232.561 233.538 0.822 0.242
16 218.000 217.204 218.092 218.930 0.427 0.042
17 221.000 222.109 222.755 223.480 1.122 0.502
18 228.000 227.625 226.993 228.237 0.442 0.104
Experiment wise 0.377 0.517 0.590
mean error (%)

Table 2.11 Experimental and predicted values from PSO algorithm for the %
elongation parameter

S. Experimental Prediction Prediction Prediction Max. Min.

no. value (%) 1 (%) 2 (%) 3 (%) error (%) error (%)
1 11.120 11.172 11.540 11.177 3.774 0.471
2 9.670 9.658 9.092 9.500 5.982 0.126
3 8.670 8.731 8.901 8.852 2.670 0.698
4 12.110 11.806 12.047 12.068 2.509 0.348
5 9.780 10.361 10.355 10.156 5.942 3.846
6 9.760 9.148 9.168 9.049 7.280 6.069
7 12.890 13.130 12.763 13.024 1.862 0.985
8 9.870 9.211 9.363 9.226 6.672 5.141
9 9.670 10.179 10.132 10.412 7.678 4.777
10 12.670 12.693 12.652 12.586 0.662 0.142
11 12.330 12.263 12.448 12.573 1.970 0.546
12 11.540 11.457 11.398 11.342 1.713 0.718
13 12.890 13.058 12.490 13.087 3.106 1.305
14 12.870 12.443 12.440 12.092 6.048 3.321
15 10.430 11.097 11.121 11.239 7.760 6.392
16 12.890 12.649 13.125 12.701 1.868 1.469
17 9.340 10.114 10.074 10.122 8.370 7.864
18 12.560 11.900 11.961 11.823 5.864 4.767
Experiment wise 3.205 3.647 3.693
mean error (%)
42 Nature-inspired optimization algorithms and soft computing

Table 2.12 Experimental and predicted results from PSO algorithm for the IS
parameter

S. Experimental Prediction Prediction Prediction Max. Min.

no. value (J) 1 (J) 2 (J) 3 (J) error (%) error (%)
1 8.000 7.958 7.927 7.964 0.912 0.455
2 7.000 7.084 6.953 6.948 1.199 0.669
3 6.000 5.947 6.098 5.949 1.629 0.858
4 9.000 8.908 8.977 8.701 3.322 0.251
5 7.000 7.159 7.210 7.353 5.047 2.276
6 7.000 6.964 6.905 7.144 2.054 0.520
7 8.000 8.037 7.984 8.473 5.912 0.195
8 8.000 7.948 8.045 7.471 6.607 0.569
9 7.000 7.012 6.930 6.846 2.195 0.177
10 8.000 8.056 8.010 7.981 0.704 0.122
11 8.000 7.912 8.034 8.026 1.096 0.323
12 7.000 7.050 6.945 7.053 0.783 0.718
13 9.000 9.054 9.000 9.177 1.972 0.002
14 8.000 7.881 8.098 8.176 2.196 1.219
15 9.000 9.035 8.951 8.780 2.448 0.394
16 9.000 8.988 9.061 8.785 2.387 0.139
17 8.000 8.041 7.771 7.799 2.869 0.509
18 7.000 6.972 7.098 7.378 5.406 0.393
Experiment wise 0.764 0.979 2.524
mean error (%)

Table 2.13 Experimental and predicted values from FFA for the UTS parameter

S. Experimental Prediction 1 Prediction 2 Prediction 3 Max. Min.

no. value (MPa) (MPa) (MPa) (MPa) error (%) error (%)
1 224.000 223.653 222.611 223.296 0.620 0.155
2 228.000 228.248 228.403 228.626 0.275 0.109
3 223.000 223.095 223.021 223.030 0.042 0.009
4 226.000 226.218 226.503 227.264 0.559 0.096
5 236.000 235.878 236.475 235.701 0.201 0.052
6 241.000 240.811 240.730 240.332 0.277 0.079
7 222.000 222.216 222.616 221.626 0.278 0.097
8 231.000 230.804 230.233 229.934 0.461 0.085
9 235.000 235.068 235.203 236.032 0.439 0.029
10 208.000 207.785 208.088 207.973 0.104 0.013
11 214.000 214.653 214.683 214.364 0.319 0.170
12 219.000 218.457 218.640 218.701 0.248 0.137
13 227.000 227.116 227.179 227.433 0.191 0.051
14 232.000 231.503 231.231 230.569 0.617 0.214
15 232.000 232.570 232.906 232.709 0.391 0.246
16 218.000 217.399 217.497 217.312 0.316 0.231
17 221.000 221.650 221.353 222.847 0.836 0.160
18 228.000 227.856 227.622 227.267 0.322 0.063
Experiment wise 0.141 0.218 0.308
mean error %
 Nature-inspired optimization algorithm 43

Table 2.14 Experimental and predicted values from FFA for the % elongation
parameter

S. Experimental Prediction 1 Prediction 2 Prediction 3 Max. Min.

no. value (%) (%) (%) (%) error (%) error (%)
1 11.120 11.809 11.615 11.520 6.194 3.598
2 9.670 9.278 9.248 9.303 4.360 3.797
3 8.670 8.669 8.861 8.827 2.205 0.015
4 12.110 11.798 11.870 12.074 2.576 0.297
5 9.780 10.083 10.012 10.257 4.880 2.374
6 9.760 9.604 9.368 9.364 4.059 1.595
7 12.890 12.890 12.921 12.853 0.288 0.004
8 9.870 9.487 9.627 9.332 5.448 2.462
9 9.670 9.945 10.024 9.954 3.665 2.844
10 12.670 12.625 12.653 12.690 0.355 0.131
11 12.330 12.210 12.191 12.043 2.331 0.972
12 11.540 11.636 11.103 11.106 3.785 0.829
13 12.890 12.619 12.770 12.599 2.254 0.935
14 12.870 12.608 12.639 12.544 2.532 1.794
15 10.430 10.630 11.617 11.645 11.645 1.920
16 12.890 13.003 12.901 12.974 0.877 0.082
17 9.340 9.928 9.806 9.986 6.920 4.984
18 12.560 12.226 11.841 12.005 5.728 2.659
Experiment wise 2.350 3.095 3.433
mean error %

Table 2.15 Experimental and predicted results from FFA for the IS parameter

S. Experimental Prediction 1 Prediction 2 Prediction 3 Max. Min.

no. value (J) (J) (J) (J) error (%) error (%)
1 8.000 7.903 8.010 7.783 2.715 0.121
2 7.000 6.991 6.862 7.085 1.968 0.123
3 6.000 6.178 6.001 5.954 2.958 0.017
4 9.000 9.111 9.005 8.947 1.239 0.050
5 7.000 6.877 7.183 7.250 3.568 1.757
6 7.000 6.965 6.849 7.151 2.160 0.501
7 8.000 7.994 8.004 7.985 0.184 0.054
8 8.000 8.011 8.024 8.079 0.992 0.135
9 7.000 6.986 6.982 6.827 2.476 0.206
10 8.000 7.997 8.029 8.028 0.360 0.036
11 8.000 8.260 7.980 8.009 3.249 0.109
12 7.000 6.791 7.040 7.045 2.987 0.570
13 9.000 8.944 8.927 8.978 0.814 0.250
14 8.000 8.146 8.242 8.274 3.426 1.824
15 9.000 8.758 8.871 8.565 4.839 1.436
16 9.000 9.006 9.055 9.050 0.608 0.071
17 8.000 7.919 7.677 7.814 4.040 1.007
18 7.000 7.184 7.266 7.156 3.797 2.235
Experiment wise 1.295 1.246 1.633
mean error (in %)
44 Nature-inspired optimization algorithms and soft computing

Table 2.14 shows that the predicted values obtained were quite close to the
expected experimental results. The entire error range was discovered to be between
0.004% and 11.645%.
The results show that projected values are very close to experiment data. In
other cases, the variation can be as low as 0.002%. The complete error range was
discovered to be between 0.017% and 4.839%.
The evaluation of prediction error for the various UFSW process parameters
and the different optimization strategies is shown in Tables 2.16–2.18.
The three runs of the PSO- and FFA-based simulations are shown along with the
maximum, lowest, and mean values for the maximum and minimum prediction
errors for the UTS. Table 2.16 shows that the FFA-based simulation forecasted the
data the most accurately, with an average maximum and minimum error of 0.541%

Table 2.16 Maximum and minimum prediction errors for the UTS

PSO algorithm Firefly algorithm

Max. Min. Avg. Max. Min. Avg.

error error error error error error
Highest value among 1.575 0.857 1.216 0.836 0.246 0.541
experiments
Lowest value among experiments 0.238 0.015 0.126 0.042 0.009 0.026
Mean value among experiments 0.762 0.223 0.492 0.361 0.111 0.236

Table 2.17 Maximum and minimum prediction errors for the elongation %

PSO algorithm Firefly algorithm

Max. Min. Avg. Max. Min. Avg.

error error error error error error
Highest value among experiments 8.370 7.864 8.117 11.645 4.984 8.314
Lowest value among experiments 0.662 0.126 0.394 0.288 0.004 0.146
Mean value among experiments 4.541 2.721 3.631 3.895 1.738 2.816

Table 2.18 Maximum and minimum prediction errors for the impact strength

PSO algorithm Firefly algorithm

Max. Min. Avg. Max. Min. Avg.

error error error error error error
Highest value among experiments 6.607 2.276 4.442 4.839 2.235 3.537
Lowest value among experiments 0.704 0.002 0.353 0.184 0.017 0.100
Mean value among experiments 2.708 0.544 1.626 2.354 0.583 1.469
 Nature-inspired optimization algorithm 45

and 0.026%, respectively. The overall mean prediction error remained as low as
0.236% in this simulation. Experiment 1 demonstrated the FFA algorithm’s best
mean error values, with a mean error percentage of 0.141%. The top-performing PSO
experiment had average errors of 0.377% and 1.142%, respectively. The FFA-based
simulation achieved a small error value of 0.009% over the three trials mentioned.
In this case, the PSO and FFA-based models both performed reasonably well;
however, the FFA-based simulation yielded the lowest mean average prediction
error of 2.816%. Experiment 1 of the FFA technique had the lowest mean error
values, with an average error of 2.35%. The FFA-based prediction model achieved
an error value of 0.004%.
Table 2.18 shows the mean, maximum, and minimum impact strength predicted
errors for tests performed with PSO and FFA. The FFA-based simulation performed
admirably in this case as well, with prediction error% ranging from 0.1 to 3.537 and the
lowest mean average prediction error of 1.469%. In experiment 2, the mean error values
were 1.246%, with an average error percentage of 1.246%. The FFA-based prediction
model achieved an error value of 0.017% across the three experiments shown in
Table 2.12. PSO produced a minimum error value of 0.002% for impact strength.

2.11 Conclusion
For the last few decades, traditional search algorithms have been used to sort out
engineering problems. Many promising algorithms have found potential results. This
chapter presented an in-depth view of nature-inspired optimization algorithm and its
application in engineering constraint problems. Various nature-inspired optimization
algorithms were considered for extensive review. Different classes of such algo-
rithms are presented. Some of the popular and widely used NIOA were elaborated
and their applications on different domains are also discussed. Application on clus-
tering routing protocols and solid-state wielding is elucidated in details. The WCA-
based clustering routing protocol eventually lessens the cost of determining the best
node as the cluster head. This novel energy-efficient clustering mechanism sustains
the network lifetime. Similarly, PSO and FFA evolutionary optimization algorithms
are used to simulate the UFSW process. The results strongly suggest that the FFA
technique can be successfully used to simulate a highly efficient model for predicting
the parameters of the UFSW process. The findings and results drawn from the study
of NIOA are of great relevance to researchers of optimization algorithms and also
help in optimizing engineering problems and practical application hence therefore
has been adequately detailed. There are a variety of research directions that could be
well thought of as a helpful expansion of this analysis.

References
[1] A. Singh, S. Sharma, and J. Singh, “Nature-inspired algorithms for wireless
sensor networks: a comprehensive survey,” Comput. Sci. Rev., vol. 39, no.
100342, p. 100342, 2021.
46 Nature-inspired optimization algorithms and soft computing

[2] K. G. Dhal, S. Ray, A. Das, and S. Das, “A survey on nature-inspired opti-

mization algorithms and their application in image enhancement domain,”
Arch. Comput. Methods Eng., vol. 26, no. 5, pp. 1607–1638, 2019.
[3] X.-S. Yang, “Nature-inspired optimization algorithms: challenges and open
problems,” J. Comput. Sci., vol. 46, no. 101104, p. 101104, 2020.
[4] S. R. Kumar and K. D. Singh, “Nature-inspired optimization algorithms:
research direction and survey,” arXiv [cs.NE], 2021.
[5] N. E. L. Y. Kouba and M. Boudour, “A brief review and comparative study
of nature-inspired optimization algorithms applied to power system control,”
in Natural Computing for Unsupervised Learning, Cham: Springer
International Publishing, 2019, pp. 35–49.
[6] R. Rajakumar, P. Dhavachelvan, and T. Vengattaraman, “A survey on nature
inspired metaheuristic algorithms with its domain specifications,” in 2016
International Conference on Communication and Electronics Systems
(ICCES), 2016.
[7] D. E. Goldberg and J. H. Holland, “Genetic algorithms and machine learn-
ing,” Mach Learn, vol. 3, no. 2, pp. 95–99, 1988.
[8] J. Koza, “Genetic programming as a means for programming computers by
natural selection,” Stat. Comput., vol. 4, no. 2, pp. 87–112, 1994.
[9] R. Storn and K. Price, “Differential evolution: a simple and efficient heur-
istic for global optimization,” J. Global Optim, vol. 11, pp. 341–359, 1997.
[10] X.-S. Yang, S. Deb, S. Fong, X. He, and Y.-X. Zhao, “From swarm intelli-
gence to metaheuristics: nature-inspired optimization algorithms,” Computer
(Long Beach Calif.), vol. 49, no. 9, pp. 52–59, 2016.
[11] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proceedings
of ICNN’95 – International Conference on Neural Networks, 2002.
[12] M. Dorigo, “Optimization, learning, and natural algorithms,” PhD Thesis,
Politecnico di Milano, 1992.
[13] X.-S. Yang, “Firefly algorithms for multimodal optimization,” in Stochastic
Algorithms: Foundations and Applications, Berlin, Heidelberg: Springer
Berlin Heidelberg, 2009, pp. 169–178.
[14] X.-S. Yang and S. Deb, “Cuckoo search via Lé vy flights,” in Proceedings of
World Congress on Nature and Biologically Inspired Computing, USA:
IEEE Publications, 2009, pp. 210–214.
[15] Z. W. Geem, J. H. Kim, and G. V. Loganathan, “A new heuristic optimization
algorithm: Harmony Search,” Simulation, vol. 76, no. 2, pp. 60–68, 2001.
[16] D. Karaboga, “An idea based on honey bee swarm for numerical optimiza-
tion”, Technical report-tr06; Erciyes university, engineering faculty, com-
puter engineering department,” Tech. Rep, 2005.
[17] X. Meng, Y. Liu, X. Gao, and H. Zhang, “A new bio-inspired algorithm:
chicken swarm optimization,” in Lecture Notes in Computer Science, Cham:
Springer International Publishing, 2014, pp. 86–94.
[18] K. Krishnanand and D. Ghose, “Detection of multiple source locations using
a glowworm metaphor with applications to collective robotics”,” in Swarm
Intelligence Symposium, 2005, pp. 84–91.
 Nature-inspired optimization algorithm 47

[19] X.-S. Yang, “A new metaheuristic bat-inspired algorithm,” in Nature

Inspired Cooperative Strategies for Optimization (NICSO 2010), Berlin,
Heidelberg: Springer Berlin Heidelberg, 2010, pp. 65–74.
[20] S. Łukasik and P. A. Kowalski, “Study of flower pollination algorithm for
continuous optimization”, in Intelligent Systems, Springer, 2014.
[21] H. Eskandar, A. Sadollah, A. Bahreininejad, and M. Hamdi, “Water cycle algo-
rithm – a novel metaheuristic optimization method for solving constrained engi-
neering optimization problems,” Comput. Struct., vol. 110–111, pp. 151–166, 2012.
[22] A. Sadollah, A. Bahreininejad, H. Eskandar, and M. Hamdi, “Mine blast
algorithm for optimization of truss structures with discrete variables,”
Comput. Struct., vol. 102–103, pp. 49–63, 2012.
[23] Y.-J. Zheng, “Water wave optimization: a new nature-inspired metaheur-
istic,” Comput. Oper. Res., vol. 55, pp. 1–11, 2015.
[24] A. Gambhir and A. Payal, “Analysis of particle swarm and artificial bee
colony optimisation-based clustering protocol for WSN,” Int. J. Comput.
Syst. Eng., vol. 5, no. 2, p. 77, 2019.
[25] A. Gambhir, A. Payal, and R. Arya, “Analysis of PSO based clustering
protocol in assorted scenarios of WSN,” in Communication and Computing
Systems, London: CRC Press, 2019, pp. 400–405.
[26] A. Gambhir, A. Payal, and R. Arya, “Performance analysis of artificial bee
colony optimization based clustering protocol in various scenarios of WSN,”
Procedia Comput. Sci., vol. 132, pp. 183–188, 2018.
[27] A. Gambhir, A. Payal, and R. Arya, “Chicken swarm optimization algorithm
perspective on energy constraints in WSN,” in 2020 IEEE 7th Uttar Pradesh
Section International Conference on Electrical, Electronics and Computer
Engineering (UPCON), 2020.
[28] S. Deb, X.-Z. Gao, K. Tammi, K. Kalita, and P. Mahanta, “Recent studies on
chicken swarm optimization algorithm: a review (2014–2018),” Artif. Intell.
Rev., vol. 53, no. 3, pp. 1737–1765, 2020.
[29] W. Osamy, A. A. El-Sawy, and A. Salim, “CSOCA: chicken swarm opti-
mization based clustering algorithm for wireless sensor networks,” IEEE
Access, vol. 8, pp. 60676–60688, 2020.
[30] C. Blum, “Ant colony optimization: introduction and recent trends,” Phys.
Life Rev., vol. 2, no. 4, pp. 353–373, 2005.
[31] S. Tripathi, “A brief review of Firefly Algorithm: application in structural
optimization problem,” J. Inst. Eng., vol. 15, no. 2, pp. 183–191, 2019.
[32] V. Kumar and D. Kumar, “A systematic review on firefly algorithm: past, present,
and future,” Arch. Comput. Methods Eng., vol. 28, no. 4, pp. 3269–3291, 2021.
[33] J. Nayak, B. Naik, D. Pelusi, and A. V. Krishna, “A comprehensive review
and performance analysis of firefly algorithm for artificial neural networks,”
in Nature-Inspired Computation in Data Mining and Machine Learning,
Cham: Springer International Publishing, 2020, pp. 137–159.
[34] A. B. Mohamad, A. M. Zain, and N. E. Nazira Bazin, “Cuckoo search
algorithm for optimization problems—a literature review and its applica-
tions,” Appl. Artif. Intell., vol. 28, no. 5, pp. 419–448, 2014.
48 Nature-inspired optimization algorithms and soft computing

[35] S. Walton, O. Hassan, K. Morgan, and M. R. Brown, “A review of the

development and applications of the cuckoo search algorithm,” in Swarm
Intelligence and Bio-Inspired Computation, New York, NY: Elsevier, 2013,
pp. 257–271.
[36] Y. Song, Z. Chen, and Z. Yuan, “New chaotic pso-based neural network
predictive control for nonlinear process”,” IEEE Transactions on Neural
Networks, vol. 18, pp. 595–601, 2007.
[37] J. Cai, X. Ma, L. Li, Y. Yang, H. Peng, and X. Wang, “Chaotic ant swarm
optimization to economic dispatch,” Electric Power Systems Research,
vol. 77, no. 10, pp. 1373–1380, 2007.
[38] K. Gopal Dhal, “A chaotic lévy flight approach in bat and firefly algorithm
for gray level image enhancement,” Int. J. Image Graph. Signal Process.,
vol. 7, no. 7, pp. 69–76, 2015.
[39] B. Alatas, “Chaotic harmony search algorithms,” Applied Mathematics and
Computation, vol. 216, no. 9, pp. 2687–2699, 2010.
[40] X.-S. Yang, “Flower pollination algorithm for global optimization,” in
Unconventional Computation and Natural Computation, Berlin, Heidelberg:
Springer Berlin Heidelberg, 2012, pp. 240–249.
[41] A. Fouad and X.-Z. Gao, “A novel modified flower pollination algorithm for
global optimization,” Neural Comput. Appl., vol. 31, no. 8, pp. 3875–3908,
2019.
[42] A. Sadollah, H. Eskandar, H. M. Lee, D. G. Yoo, and J. H. Kim, “Water cycle
algorithm: a detailed standard code,” SoftwareX, vol. 5, pp. 37–43, 2016.
[43] M. Nasir, A. Sadollah, Y. H. Choi, and J. H. Kim, “A comprehensive review
on water cycle algorithm and its applications,” Neural Comput. Appl.,
vol. 32, no. 23, pp. 17433–17488, 2020.
[44] J. Tillett, T. Rao, F. Sahin, and R. Rao, “Darwinian particle swarm optimi-
zation,” in Proceedings of the 2nd Indian International Conference on
Artificial Intelligence (IICAI-05), 2005.
[45] R. V. Kulkarni and G. K. Venayagamoorthy, “Particle swarm optimization
in wireless-sensor networks: a brief survey,” IEEE Trans. Syst. Man Cybern.
C Appl. Rev., vol. 41, no. 2, pp. 262–267, 2011.
[46] P. C. S. Rao, P. K. Jana, and H. Banka, “A particle swarm optimization
based energy efficient cluster head selection algorithm for wireless sensor
networks,” Wirel. Netw., vol. 23, no. 7, pp. 2005–2020, 2017.
[47] B. Singh and D. K. Lobiyal, “A novel energy-aware cluster head selection
based on particle swarm optimization for wireless sensor networks,” Hum.-
Centric Comput. Inf. Sci., vol. 2, no. 1, p. 13, 2012.
[48] N. M. A. Latiff, C. C. Tsimenidis, and B. S. Sharif, “Energy-aware clustering
for wireless sensor networks using particle swarm optimization,” in 2007
IEEE 18th International Symposium on Personal, Indoor and Mobile Radio
Communications, 2007.
[49] J.-L. Liu and C. V. Ravishankar, “LEACH-GA: genetic algorithm-based
energy-efficient adaptive clustering protocol for wireless sensor networks,”
Int. J. Mach. Learn. Comput., pp. 79–85, 2011.
 Nature-inspired optimization algorithm 49

[50] T. Liao, T. Stützle, M. A. Montes de Oca, and M. Dorigo, “A unified ant

colony optimization algorithm for continuous optimization,” Eur. J. Oper.
Res., vol. 234, no. 3, pp. 597–609, 2014.
[51] T. Camilo, C. Carreto, J. S. Silva, and F. Boavida, “An energy-efficient ant-
based routing algorithm for wireless sensor networks,” in Ant Colony
Optimization and Swarm Intelligence, Berlin, Heidelberg: Springer Berlin
Heidelberg, 2006, pp. 49–59.
[52] S. Okdem and D. Karaboga, “Routing in wireless sensor networks using an
ant colony optimization (ACO) router chip,” Sensors (Basel), vol. 9, no. 2,
pp. 909–921, 2009.
[53] D. Karaboga and B. Basturk, “Artificial bee colony (ABC) optimization
algorithm for solving constrained optimization problems,” in Lecture Notes
in Computer Science, Berlin, Heidelberg: Springer Berlin Heidelberg, 2007,
pp. 789–798.
[54] D. Karaboga, B. Gorkemli, C. Ozturk, and N. Karaboga, “A comprehensive
survey: artificial bee colony (ABC) algorithm and applications,” Artif. Intell.
Rev, vol. 42, no. 1, pp. 21–57, 2014.
[55] S. M. Guru, S. K. Halgamuge, and S. Fernando, “Particle swarm optimisers for
cluster formation in wireless sensor networks,” in 2005 International Conference
on Intelligent Sensors, Sensor Networks and Information Processing, 2005.
[56] A. Gambhir, A. Payal, and R. Arya, “Water cycle algorithm based optimized
clustering protocol for wireless sensor network,” J. Interdiscip. Math.,
vol. 23, no. 2, pp. 367–377, 2020.
[57] A. Gambhir, R. Arya, and A. Payal, “Performance analysis of SEP, I-SEP,
PSO and WCA-based clustering protocols in WSN,” Int. J. Intell. Eng.
Inform., vol. 7, no. 6, p. 545, 2019.
[58] M. A. Wahid, A. N. Siddiquee, and Z. A. Khan, “Aluminum alloys in marine
construction: characteristics, application, and problems from a fabrication
viewpoint,” Mar. Syst. Ocean Technol., vol. 15, no. 1, pp. 70–80, 2020.
[59] M. A. Wahid, Z. A. Khan, A. N. Siddiquee, R. Shandley, and N. Sharma,
“Analysis of process parameters effects on underwater friction stir welding
of aluminum alloy 6082-T6,” Proc. Inst. Mech. Eng. Pt. B: J. Eng. Manuf.,
vol. 233, no. 6, pp. 1700–1710, 2019.
[60] M. A. Wahid, Z. A. Khan, and A. N. Siddiquee, “Review on underwater
friction stir welding: a variant of friction stir welding with great potential of
improving joint properties,” Trans. Nonferrous Met. Soc. China, vol. 28, no.
2, pp. 193–219, 2018.
[61] M. A. Wahid, S. Masood, Z. A. Khan, A. N. Siddiquee, I. A. Badruddin, and
A. Algahtani, “A simulation-based study on the effect of underwater friction
stir welding process parameters using different evolutionary optimization
algorithms,” Proc Inst Mech Eng Part C, vol. 234, no. 2, pp. 643–657, 2020.
[62] M. A. Wahid, P. Goel, Z. A. Khan, K. M. Agarwal, and E. Hasan Khan,
“Underwater friction stir welding of AA6082-T6: thermal analysis,” in
Advances in Engineering Materials, Singapore: Springer Singapore, 2021,
pp. 365–375.
50 Nature-inspired optimization algorithms and soft computing

[63] M. A. Wahid, S. K. Shihab, R. Shandley, A. Jacob, and T. Majeed, “Friction

stir welding process vis-a-vis human health,” in Design Science and
Innovation, Singapore: Springer Singapore, 2021, pp. 1003–1010.
[64] N. Sharma, Z. A. Khan, A. N. Siddiquee, and M. A. Wahid, “Multi-response
optimization of friction stir welding process parameters for dissimilar joining
of Al6101 to pure copper using standard deviation based TOPSIS method,”
Proc. Inst. Mech. Eng. Part C, vol. 233, no. 18, pp. 6473–6482, 2019.
[65] R. Palanivel, R. F. Laubscher, S. Vigneshwaran, and I. Dinaharan,
“Prediction and optimization of the mechanical properties of dissimilar
friction stir welding of aluminum alloys using design of experiments,” Proc.
Inst. Mech. Eng. Pt. B: J. Eng. Manuf., vol. 232, no. 8, pp. 1384–1394, 2018.
[66] R. L. Malghan, K. M. C. Rao, A. K. Shettigar, S. S. Rao, and R. J. D’Souza,
“Application of particle swarm optimization and response surface metho-
dology for machining parameters optimization of aluminium matrix com-
posites in milling operation,” J. Braz. Soc. Mech. Sci. Eng., vol. 39, no. 9,
pp. 3541–3553, 2017.
[67] A. Majumder, A. Das, and P. K. Das, “A standard deviation based firefly
algorithm for multi-objective optimization of WEDM process during
machining of Indian RAFM steel,” Neural Comput. Appl., vol. 29, no. 3,
pp. 665–677, 2018.
[68] P. C. Pinto, T. A. Runkler, and J. M. Sousa, “Wasp swarm algorithm for
dynamic max-sat problems,” in International Conference on Adaptive and
Natural Computing Algorithms, Springer, 2007, pp. 350–357.
[69] B. Wang, X. Jin, and B. Cheng, “Lion pride optimizer: an optimization
algorithm inspired by lion pride behavior,” Science China Information
Sciences, vol. 55, no. 10, pp. 2369–2389, 2012.
[70] J. C. Bansal, H. Sharma, S. S. Jadon, and M. Clerc, “Spider monkey opti-
mization algorithm for numerical optimization,” Memetic Computing, vol. 6,
no. 1, pp. 31–47, 2014.
[71] S. Mirjalili and A. Lewis, “The whale optimization algorithm,” Adv. Eng.
Softw., vol. 95, pp. 51–67, 2016.
[72] H.-S. Wu and F.-M. Zhang, “Wolf pack algorithm for unconstrained global
optimization,” Math. Probl. Eng., vol. 2014, pp. 1–17, 2014.
[73] J. C. Tillett, R. M. Rao, F. Sahin, and T. M. Rao, “Particle swarm optimi-
zation for the clustering of wireless sensors,” in Proc. SPIE 5100, Digital
Wireless Communications V, 2003.
[74] N. A. A. Latiff, N. M. Abdullatiff, and R. B. Ahmad, “Extending wireless sensor
network lifetime with base station repositioning,” in 2011 IEEE Symposium on
Industrial Electronics and Applications, IEEE, 2011, pp. 241–246.
[75] Y. Xu and Y. Ji, “A clustering algorithm of wireless sensor networks based
on PSO,” in Artificial Intelligence and Computational Intelligence, Berlin,
Heidelberg: Springer Berlin Heidelberg, 2011, pp. 187–194.
[76] S. Thilagavathi and B. G. Geetha, “Energy aware swarm optimization with
intercluster search for wireless sensor network,” Scientific World Journal,
vol. 2015, p. 395256, 2015.
 Nature-inspired optimization algorithm 51

[77] X. Cao, H. Zhang, J. Shi, and G. Cui, “Cluster heads election analysis for
multi-hop wireless sensor networks based on weighted graph and particle
swarm optimization,” in 2008 Fourth International Conference on Natural
Computation, 2008.
[78] S. Jin, M. Zhou, and A. S. Wu, “Sensor network optimization using a genetic
algorithm,” in Proceedings of the 7th World Multiconference on Systemics,
Cybernetics and Informatics, 2003, pp. 109–116.
[79] D. S. Hussain and O. Islam, “Genetic algorithm for energy-efficient trees in
wireless sensor networks,” in Advanced Intelligent Environments, Boston,
MA: Springer US, 2009, pp. 139–173.
[80] A. Norouzi, F. S. Babamir, and A. H. Zaim, “A new clustering protocol for
wireless sensor networks using genetic algorithm approach,” Wirel. Sens.
Netw., vol. 03, no. 11, pp. 362–370, 2011.
[81] W. Luo, “A quantum genetic algorithm based QoS routing protocol for
wireless sensor networks,” in 2010 IEEE International Conference on
Software Engineering and Service Sciences, 2010.
[82] H.-S. Seo, S.-J. Oh, and C.-W. Lee, “Evolutionary genetic algorithm for
efficient clustering of wireless sensor networks,” in 2009 6th IEEE
Consumer Communications and Networking Conference, 2009.
[83] A.-A. Salehpour, B. Mirmobin, A. Afzali-Kusha, and S. Mohammadi, “An
energy efficient routing protocol for cluster-based wireless sensor networks
using ant colony optimization,” in 2008 International Conference on
Innovations in Information Technology, 2008.
[84] S. Mao, C. Zhao, Z. Zhou, and Y. Ye, “An improved fuzzy unequal clus-
tering algorithm for wireless sensor network,” Mob. Netw. Appl., vol. 18, no.
2, pp. 206–214, 2013.
[85] A. M. S. Almshreqi, B. M. Ali, M. F. A. Rasid, A. Ismail, and P. Varahram,
“An improved routing mechanism using bio-inspired for energy balancing in
wireless sensor networks,” in The International Conference on Information
Network 2012, 2012.
[86] D. Karaboga, S. Okdem, and C. Ozturk, “Cluster based wireless sensor
network routing using artificial bee colony algorithm,” Wirel. Netw., vol. 18,
no. 7, pp. 847–860, 2012.
[87] Y. Kumar and G. Sahoo, “A two-step artificial bee colony algorithm for
clustering,” Neural Comput. Appl., vol. 28, no. 3, pp. 537–551, 2017.
[88] P. P, M. Garg, and N. Jain, “An energy efficient routing protocol using ABC
to increase survivability of WSN,” Int. J. Comput. Appl., vol. 143, no. 2,
pp. 37–42, 2016.
[89] S. Potthuri, T. Shankar, and A. Rajesh, “Lifetime improvement in wireless
sensor networks using hybrid differential evolution and simulated annealing
(DESA),” Ain Shams Eng. J., vol. 9, no. 4, pp. 655–663, 2018.
[90] S. Kaur and R. Mahajan, “Hybrid metaheuristic optimization based energy
efficient protocol for wireless sensor networks,” Egypt. Inform. J., vol. 19,
no. 3, pp. 145–150, 2018.
This page intentionally left blank
 Chapter 3
Application aspects of nature-inspired
optimization algorithms
Abhinav Kumar1, Subodh Srivastava1, Vinay Kumar1 and
Niharika Kulshrestha2

In today’s time, there are numerous nature-inspired optimization algorithms

(NIOAs) that help to achieve fast computation with maximum profit and minimum
cost. They are widely used in various fields of engineering, management, and sci-
ence due to their effectiveness in solving complex optimization problems. This
chapter attempts to address the application aspects of NIOAs. It begins with the
definition of NIOAs. In addition to that, it also deals with its implementation and
parameter tunning. Furthermore, it also presents a study on constrained and
unconstrained optimization. Later, it covers the feature selection and practical
engineering applications of NIOAs.

3.1 Introduction

NIOAs are sets of nature-based computational algorithms [1]. Basically, these are a
type of metaheuristic algorithms that are adapted or governed by natural phenom-
ena. It is categorized as bio-inspired (BIOAs), evolution based, and natural science
inspired (physics and chemistry) optimization algorithms (NSIOAs) [2] as shown in
Figure 3.1.
BIOAs are divided into evolution and swarm intelligence-based optimization
algorithm (OA). Genetic algorithm (GA) [3] and differential evolution (DE) [4] fall
under evolution-based OA. Swarm intelligence-based OAs are particle swarm
optimization (PSO) algorithm [5], fire fly (FF) [6], artificial bee colony (ABC) [7],
bacterial foraging (BF) [8], ant colony optimization (ACO) [9], bat algorithm (BA)
[10], cuckoo search (CS) [11], and so on. NSIOAs include simulated annealing
(SA) [12], gravitational search (GS) [13], and big bang big crunch (BBBC) [14]. As
shown in Figure 3.2, NIOAs [15] can be significantly used for engineering, finance

1
Department of Electronics and Communication Engineering, National Institute of Technology – Patna,
India
2
Department of Physics, GLA University, India
54 Nature-inspired optimization algorithms and soft computing

Nature-inspired optimization algorithms (NIOAs)

Bio-inspired optimization algorithms (BIOAs) Natural science-inspired optimization algorithms

(NSIOAs)

Based on evolution Based on swarm behaviors

Simulated annealing

Genetic algorithm Particle swarm Gravitational search

Differential evolution Ant colony Big bang big crunch

Artificial bee colony

Bacterial foraging

Cuckoo search

Firefly algorithm

Figure 3.1 Different categories of nature-inspired optimization algorithms

Environmental

Finance Healthcare

Nature-
inspired Industrial
Engineering optimization designs
algorithms
(NIOAs)

Figure 3.2 Different applications of NIOAs

business activities, environmental management, healthcare, and industrial designs.

NIOAs offer a best possible or optimal solution to an application that maximize the
performance in terms of profit, output, and efficiency. Furthermore, it minimizes
the overall time and energy consumption. It works on set of mathematical models
or algorithms that helps to tune the parameters used in the application.
In engineering section, OAs deals with problems of electrical, electronics and
communication, mechanical, civil engineering, and power systems. However, the
growth of engineering domain is not limited to core section only. The NIOAs are
also a solution provider in other aspects of engineering such as, expert systems,
artificial intelligence (AI) [16], robotics, machine learning (ML) [17], deep learn-
ing (DL) [18], social media, pattern recognition, speech processing, bioinformatics,
biomedical, computer vision, and signal and image processing [19]. NIOAs assist
 Application aspects of nature-inspired optimization algorithms 55

to optimize the issues of industrial designs such as manufacturing, operation

research, and maintenance. In healthcare, NIOAs club with internet of things (IoT)
[20], and big cloud to readdress the current issues and challenges like its scheduling
and management [21]. Furthermore, stock prediction, portfolio, and risk manage-
ment optimization are the main areas of financials activities where NIOAs play an
important role. NOAs have been applied in environmental management for solving
complex optimization problems related to water resources, air quality index, and
logistic and waste management. For instance, GA and PSO algorithms have been
used for water resources management, and ABC algorithm has been used for air
quality management. Thus, optimization is necessary part of our day-to-day life
like from engineering design and development to business setup and from internet
browsing to vacation planning, as real-world applications usually have limited
resources, time, and money. With a variety of constraints, we need to develop ways
to use these precious resources as efficiently as possible.
Before dive into chapter, some common terminologies related to NIOAs [15]
are defined. After that, some of the most commonly used NOAs such as DE [4], GA
[3], PSO [5], ACO [22], ABC [7], CS [11], and [12] SA are discussed.

(A) Common terminologies

(a) Objective function (OF): The function that is being minimized or
maximized. Sometimes it may be known as the cost function, loss
function, or fitness function.
(b) Constraints: The limitations or restrictions that must be satisfied for the
solution to be valid. Constraints can be equality constraints or inequality
constraints.
(c) Decision variables: The elements that can be adjusted to optimize the
objective function. These variables are often denoted as x, y, or z.
(d) Feasible solution: A solution that satisfies all the constraints or
restrictions.
(e) Infeasible solution: A solution that violates one or more constraints.
(f) Global optimum: The best possible solution to the optimization
problem.
(g) Local optimum: A solution that is optimal within a local neighborhood
but may not be the best overall solution.
(h) Gradient: The direction and magnitude of the steepest increase or
decrease of the objective function. It is often used to guide the search
for the optimal solution.
(i) Convergence or stopping criteria: The process of getting closer and
closer to the optimal solution. Convergence criteria are used to deter-
mine when the optimization algorithm has found a satisfactory solution.
(j) Iterations: The number of times the optimization algorithm evaluates
the objective function and adjusts the decision variables to search for
the optimal solution.
(k) Candidate solution: Candidate solution reflects a member, which is the
part of set of all possible feasible solution.
56 Nature-inspired optimization algorithms and soft computing

(B) Some of the most commonly used NOAs

(a) Differential evolution
DE [4] is an evolution-based metaheuristic algorithm that uses the
alterations amid arbitrarily chosen solutions in the population to gen-
erate new solutions. It originates with a population of candidate solu-
tions, called “individuals or entities” [23]. Then, it arbitrarily selects
three entities from the population and produces a new entity by adding a
weighted difference among two of the selected entities to a third
selected entity. The resulting entity is equated to its parent in the
population, and the better of the two is retained for the next generation.
(b) Genetic algorithm
GA [24] is a type of evolution-based NIOAs that simulates the pro-
gression of natural selection. It utilizes the principle of survival of the
fittest to choose the utmost optimal solution from a set of solutions. It
works by generating a population of potential solutions to a problem
and continually applying genetic operators or parameters such as
mutation, crossover, and selection to evolve the population until an
optimal solution is obtained.
(c) Ant colony optimization
ACO [9] algorithms are a type of BIOAs that follows the behavior of ants
in finding the shortest path between their colony and a food source.
Primarily, it uses a pheromone as a biochemical messenger and the con-
centration of pheromone indicates the quality of solutions to the issue. The
greater pheromone concentrations indicate routes and paths that are better
at providing answers to issues like discrete combinatorial problems. This is
because solutions are tied to the pheromone concentration from an imple-
mentation objective. The choice is subtly influenced by some a priori
information about the route, including how frequently the distance is used.
(d) Artificial bee colony
ABC mimic the foraging behavior of honey bees [7]. It is constituted of
three kinds of bees: employed, onlooker, and scout bees. The employed
bees explore for food sources (i.e., potential solutions) in the vicinity of
their current food source, while the onlooker bees observe the waggle
dance of the employed bees to select a promising food source to explore.
The scout bees search for new food sources in areas of the solution space
that have not been explored yet. The algorithm works by iteratively
updating the positions of the employed, onlooker, and scout bees in the
solution space, with the goal of finding the optimal solution.
(e) Particle swarm optimization
PSO [5] follows the principle of swarm intelligence that includes the
natures of bird, flocks, and fish schools. In PSO, a population of parti-
cles is used to explore the solution space of an optimization problem.
Each particle signifies a possible solution, and its position in the solu-
tion space corresponds to the parameters of the solution. The particles
move over the solution space and update their positions and velocities
 Application aspects of nature-inspired optimization algorithms 57

according to their own best-known position and the best-known position

of the swarm. The velocity update is regulated by the particle’s own
experience (i.e., its personal best position) and the experience of the
swarm as a whole (i.e., the global best position). The particle’s position
is then updated based on the new velocity. This process is repeated
iteratively, with the goal of finding the optimal solution.
(f) Cuckoo search
CS [11] follows the brood mechanism of cuckoo species. In CS [25], a
population of cuckoos represents potential solutions to an optimization
problem. Each cuckoo lays an egg, which represents a potential new solu-
tion. The cuckoo then discards its own egg and replaces it with the new egg
if the new solution is better than its own. The algorithm also includes a local
search mechanism to explore the solution space around the current solution.
(g) Fire fly
In FF [6], a population of fireflies represents potential solutions to an
optimization issue. Each firefly is attracted to other fireflies in the
population based on their brightness (i.e., fitness). The firefly moves to
the brighter firefly and updates its position accordingly. The brightness
of each firefly is estimated by the objective function being optimized.
(h) Simulated annealing
SA [12] is based on the characteristics of the metal annealing process. It
originated with an initial solution and iteratively produces new candidate
solutions by randomly perturbing the recent solution. The candidate
solution is then acknowledged or disallowed based on a probability
function that depends on the quality of the candidate solution and a
temperature parameter that decreases over time. The acknowledged
probability function is founded on the Metropolis-Hastings algorithm
[26] from statistical mechanics. At higher temperatures, the algorithm is
more likely to accept solutions that are worse than the current solution,
which allows the algorithm to escape from local optima. As the tem-
perature decreases, the algorithm becomes less likely to accept worse
solutions, which allows it to converge towards the optimal solution.
As mentioned above, there are many applications that can be optimized with
the aid of NIOAs. Since the role of image processing is very vast in the areas such
as diagnosis of disease through medical images [27], forgery detection [28], crowd
management [29], and so on. So, this chapter presents an overview of optimization
algorithms in image processing.

3.2 Application domains of nature-inspired optimization

algorithms
Image is a collection of pixel values [30]. Image processing is a process of image
manipulation where an image is transformed into digital form by a suitable means
to extract the required information from it. As shown in Figure 3.3, pre-processing,
58 Nature-inspired optimization algorithms and soft computing

Image denoising

Image
Pre-processing
enhancement

Digital image processing

Image
Segmentation
restoration
Feature
extraction

Feature selection

Classification

Figure 3.3 Basic building of digital image processing

segmentation, feature extraction, feature selection, and classification are the basic
building blocks of image processing [31]. Pre-processing refers to the task of image
resizing, normalization, cropping, noise removal, image quality improvement, and
image restoration [32]. Pre-processing steps enhance the features of image.
Segmentation helps to find the region of interests (ROIs) by dividing the images
into a group of segments [33]. ROIs indicate the abnormal region in the case of
medical images. Feature extraction is the procedure of mining the features of an
image that assists in differentiation between two objects [34]. It is derived from the
internal attributes and the characteristics of the image. Feature selection is a sta-
tistical approach which finds the redundant and non-redundant features related to
the image [35]. It is usually used for dimension deduction. Classification [36]
describes the class or label of the image. It is estimated on the basis of extracted and
selected features. For example, in the case of image cancer classification, the
classification process categorized the images as cancerous or non-cancerous.
Nowadays, the steps involved in image processing are clubbed with ML and
DL to accelerate the accuracy and computation rate. Yet, it cannot be done without
the aid of an optimization technique. OAs are employed in various image proces-
sing application like to increase image quality, lessen noise, improve segmentation,
to optimize the feature set and classifications parameters. For instance, DE [37],
GA [38], PSO [39], ACO [40,41], and ABC [42] algorithms have been used for
image enhancement, segmentation [37,43], feature selection, classification [40],
and clustering, respectively.

3.2.1 Optimization in image denoising

Image denoising is a means to eliminate the noise from an image [31]. Noise can be
caused at the time of image acquisition or transmission by various factors such as
sensor noise, compression artifacts, or environmental factors. The general mathe-
matical model of an image [32] can be read as:
y ¼ mh1 :x þ ah2 (3.1)
 Application aspects of nature-inspired optimization algorithms 59

Here, y is the observed image; x is the noiseless image; ah2 is additive noise,
which due to machine calibration or sensors; and mh1 is inherent characteristic of
image known as multiplicative noise.
Based on the evolution of imaging modalities [27] like mammography (X-ray)
[32], ultrasound [44], magnetic resonance imaging (MRI) [45], computed tomo-
graphy (CT) [42], positron emission tomography (PET), and micro biopsy [33], the
noises can be categorized as Quantum (Poisson) [32], Speckle (Rayleigh) [44],
Rician [46], and Gaussian, respectively. The most commonly used denoising filters
[31] are median [47], wiener [48], wavelet [49], bilateral [50], non-local means
[51], total variation (TV) [52], anisotropic diffusion [53], fourth-order partial dif-
ferential equation [54], and complex diffusion [55]. All these filters are constituted
with a number of image parameter. The parameters are selected manually, which
increases the time complexity and affects the denoising filters accuracy. So, there is
a need of optimization that can improve the efficiency of denoising filters by
selecting the optimal filter parameters and denoising rules, and also minimize the
error between the noisy and denoised image.

3.2.2 Optimization in image enhancement

Image enhancement [31] is a procedure of refining the quality of an image by
increasing its contrast, sharpness, brightness, color balance, and other visual fea-
tures. Histogram equalization [56], unsharp masking [55], high boost filtering [57],
contrast limited adaptive histogram [31], equalization, contrast stretching [30], and
so on are primarily developed image enhancement techniques. The purpose of any
enhancement techniques is to redefine or restore the pixel values. Since, the
enhancement techniques comprise manual tuned parameters. Thus, optimization is
essential in order to advance the efficacy of enhancement techniques by choosing
the optimal parameters values. The NIOAs like GA, PSO, ABC, and ACO may be
used for contrast, brightness, color image enhancement, and image sharpening,
respectively. Upendra et al. [56] have proposed an image enhancement technique
that combined GA with histogram equalization to enhance the contrast of images.
The idea behind of ACO for image contrast enhancement is as follows:
(i) It searches for the best values of the contrast parameters by mimicking the
foraging behavior of ants.
(ii) The algorithm’s pheromone trail is a representation of the contrast level, and
the ants’ behavior is based on the intensity of the pheromone trail.
(iii) By applying this method, the algorithm can enhance the image’s contrast by
increasing the brightness level of the pixels in the image.

3.2.3 Optimization in image segmentation

Image segmentation [31] is a method of partitioning an image into several segments
or regions that is governed by certain characteristics, such as color, texture, or
shape. It is crucial in image processing as it enables the identification of ROIs in an
image. The image segmentation techniques [30] are categorized as thresholding,
edge detection, region growing, clustering, watershed, and graph-based segmentation
60 Nature-inspired optimization algorithms and soft computing

techniques. These techniques can be castoff individually or in combination with each

other to attain the expected segmentation results. The selection of technique depends
on the specific requirements of the application and the characteristics of the image
being segmented. But due to complex image structures, segmentation fails to seg-
ment desired ROIs. Therefore, OAs can play a vital role.
The optimization problem can be expressed as an objective function that
measures the quality of the segmentation and NIOAs can be applied to optimize
this function. In simple words, the fitness function is designed to measure the
similarity between the segmented image and the ground truth. The NIOAs advan-
ces the accuracy and proficiency of image segmentation by choosing the optimal
threshold values, clustering parameters, and region growing rules. Bahadur et al.
[45] have demonstrated MRI brain tumor image segmentation and classification
using GA.

3.2.4 Optimization in image feature extraction and selection

Image feature extraction and selection have played a significant role in many com-
puter vision applications [58], such as object detection, recognition, and tracking [31].
The goal of feature extraction is to draw out relevant and discriminative features from
the image, while feature selection aims to choose the most informative features for
the task at hand. The features can be extracted by using image properties loke texture,
shape, and size. Some common techniques for image feature extraction and selection
[59] are scale-invariant feature transform (SIFT) [60], local binary patterns [34],
histogram of oriented gradients [61], speeded-up robust features [62], principal
component analysis [63], recursive feature elimination [64], linear discriminant ana-
lysis [65], and mutual information (MI) [66]. The choice of feature extraction and
selection techniques depends on the specific task and the characteristics of the image
data. It is often necessary to experiment with different techniques and combinations
of techniques to find the best approach for a particular application. NIOAs plays a
crucial role in image feature extraction and selection because it can significantly
improve the performance, accuracy, efficiency, and scalability of computer vision
[58] algorithms making them more suitable for real-world applications. NIOAs can
be used to find the best subset of features that maximizes the objective function with
the aid of parameter tuning, dimensionality reduction. For example, the SIFT algo-
rithm [60] has parameters that control the size of the key points, the number of
orientations, and the threshold for eliminating low-contrast key points. OAs such as
grid search, random search, or Bayesian optimization can be used to find the optimal
values for these parameters [67].

3.2.5 Optimization in image classification

Image classification [31] is the procedure of assigning a label or a class to an image
based on its features. Its goal is to accurately identify and classify images into
predefined categories, such as objects, scenes, or activities. Furthermore, the
extended form of classification helps in localization and detection of abnormalities
from medical images, object recognition from day-to-day activity, and face
 Application aspects of nature-inspired optimization algorithms 61

detection [38] from crowd management. The traditional approach of image classi-
fication is to draw out the features from the image and then use a ML algorithm to
learn a mapping between the features and the labels. However, this approach
requires domain expertise in feature engineering and may not generalize well to
new data. Support vector machine (SVM) [35], K-nearest neighbor [68], decision
tree, random forest, convolutional neural network (CNN) [69], mask regional
convolutional neural network [70], and so on are classification techniques.
Classification techniques have some limitations in handling complex and non-
linear relationships between the features and the labels. So, the OAs like GA, PSO,
CS, and ACO can be used to optimize the classification parameters where the
fitness function can be defined as the classification accuracy of a classifier trained
on the extracted features.

3.3 Implementation
NIOAs have several parameters that need to be set correctly to ensure good per-
formance. These parameters [71] include the population size, crossover with
mutation rates (for GA), inertia weight with learning factors (for PSO), and pher-
omone evaporation and deposition rates (for ACO). Implementation of NIOAs
involves parameter tuning, representation of the solution, fitness function, and
stopping criteria [72]. Parameter tuning helps to find the best values for these
above-mentioned parameters. It relies on the specific issues being solved, and
different values can lead to different solutions. The solution representation is
another crucial aspect of implementing these NIOAs. The solution needs to be
encoded in a way that is suitable for the specific algorithm being used. For exam-
ple, in GA, solutions are typically exhibited as binary strings, while in PSO, solu-
tions are represented as vectors. The solution representation can greatly affect the
performance of the algorithm, and different representations can lead to different
solutions. The fitness function is the objective function that the OAs are trying to
optimize. It evaluates how good a particular solution is. It needs to be expressed in
a way that exactly captures the problem being solved. In some cases, it can be
difficult to design, and it may require domain-specific knowledge. Furthermore,
stopping criteria determine when the OAs should stop searching for a solution. It
may be the number of assigned iterations, the maximum time allowed, or when a
specific solution is found. Choosing the right stopping criteria is crucial to prevent
overfitting or underfitting the problem being solved.
Figure 3.4 demonstrates the generalized implementation steps to perform an
optimization in image denoising. It starts from input noisy image followed by the
definition objective function; choosing of an OAs; initialization of the OAs; itera-
tion of the OAs; determination of stopping criteria; and it ends with output denoised
image. Table 3.1 illustrates the implementation steps of PSO-based TV denoising
filter. The optimization begins with defining an objective function. The OF is
defined in terms of noisy and denoised image. After that, the PSO particles are
initialized and its fitness is evaluated. Based on the fitness evaluation, PSO
62 Nature-inspired optimization algorithms and soft computing

Choose an
Define the objective
Input noisy image optimization
function
algorithm

Initialize the Iterate the Determine stopping

algorithm algorithm criteria

Output denoised
image

Figure 3.4 Generalized implementation steps of optimization algorithms for

image denoising

Table 3.1 Generalized implementation steps of PSO in TV denoising filter

Step I: Define the objective function

The goal is to find an image that is close to the noisy image while also being smooth. The
objective function in PSO consists two terms: (I) a data fidelity term that measures the
difference among the noisy image and the denoised image and (II) a regularization term that
promotes smoothness.
Step II: Initialize the particles
The population of particles in PSO represents potential solutions to the optimization
problem. Each particle has a position and a velocity vector that are initialized randomly.
Step III: Evaluate the fitness
The fitness of each particle is estimated by using its position vector, which corresponds to a
denoised image. The fitness is computed using the objective function.
Step IV: Update the particles
Each particle’s velocity and position vectors which are updated based on its own best
position so far (i.e., the position with the best fitness that this particle has reached) and the
global best position so far (i.e., the position with the best fitness among all particles in the
population). The velocity update is done by the particle’s previous velocity, the difference
among its recent position and its best position, and the difference between the global best
position and its recent position.
Step V: Evaluate the fitness of the updated particles
After the particles’ positions and velocities have been updated, their fitness is re-evaluated
based on the updated position vectors.
Step VI: Repeat the update and evaluation steps
Steps 4 and 5 are repeated until a stopping criterion is met. This criterion can be based on a
maximum number of iterations, a minimum change in the objective function or the fitness of
the particles, or some other criterion.
Step VII: Output the denoised image
Once the PSO algorithm has converged, the best particle (i.e., the one with the best fitness)
represents the denoised image, and its position vector can be outputted as the final estimate.
 Application aspects of nature-inspired optimization algorithms 63

particles are updated. These steps are repeated until the matching of stopping cri-
teria. At the end, the optimized denoised image is obtained as output. PSO particles
are categorized as position and velocity vector as stated in the definition.
The implementation of the NIOAs needs to be carefully designed and opti-
mized to ensure a good performance. Furthermore, the algorithm needs to be
designed to take advantage of parallel processing when available, as well as other
optimizations such as local search and hybridization with other algorithms. It also
needs to be scalable and adaptable to different problem sizes and types.

3.3.1 Parameter tuning

NIOAs are broadly used to solve complex optimization issues. The performance of
NIOAs is highly reliant on the values of their parameters [71]. Because, parameters
control the behavior of the search process, such as the rate of mutation in GA or the
velocity update in PSO. The choice of appropriate parameter values can sig-
nificantly affect the convergence rate and the quality of the solutions obtained by
the algorithm. Therefore, parameter tuning is a critical step in the application of
these algorithms. The parameters include population size, mutation, crossover rate,
selection strategy, and convergence criteria.
(a) Population Size (PS): It determines the quantity of candidate solutions that
are evaluated in every iteration. A larger PS can improve the accuracy of the
search but also increases the computational cost. While a smaller PS can
reduce the computational cost but may lead to premature convergence. The
PS should be selected according to the problem’s complexity, the available
computational resources, and the algorithm’s performance.
(b) Mutation Rate (MR): MR controls the degree of randomness in the search
process. A higher MR increases the diversity of the population that leads the
algorithm escape local optima. However, a high mutation rate can also reduce
the convergence rate and lead to a longer search time. The mutation rate should
be selected based on the problem’s landscape and the algorithm’s performance.
(c) Crossover Rate: Crossover rate is a parameter that determines the frequency
of recombination between candidate solutions. A higher crossover rate
explores the search space more efficiently, but it may also increase the
probability of premature convergence. A lower crossover rate may improve
convergence but may also decrease the diversity of the population. The
crossover rate should be selected based on the problem’s landscape and the
algorithm’s performance.
(d) Selection Strategy: Selection strategy is a parameter that determines how can-
didate solutions are selected for the next iteration. The most common selection
strategies are roulette wheel and tournament selection strategies. Roulette wheel
selection selects solutions based on their fitness, whereas tournament selection
selects solutions based on their ranking. The selection strategy should be selected
based on the problem’s characteristics and the algorithm’s performance.
(e) Convergence Criteria: Convergence criteria is a parameter that determines
when the algorithm should stop. The most common convergence criteria are
64 Nature-inspired optimization algorithms and soft computing

the maximum number of iterations and the target fitness value. The con-
vergence criteria should be selected based on the problem’s complexity and
the algorithm’s performance.
The parameters of NIAOs are of two kinds: global and local parameters. Global
parameters affect the behavior of the entire algorithm, such as the PS in GA or the
number of iterations in ACO. Furthermore, local parameters affect the behavior of
individual components of the algorithm, such as the crossover probability in GA or
the pheromone evaporation rate in ACO. There are various methods for parameter
tuning [67] in NIOAs, including manual tuning, grid search, random search, and
metaheuristic optimization. Each of these methods has its advantages and dis-
advantages, and the selection of the technique depends on the specific problem at
hand. A well-tuned NIOAs can significantly improvise the proficiency and exact-
ness of the optimization process.

3.3.2 Manual tuning

Manual tuning is the most straightforward and commonly used technique for para-
meter tuning. In this technique, the parameters are selected by the domain knowledge
and experience of the user. The user adjusts the parameters of the algorithm itera-
tively until the desired performance is achieved. It is a time-consuming process and
may require multiple trials and errors to obtain the optimal parameter values.

3.3.3 Grid search

It is a simple and organized approach for parameter tuning. In this technique, a
range of parameter values is defined for each parameter, and the algorithm is run
for all combinations of parameter values. The performance is evaluated for each
combination, and the optimal parameter values are selected based on the results. It
is a computationally expensive technique and may not be feasible for problems
with a large number of parameters or a high computational cost.

3.3.4 Random search

It is a stochastic approach for parameter tuning. In this technique, the parameters are
selected randomly within predefined ranges. The algorithm is run for each combi-
nation of parameter values, and the performance is evaluated. The optimal parameter
values are selected based on the results. It is less computationally expensive than grid
search but may require more iterations to obtain the optimal parameter values.

3.3.5 Metaheuristic optimization

It is a powerful technique for parameter tuning. In this technique, a metaheuristic
algorithm searches the optimal parameter values. The metaheuristic algorithm can
be any NIOAs, such as GA, PSO, ACO, and CS. The metaheuristic algorithm
optimizes a fitness function that evaluates the performance of the original algo-
rithm for each parameter combination. It can be computationally expensive but can
provide better results than manual tuning, grid search, or random search.
 Application aspects of nature-inspired optimization algorithms 65

Table 3.2 Time complexity relationship with the parameters

Parameters tuning Time complexity

methods
Manual tuning It may be linear to exponential, depending on the number of
hyperparameters and the number of values each hyperparameter can
take.
Grid search The time complexity is Oðnm Þ; where n is the number of
hyperparameters and m is the number of values each hyperparameter
can take.
Random search The time complexity is Oðk Þ; where k is the number of samples
taken. It depends on the number of samples taken and the size of the
search space.
Metaheuristic The time complexity can range from Oðn log nÞ to Oðn2 Þ, where n is
optimization the number of hyperparameters.

Table 3.2 shows the time complexity chart of manual tuning, grid search,
random search, and metaheuristic optimization-based parameters tuning.

3.4 Constrained and unconstrained optimization

Optimization problems can be categorized into two types: constrained [73] and
unconstrained optimization [74]. Constraints are set of circumstances that must be
satisfied in order to achieve a valid solution. In both cases, NIOAS have become
popular due to their effectively solving procedure. Figure 3.5 exhibits the cate-
gorization of NIOAs problems according to constrained- and unconstrained-based
optimization.

3.4.1 Constrained optimization

It refers to the problem of finding the optimal solution subject to one or more
constraints. The constraints may be in the form of equalities or inequalities. The
main challenge in constrained optimization is to find a feasible solution that
satisfies all the constraints and is also optimal [75].
NIOAs are well-suited for constrained optimization problems because they can
efficiently search the solution space and find feasible solutions that satisfy the
constraints. The constraint-based algorithms [73] can be classified into two groups:
penalty function and constraint handling methods.

3.4.1.1 Penalty function methods

It adds a penalty term to the OF to penalize violations of the constraints. The
penalty term is designed to be large when the constraints are violated, and it
decreases as the constraints are satisfied. The algorithm then searches for the
solution that minimizes the penalized objective function.
66 Nature-inspired optimization algorithms and soft computing

NIOAs
problems

Constrained Unconstrained
optimization optimization

Constraint
Penalty function handling Population- Single-point
methods methods based methods methods

Figure 3.5 Categorization NIOAs on the basis of constrained and unconstrained

optimization problems

3.4.1.2 Constraint handling methods

It uses specific techniques to handle the constraints directly. These techniques may
include repairing infeasible solutions, transforming the constraints into the objective
function, or incorporating the constraints into the search process. NIAOs that are com-
monly used for constrained optimization problems [76] include GA, PSO, DE, and ACO.

3.4.2 Unconstrained optimization

It refers to the problem of finding the optimal solution without any constraints. The
main challenge in unconstrained optimization is to find the global optimum, which
is the point that minimizes the objective function.
The unconstrained-based OAs are population-based methods and single-point
methods.

3.4.2.1 Population-based methods

It maintains a population of candidate solutions and search for the global optimum
by evolving the population through selection, crossover, and mutation operations.
Examples of population-based methods include GA, PSO, and ACO.
3.4.2.2 Single-point methods
It searches for the global optimum by iteratively updating a single candidate
solution. Examples of single-point methods include gradient descent, Newton’s
method, and SA.

3.5 How to deal with constraints

Constraints are restrictions on the decision variables that must be satisfied in an
optimization problem [75]. For example, in a manufacturing process, there may be
 Application aspects of nature-inspired optimization algorithms 67

constraints on the production capacity, material availability, and energy con-

sumption. In financial planning, there may be constraints on the investment port-
folio, interest rates, and inflation. In biology, there may be constraints on the
nutrient availability, population growth, and predation. NIOAs such as GA [77],
PSO [78], and ACO [79] are based on imitating the behavior of natural systems to
find optimal solutions to optimization problems. However, these algorithms may
not be suitable for problems with constraints, as they may violate the constraints
while searching for solutions. Therefore, special techniques are needed to deal with
constrained optimization problems [73]. There are several approaches to handling
constraints in nature-inspired optimization algorithms.

3.5.1 Penalty functions

The penalty function [80] adds a penalty term to the OF whenever a constraint is
violated, thereby discouraging the optimization algorithm from producing solutions
that violate the constraints. There are several types of penalty functions that can be
used to deal the constraints.
(a) L1 Penalty (Lasso): The L1 penalty function adds the absolute values of the
coefficients to the OF [81]. It is used to encourage sparse solutions, where
many of the coefficients are zero.
(b) L2 Penalty (Ridge): The L2 penalty function adds the squared values of the
coefficients to the OF [81]. It is used to prevent overfitting by reducing the
magnitude of the coefficients.
(c) Elastic Net Penalty: It is a combination of the L1 and L2 penalties [82]. It is
used to achieve both sparsity and regularization.
(d) Huber Penalty: The Huber penalty function is a combination of the L1 and L2
penalties, which is less sensitive to outliers in the data [83].
(e) Penalty Barrier Method: It is a type of penalty function that adds a barrier
term to the OF [73]. It is used to constrain the variables to a feasible region.
(f) Augmented Lagrangian Method: The augmented Lagrangian method is a
penalty function that adds a Lagrangian multiplier term to the OF [76]. It is
used to impose constraints on the optimization problem.
(g) Trust-Region Penalty Method: This penalty method is a penalty function that
adds a penalty term to the OF [84]. It is used to impose constraints on the
optimization problem by defining a trust region around the current iterate.
These penalty functions may be used in several optimization issues that includes
linear [85] and nonlinear programming [86], and may offers a well optimized
answer. The penalty function can be designed in different ways, such as a linear or
quadratic penalty function, or a more complex function that depends on the degree
of constraint violation.

3.5.2 Linear penalty function

A linear penalty function is a function that increases linearly with the violation of
the constraints [85]. For example, if the optimization problem has a constraint that
68 Nature-inspired optimization algorithms and soft computing

a variable x must be greater than or equal to some value b, then a linear penalty
function might add a term of the form. It may be mathematically noted as:
PðxÞ ¼ maxð0; b  xÞ (3.2)
This penalty function is zero when x is greater than or equal to b, and is equal
to ðb  xÞ when x is less than b. The linear penalty function increases proportion-
ally with the amount of constraint violation.

3.5.3 Quadratic penalty function

A quadratic penalty function is a function that increases quadratically with the
violation of the constraints [87]. For example, a quadratic penalty function might
add a term as shown in (3.3):
 
PðxÞ ¼ max 0; ðb  xÞ2 (3.3)

This penalty function is zero when x is greater than or equal to b and is equal to
ðb  xÞ2 when x is less than b. The quadratic penalty function increases much faster
than the linear penalty function with the amount of constraint violation.

3.5.4 Constraint handling techniques

Constraint handling techniques [79] include methods such as repair, feasibility-
based selection, and constraint domination. Actually, repair, feasibility-based
selection, and constraint domination are not considered as constraint handling
techniques in optimization, but they are related concepts that are used in constraint
satisfaction problems (CSPs) [88] and constraint programming (CP) [89]. In CSPs,
repair methods are used to modify the current solution by repairing the violated
constraints.
(i) Repair methods use heuristics to modify the current solution and make it fea-
sible, without requiring a complete search of the solution space. Repair methods
can be used iteratively until a feasible solution is found, or until a maximum
number of iterations is reached. Feasibility-based selection is a technique used in
CSPs [90] to select variables for assignment based on their feasibility.
(ii) Feasibility-based selection selects variables that are more likely to lead to a
feasible solution, based on the constraints that have already been assigned.
Constraint domination is a technique used in CP to simplify the constraint
graph by removing redundant constraints.
(iii) Constraint domination identifies constraints that are dominated by other
constraints and removes them from the constraint graph, reducing the size of
the search space and improving the efficiency of the algorithm.
While these techniques are not considered as constraint handling techniques in
optimization, they are still important in solving optimization problems with con-
straints. In some cases, they can be combined with constraint handling techniques
to improve the overall performance of the optimization algorithm. Some of the
 Application aspects of nature-inspired optimization algorithms 69

constraint handling [73] techniques along with penalty function method are listed
below:
(a) Lagrange Multiplier Methods: Lagrange multiplier methods add a Lagrange
multiplier term to the OF that enforces the constraints as equality constraints
[76]. The Lagrange multiplier is a scalar parameter that is adjusted during
optimization to ensure that the constraints are satisfied.
(b) Sequential Quadratic Programming (SQP): SQP methods solve a sequence
of sub-problems that approximate the original problem while satisfying the
constraints [91]. The sub-problems are solved using quadratic programming,
which can handle nonlinear constraints.
(c) Interior Point Methods: Interior point methods solve the optimization pro-
blem by moving towards the interior of the feasible region [92]. The constraints
are handled by adding barrier functions to the OF, which ensures that the
optimization variables remain within the feasible region.
(d) Active Set Methods: Active set methods involve solving a sequence of sub-
problems in which some of the constraints are active (i.e., they are satisfied
with equality) and others are inactive (i.e., they are satisfied with inequality)
[93]. The active set is updated iteratively until a feasible solution is found.

3.5.5 Hybrid constraint handling techniques

Hybrid constraint handling techniques (HCHTs) [94] are a combination of different
constraint handling techniques used in NIOAs to improve the optimization per-
formance. For example, a GA can be combined with a local search method that
repairs solutions that violate constraints. Some examples of HCHTs [95] used in
nature-inspired optimization algorithms are:
(a) Penalty-Based Hybrid Constraint Handling: This approach combines a penalty
function with other techniques, such as active set methods, to handle constraints.
Penalty function, penalizes constraint violations, while the active set method is
used to update the constraints and determine which constraints are active.
(b) Barrier-Based Hybrid Constraint Handling: This approach combines a
barrier method with other techniques, such as augmented Lagrangian methods
[76], to handle constraints. The barrier method is used to prevent the optimi-
zation solution from violating the constraints, while the augmented Lagrangian
method is used to handle nonlinear constraints.
(c) Hybrid Genetic Algorithm: This method combines GA with other techniques,
such as penalty functions, to handle constraints [96]. The genetic algorithm is
used to generate a set of feasible solutions, while the penalty function is used to
penalize constraint violations and improve the quality of the solutions.
(d) Hybrid Particle Swarm Optimization: This technique combines PSO with other
methods, such as active set methods, to handle constraints [97]. The particle swarm
optimization algorithm is used to search for the optimal solution, while the active
set method is used to update the constraints and ensure feasibility.
(e) Hybrid Ant Colony Optimization: This approach combines ACO with other
techniques, such as penalty functions, to handle constraints [98]. The ant
70 Nature-inspired optimization algorithms and soft computing

colony optimization algorithm is used to generate a set of feasible solutions,

while the penalty function is used to penalize constraint violations and improve
the quality of the solutions.
HCHTs can be effective in improving the optimization performance of nature-
inspired optimization algorithms, and the selection of which HCHT to use depends
on the explicit problem and the desired level of constraint satisfaction.

3.6 Feature selection

Feature selection is a vital step in many ML [59] and DL [99] tasks, as it can
significantly advance the accuracy of the resulting models by deducting the number
of features that are irrelevant or redundant. Feature selection algorithms aim to
identify the subset of features that are most informative for the given task [31]. The
traditional methods for feature selection like the sequential forward selection and
backward elimination, have limitations in handling high-dimensional data. NIOAs
have been successfully implemented to feature selection problems, as they can
effectively explore the search space and identify relevant feature subsets [100].

3.6.1 Feature selection based on GA

Feature set is obtained after the process of feature extraction. Since, these features
set contains all redundant and non-redundant information. So, such features which
are utmost required for classification or detection purposes that are designated with
the aid of feature selection. But all these processes are complex in nature, so to
deduce complexity and increase accuracy, it must be optimized. GA uses the
genetic operators to optimize the feature set [101]. The steps involved to optimize
the feature set with GA is demonstrated in Table 3.3.

Table 3.3 GA-based generalized feature selection step

Initialization: Initialize a population of individuals, where each individual signifies a

potential solution (i.e., a subset of features).
Fitness evaluation: Evaluate the fitness of each individual by using the quality of the
selected features.
Selection: Select the fittest individuals to form a mating pool for the next iteration.
Crossover: Perform crossover between pairs of parents to create offspring, where the
offspring inherit some of the features from each parent.
Mutation: Mutate some of the offspring to introduce new features or remove existing
features.
Evaluation: Evaluate the fitness of the offspring.
Replacement: Replace the least fit individuals in the population with the offspring.
Termination: The algorithm stops when a stopping criterion is met, such as a maximum
number of iterations or a convergence criterion.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.
 Application aspects of nature-inspired optimization algorithms 71

Table 3.4 Overview of feature selection steps with the aid of PSO

Initialization: Initialize a swarm of particles, where each particle signifies a potential

solution (i.e., a subset of features).
Velocity and position update: Each particle adjusts its velocity and position based on its
recent position, its best position so far, and the best position found by the swarm. This update
is based on a set of equations that balance exploration and exploitation.
Fitness evaluation: Evaluate the fitness of each particle by using the quality of the selected
features.
Local search: Each particle performs a local search to improve its solution by adding or
removing features.
Global update: After each iteration, the best solution found so far is communicated to all
particles to update their search behavior.
Termination: The algorithm stops when a stopping criterion is matched like total number of
iterations or a convergence norm.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.

3.6.2 Feature selection based on PSO

PSO can be used to optimize the selection of features based on different criteria,
such as mutual information, correlation, or classification error. For example, PSO
has been used to optimize the selection of features for classification of medical
images [102]. The primarily steps to readdress the issues of feature selection by
PSO is presented in Table 3.4.

3.6.3 Feature selection based on ACO

In the context of feature selection by ACO, each feature is represented as a node in
a graph, and the pheromone trail represents the importance of each feature [103].
Table 3.5 demonstrates the generalized steps used by ACO to perform feature
selection.

3.6.4 Feature selection based on ABC

ABC can be used to perform feature selection by optimizing a fitness function that
evaluates the quality of the selected features [104]. ABC mimics the behavior of
ants. Table 3.6 exhibits the basic steps for feature selection using ABC.

3.6.5 Feature selection based on CS

CS has a widespread range of applications in feature selection [105]. Many MLs
and DLs [20] follow the CS for the feature selection. The generalized feature
selection steps based on CS are shown in Table 3.7.
72 Nature-inspired optimization algorithms and soft computing

Table 3.5 Steps involved in feature selection by ACO

Initialization: Initialize the population of ants, where each ant signifies a potential solution
(i.e., a subset of features).
Pheromone trails: Each ant constructs a pheromone trail that exhibits the quality of its
solution, where the pheromone level increases with the fitness value of the solution.
Probabilistic transition: Each ant probabilistically selects the next feature to include in its
solution based on the pheromone level and the importance of the feature.
Local search: Each ant performs a local search to improve its solution by adding or
removing features.
Global update: After each iteration, the pheromone levels are updated based on the best
solution found so far.
Termination: The algorithm stops when a stopping criterion is satisfied in terms of number
of iterations.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.

Table 3.6 Basic steps for feature selection using ABC

Initialization: Initialize the population of bees, where each bee characterizes a potential
solution (i.e., a subset of features).
Employed bees: Each employed bee evaluates a fitness function for its solution and then
searches for a neighboring solution to improve its fitness. The neighboring solution is
generated by randomly selecting a feature to add or remove from the current solution.
Onlooker bees: It select a solution based on its fitness value and then search for a
neighboring solution to improve its fitness. The selection probability is relative to the fitness
value of the solution.
Scout bees: If an Employed or onlooker bee has not found a better solution after a certain
number of iterations, it becomes a scout bee and generates a new solution randomly.
Termination: The algorithm stops when a stopping criterion matched.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.

Table 3.7 Steps performed for feature selection by CS

Initialization: Initialize the population of cuckoos, where each cuckoo signifies a potential
solution (i.e., a subset of features).
Levy Flight: Each cuckoo generates a new solution by performing a random walk in the
feature space, which is modelled by a Levy Flight distribution.
Evaluate fitness: Each cuckoo evaluates a fitness function for its solution and compares it to
the fitness values of the other cuckoos in the population.
Replace eggs: If a cuckoo’s solution is better than another cuckoo’s solution, it replaces the
other cuckoo’s solution with a new solution generated by a Levy Flight.
Abandon eggs: Some cuckoos may lay eggs in the nests of other cuckoos. If a cuckoo finds
a better solution by laying an egg in another cuckoo’s nest, it abandons its own nest and
adopts the other cuckoo’s nest.
Termination: The algorithm iterates until stopping condition is matched.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.
 Application aspects of nature-inspired optimization algorithms 73

Table 3.8 FF-based feature selection steps

Initialization: Initialize the population of fireflies, where each firefly represents a potential
solution (i.e., a subset of features).
Attraction: Each firefly moves towards other fireflies that have a higher fitness value,
which is modelled by an attractiveness function that depends on the distance between the
fireflies and their fitness values.
Randomization: Each firefly also moves randomly to explore the search space and escape
from local optima.
Intensity: The intensity of each firefly’s light is relative to its fitness value, and the intensity
decreases with distance from the firefly.
Absorption: If a firefly’s light intensity is greater than another firefly’s light intensity, it
absorbs the other firefly’s position and updates its solution.
Termination: It will terminate after the matching of convergence norm.
Once the optimization process is completed, the subset of features with the highest fitness
value is selected as the optimal subset.

3.6.6 Feature selection based on FF

FF exhibits the properties of natural fire flies. FF can be used for dimension
reduction [106]. Table 3.8 shows the steps involved in feature selection by FF.
Additionally, GA, PSO ACO, ABC, CS, and FF may require a large number of
iterations to converge to an optimal solution, which can increase the computational
cost of the algorithm.

3.7 Practical engineering applications

The applications of NIOAs are numerous and varied, ranging from power dis-
tribution networks to building energy systems. As such, they are an essential tool in
the toolbox of modern engineers. Some of the practical engineering applications of
NIOAs [71,107] are listed below. Furthermore, Table 3.9 demonstrates an overview
on the practical engineering applications with the aid of NIOAs.
(a) ACO: It is enthused by the foraging behavior of ants. This algorithm has been
broadly used in engineering applications such as optimization of tele-
communication networks, vehicle routing, and job scheduling. One practical
application of ACO is in the optimization of power distribution networks. ACO
can be used to find the shortest path between power stations, reducing energy
losses, and improving the efficiency of the network.
(b) PSO: It is governed by the flocking behavior of birds and the schooling beha-
vior of fish. It has been used in a several engineering applications, including
antenna design, robot control, and image processing. The real-time example of
PSO is in the optimization of wind turbines. PSO can be used to optimize the
blade shape and angle of attack, improving the efficiency and power output of
the wind turbine.
74 Nature-inspired optimization algorithms and soft computing

Table 3.9 Overview on the practical engineering applications with the aid of
NIOAs

NIOAs Practical engineering applications

DE ● Parameter tuning of electric motors
● Design of heat exchangers
● Optimization of gearboxes
● Process control of wastewater treatment plants
GA ● Structural design of aircraft components
● Optimization of power generation systems
● Design of control systems for autonomous vehicles
● Optimization of manufacturing processes
PSO ● Optimization of water distribution networks
● Design of antenna arrays
● Scheduling of production systems
● Optimization of energy consumption in buildings
ACO ● Routing of vehicles in transportation networks
● Optimization of wireless sensor networks
● Design of electrical power networks
● Optimization of logistics and supply chain management
ABC ● Optimization of wind farm layout
● Design of power systems
● Optimization of wireless communication systems
● Parameter estimation in chemical processes
FF ● Optimization of structural design
● Optimization of machine learning algorithms
● Scheduling of transportation systems
● Optimization of neural networks
CS ● Design of power systems
● Optimization of artificial neural networks
● Image processing and analysis
● Optimization of medical diagnosis models
BF ● Signal and Image processing
● Sensor networks
● Control systems
● Electrical power systems
● Chemical and mechanical engineering
● Bioinformatics
SA ● Optimization of supply chain management systems
● Design of artificial neural networks
(Continues)
 Application aspects of nature-inspired optimization algorithms 75

Table 3.9 (Continued)

NIOAs Practical engineering applications

● Optimization of renewable energy systems

● Scheduling of production systems
GS ● Optimization of robot path planning
● Optimization of hydrodynamic flow control systems
● Design of communication networks
● Optimization of machine learning algorithms
BBBC ● Optimization of water resource management systems
● Design of flexible manufacturing systems
● Optimization of supply chain networks
● Parameter estimation in hydrological models

(c) GA: It is inspired by the process of natural selection and genetics. Its working
areas in engineering applications include structural design, material selection,
and manufacturing optimization. The practical example of GA is in the opti-
mization of aircraft wing design. GA can be used to optimize the wing shape
and size, reducing drag, and improving the lift-to-drag ratio of the aircraft.
(d) ABC: It is evolved by the foraging behavior of honey bees. It has been used in a
variety of engineering areas like image processing, power system optimization,
and parameter tuning. One practical real application of ABC is in the optimization
of water distribution networks. ABC can be used to optimize the placement and
sizing of pipes, reducing leaks, and improving the efficiency of the network.
(e) FA: It is originated by the flashing behavior of fireflies. It can be used in robotics,
power system optimization, and signal processing. One of the most common prac-
tical applications of FA is in the optimization of building energy systems. FA can be
used to optimize the control of heating, ventilation, and air conditioning systems,
reducing energy consumption, and improving the comfort of the occupants.
(f) BF: It mimics the behavior of bacteria foraging for food. Some examples of
practical engineering applications are signal-image processing, sensor net-
works, control systems, electrical power systems, bioinformatics, environ-
mental, chemical, and mechanical engineering. For instance, it can be used to
optimize the load dispatch problem, which involves scheduling the generation
and distribution of electrical power in a system. In transportation engineering, it
optimizes the traffic signal timings and routing in transportation networks.

3.8 Conclusion
NIOAs have been shown to be effective in solving a wide range of engineering pro-
blems. These algorithms take inspiration from natural phenomena and use them to
create optimization algorithms that can be used to find the best solution to a problem.
76 Nature-inspired optimization algorithms and soft computing

The application areas of these algorithms include engineering optimization, robotics,

finance, image processing, machine learning, and environmental management. These
can solve complex optimization problems and finding the optimal solution.
In this chapter, various application aspects of NIOAs in the area of image
processing like denoising, enhancement, segmentation, feature extraction-selection,
and classification have been discussed. Furthermore, parameter tuning is essential
for the performance of NIOAs. The optimal values for the algorithm’s parameters
depend on the problem’s characteristics and the available computational resources.
A well-tuned algorithm can significantly improve the efficiency and accuracy of
the optimization process. In addition to that, NIOAs have become popular in sol-
ving constrained and unconstrained optimization problems due to their effective-
ness in efficiently searching the solution space and finding the global optimum. The
algorithms can be adapted to handle constraints in constrained optimization pro-
blems through penalty function methods and constraint handling methods. In
unconstrained optimization problems, the algorithms can be classified into
population-based methods and single-point methods. Overall, NIOAs provides a
powerful tool for solving a wide range of optimization problems.

List of Abbreviations
ABC Artificial Bee Colony
ACO Ant Colony Optimization
AI Artificial Intelligence
BA Bat Algorithm
BBBC Big Bang Big Crunch
BIOAs Bio-Inspired Optimization Algorithms
CNN Convolutional Neural Network
CP Constraint Programming
CS Cuckoo Search
CSPs Constraint Satisfaction Problems
DE Differential Evolution
DL Deep Learning
FF Firefly
GA Genetic Algorithm
GS Gravitational Search
HCHTs Hybrid Constraint Handling Techniques
IoT Internet of Things
ML Machine Learning
MR Mutation Rate
MI Mutual Information
NIOAs Nature-Inspired Optimization Algorithms
 Application aspects of nature-inspired optimization algorithms 77

NSIOAs Natural Science Inspired Optimization Algorithms

OF Objective Function
OAs Optimization Algorithms
PSO Particle Swarm Optimization
PS Population Size
ROIs Region of Interests
SA Simulated Annealing
SIFT Scale-Invariant Feature Transform
SQP Sequential Quadratic Programming
TV total variation

References
[1] W. R. Ashby, Principles of Self-Organizations: Transaction. Pergamon
Press, 1962.
[2] I. Fister Jr, X.-S. Yang, I. Fister, J. Brest, and D. Fister, “A brief review of nature-
inspired algorithms for optimization,” arXiv Prepr. arXiv1307.4186, 2013.
[3] S. Forrest, “Genetic algorithms,” ACM Comput. Surv., vol. 28, no. 1,
pp. 77–80, 1996.
[4] S. Das and P. N. Suganthan, “Differential evolution: a survey of the state-
of-the-art,” IEEE Trans. Evol. Comput., vol. 15, no. 1, pp. 4–31, 2010.
[5] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Proceedings
of ICNN’95-International Conference on Neural Networks, 1995, vol. 4,
pp. 1942–1948.
[6] X.-S. Yang, “Firefly algorithms for multimodal optimization,” in
Stochastic Algorithms: Foundations and Applications: 5th International
Symposium, SAGA 2009, Sapporo, Japan, October 26–28, 2009.
Proceedings 5, 2009, pp. 169–178.
[7] D. Karaboga and B. Basturk, “A powerful and efficient algorithm for
numerical function optimization: artificial bee colony (ABC) algorithm,” J.
Glob. Optim., vol. 39, pp. 459–471, 2007.
[8] S. Das, A. Biswas, S. Dasgupta, and A. Abraham, “Bacterial foraging
optimization algorithm: theoretical foundations, analysis, and applica-
tions,” Found. Comput. Intell. Glob. Optim., vol. 3, pp. 23–55, 2009.
[9] M. Dorigo, M. Birattari, and T. Stutzle, “Ant colony optimization,” IEEE
Comput. Intell. Mag., vol. 1, no. 4, pp. 28–39, 2006.
[10] X.-S. Yang and A. Hossein Gandomi, “Bat algorithm: a novel approach for
global engineering optimization,” Eng. Comput., vol. 29, no. 5, pp. 464–
483, 2012.
[11] A. H. Gandomi, X.-S. Yang, and A. H. Alavi, “Cuckoo search algorithm: a
metaheuristic approach to solve structural optimization problems,” Eng.
Comput., vol. 29, pp. 17–35, 2013.
78 Nature-inspired optimization algorithms and soft computing

[12] S. Kirkpatrick, C. D. Gelatt Jr, and M. P. Vecchi, “Optimization by simu-

lated annealing,” Science (80-. )., vol. 220, no. 4598, pp. 671–680, 1983.
[13] E. Rashedi, H. Nezamabadi-Pour, and S. Saryazdi, “GSA: a gravitational
search algorithm,” Inf. Sci. (Ny)., vol. 179, no. 13, pp. 2232–2248, 2009.
[14] O. K. Erol and I. Eksin, “A new optimization method: big bang–big
crunch,” Adv. Eng. Softw., vol. 37, no. 2, pp. 106–111, 2006.
[15] X.-S. Yang, Nature-Inspired Optimization Algorithms. Academic Press, 2020.
[16] E. V. Onet and E. Vladu, “Nature inspired algorithms and artificial intel-
ligence,” J. Comput. Sci. Control. Syst., no. 1, pp. 66, 2008.
[17] X.-S. Yang and X.-S. He, Nature-Inspired Computation in Data Mining and
Machine Learning, 1st Edition. Springer Cham Publishers, ISBN: 978-3-030-
28555-5, September 2020, https://doi.org/10.1007/978-3-030-28553-1.2020.
[18] H. Chiroma, A. Gital, N. Rana, et al., “Nature inspired meta-heuristic
algorithms for deep learning.: no recent progress and novel perspective,” in
Advances in Computer Vision: Proceedings of the 2019 Computer Vision
Conference (CVC), 2020, vol 1 1, pp. 59–70.
[19] S. R. Jino Ramson, K. Lova Raju, S. Vishnu, and T. Anagnostopoulos,
“Nature inspired optimization techniques for image processing—a short
review,” in Nature Inspired Optimization Techniques for Image Processing
Applications. Cham: Springer, 2019, pp. 113–145.
[20] R. J. S. Raj, S. J. Shobana, I. V. Pustokhina, D. A. Pustokhin, D. Gupta, and
K. Shankar, “Optimal feature selection-based medical image classification
using deep learning model in internet of medical things,” IEEE Access,
vol. 8, pp. 58006–58017, 2020.
[21] V. Sharma and A. K. Tripathi, “A systematic review of meta-heuristic
algorithms in IoT based application,” Array, vol. 14, p. 100164, 2022.
[22] M. Dorigo and G. Di Caro, “Ant colony optimization: a new meta-
heuristic,” in Proceedings of the 1999 Congress on Evolutionary
Computation-CEC99 (Cat. No. 99TH8406), 1999, vol. 2, pp. 1470–1477.
[23] M. F. Ahmad, N. A. M. Isa, W. H. Lim, and K. M. Ang, “Differential
evolution: a recent review based on state-of-the-art works,” Alexandria
Eng. J., vol. 61, no. 5, pp. 3831–3872, 2022.
[24] J. H. Holland, Adaptation in Natural and Artificial Systems: an
Introductory Analysis with Applications to Biology, Control, and Artificial
Intelligence. MIT Press, 1992.
[25] X.-S. Yang and S. Deb, “Cuckoo search via Lévy flights,” in 2009 World
Congress on Nature and Biologically Inspired Computing (NaBIC), 2009,
pp. 210–214.
[26] S. Chib and E. Greenberg, “Understanding the metropolis-hastings algo-
rithm,” Am. Stat., vol. 49, no. 4, pp. 327–335, 1995.
[27] P. Kumar, S. Srivastava, and R. Srivastava, “Basic understanding of med-
ical imaging modalities,” in High-Performance Medical Image Processing,
Apple Academic Press, 2021, pp. 1–17.
[28] A. K. Rai and S. Srivastava, “A thorough investigation on image forgery
detection,” C. Model. Eng. \& Sci., vol. 134, no. 3, pp. 1489–1528, 2023.
 Application aspects of nature-inspired optimization algorithms 79

[29] D. Sharma, A. P. Bhondekar, A. K. Shukla, and C. Ghanshyam, “A review

on technological advancements in crowd management,” J. Ambient Intell.
Humaniz. Comput., vol. 9, no. 3, pp. 485–495, 2018.
[30] K. R. Castleman, Digital Image Processing. Prentice Hall Press, 1996.
[31] R. C. Gonzalez, Digital Image Processing. Pearson Education India, 2009.
[32] R. Srivastava and S. Srivastava, “Restoration of Poisson noise corrupted
digital images with nonlinear PDE based filters along with the choice of
regularization parameter estimation,” Pattern Recognit. Lett., vol. 34, no.
10, pp. 1175–1185, 2013.
[33] R. Kumar, R. Srivastava, and S. Srivastava, “Microscopic biopsy image
segmentation using hybrid color K-means approach,” Int. J. Comput. Vis.
Image Process., vol. 7, no. 1, pp. 79–90, 2017.
[34] V. P. Singh, S. Srivastava, and R. Srivastava, “Effective mammogram
classification based on center symmetric-LBP features in wavelet domain
using random forests,” Technol. Heal. Care, vol. 25, no. 4, pp. 709–727,
2017.
[35] S. Srivastava, N. Sharma, and S. K. Singh, “Empirical analysis of super-
vised and unsupervised filter based feature selection methods for breast
cancer classification from digital mammograms,” Int. J. Comput. Appl.,
vol. 88, no. 8, 2014.
[36] K. Jani, R. Srivastava, and S. Srivastava, “Computer aided medical image
analysis for capsule endoscopy using multi-class classifier,” in 2019 IEEE
5th International Conference for Convergence in Technology (I2CT), 2019,
pp. 1–5.
[37] S. Sarkar, S. Das, and S. S. Chaudhuri, “Multi-level thresholding with a
decomposition-based multi-objective evolutionary algorithm for segment-
ing natural and medical images,” Appl. Soft Comput., vol. 50, pp. 142–157,
2017.
[38] P. Sukhija, S. Behal, and P. Singh, “Face recognition system using genetic
algorithm,” Procedia Comput. Sci., vol. 85, pp. 410–417, 2016.
[39] M. Sabeti, R. Boostani, and B. Davoodi, “Improved particle swarm opti-
misation to estimate bone age,” IET Image Process., vol. 12, no. 2,
pp. 179–187, 2018.
[40] G. JayaBrindha and E. S. G. Subbu, “Ant colony technique for optimizing
the order of cascaded SVM classifier for sunflower seed classification,”
IEEE Trans. Emerg. Top. Comput. Intell., vol. 2, no. 1, pp. 78–88, 2017.
[41] H.-F. Kuo and Frederick, “Ant colony optimization-based freeform sources
for enhancing nanolithographic imaging performance,” IEEE Trans.
Nanotechnol., vol. 15, no. 4, pp. 599–606, 2016.
[42] A. Mostafa, A. Fouad, M. A. Elfattah, et al., “CT liver segmentation using
artificial bee colony optimisation,” Procedia Comput. Sci., vol. 60,
pp. 1622–1630, 2015.
[43] A. K. Bhandari, A. Kumar, and G. K. Singh, “Tsallis entropy based multi-
level thresholding for colored satellite image segmentation using evolu-
tionary algorithms,” Expert Syst. Appl., vol. 42, no. 22, pp. 8707–8730, 2015.
80 Nature-inspired optimization algorithms and soft computing

[44] A. Kumar and S. Srivastava, “Restoration and enhancement of breast ultra-

sound images using extended complex diffusion based unsharp masking,”
Proc. Inst. Mech. Eng. Part H J. Eng. Med., vol. 236, no. 1, pp. 12–29, 2022.
[45] N. B. Bahadure, A. K. Ray, and H. P. Thethi, “Comparative approach of
MRI-based brain tumor segmentation and classification using genetic
algorithm,” J. Digit. Imaging, vol. 31, pp. 477–489, 2018.
[46] V. Kumar and S. Srivastava, “Performance analysis of reshaped Gabor
filter for removing the Rician distributed noise in brain MR images,” Proc.
Inst. Mech. Eng. Part H J. Eng. Med., vol. 236, no. 8, pp. 1216–1231, 2022.
[47] P. M. Narendra, “A separable median filter for image noise smoothing,”
IEEE Trans. Pattern Anal. Mach. Intell., no. 1, pp. 20–29, 1981.
[48] F. Jin, P. Fieguth, L. Winger, and E. Jernigan, “Adaptive Wiener filtering
of noisy images and image sequences,” in Proceedings 2003 International
Conference on Image Processing (Cat. No. 03CH37429), 2003, vol. 3,
pp. III–349.
[49] J. D. Villasenor, B. Belzer, and J. Liao, “Wavelet filter evaluation for
image compression,” IEEE Trans. Image Process., vol. 4, no. 8, pp. 1053–
1060, 1995.
[50] M. Elad, “On the origin of the bilateral filter and ways to improve it,” IEEE
Trans. Image Process., vol. 11, no. 10, pp. 1141–1151, 2002.
[51] A. Buades, B. Coll, and J.-M. Morel, “Non-local means denoising,” Image
Process. Line, vol. 1, pp. 208–212, 2011.
[52] S. Srivastava, N. Sharma, R. Srivastava, and S. K. Singh, “Restoration of digital
mammographic images corrupted with quantum noise using an adaptive total
variation (TV) based nonlinear filter,” in 2012 International Conference on
Communications, Devices and Intelligent Systems (CODIS), 2012, pp. 125–128.
[53] R. R. Kumar, A. Kumar, and S. Srivastava, “Anisotropic diffusion based
unsharp masking and crispening for denoising and enhancement of MRI
images,” in 2020 International Conference on Emerging Frontiers in
Electrical and Electronic Technologies (ICEFEET), 2020, pp. 1–6.
[54] S. Srivastava, R. Srivastava, N. Sharma, S. K. Singh, and S. Sharma, “A
fourth-order PDE-based non-linear filter for speckle reduction from optical
coherence tomography images,” Int. J. Biomed. Eng. Technol., vol. 10, no.
1, pp. 55–69, 2012.
[55] A. Kumar, P. Kumar, and S. Srivastava, “A skewness reformed complex
diffusion based unsharp masking for the restoration and enhancement of
Poisson noise corrupted mammograms,” Biomed. Signal Process. Control,
vol. 73, pp. 103421, 2022.
[56] U. K. Acharya and S. Kumar, “Genetic algorithm based adaptive histogram
equalization (GAAHE) technique for medical image enhancement,” Optik
(Stuttg)., vol. 230, pp. 166273, 2021.
[57] R. Srivastava, J. R. P. Gupta, H. Parthasarthy, and S. Srivastava, “PDE based
unsharp masking, crispening and high boost filtering of digital images,” in
Contemporary Computing: Second International Conference, IC3 2009,
Noida, India, August 17–19, 2009. Proceedings 2, 2009, pp. 8–13.
 Application aspects of nature-inspired optimization algorithms 81

[58] S. Winkler, “Vision models and quality metrics for image processing
applications,” Verlag nicht ermittelbar, 2001.
[59] S. Khalid, T. Khalil, and S. Nasreen, “A survey of feature selection and
feature extraction techniques in machine learning,” in 2014 Science and
Information Conference, 2014, pp. 372–378.
[60] Z. Daixian, “SIFT algorithm analysis and optimization,” in 2010
International Conference on Image Analysis and Signal Processing, 2010,
pp. 415–419.
[61] N. Dalal and B. Triggs, “Histograms of oriented gradients for human
detection,” in 2005 IEEE Computer Society Conference on Computer
Vision and Pattern Recognition (CVPR’05), 2005, vol. 1, pp. 886–893.
[62] H. Bay, A. Ess, T. Tuytelaars, and L. Van Gool, “Speeded-up robust features
(SURF),” Comput. Vis. image Underst., vol. 110, no. 3, pp. 346–359, 2008.
[63] H. Abdi and L. J. Williams, “Principal component analysis,” Wiley
Interdiscip. Rev. Comput. Stat., vol. 2, no. 4, pp. 433–459, 2010.
[64] B. F. Darst, K. C. Malecki, and C. D. Engelman, “Using recursive feature
elimination in random forest to account for correlated variables in high
dimensional data,” BMC Genet., vol. 19, no. 1, pp. 1–6, 2018.
[65] S. Balakrishnama and A. Ganapathiraju, “Linear discriminant analysis—a
brief tutorial,” Inst. Signal Inf. Process., vol. 18, no. 1998, pp. 1–8, 1998.
[66] A. Kraskov, H. Stögbauer, and P. Grassberger, “Estimating mutual infor-
mation,” Phys. Rev. E, vol. 69, no. 6, pp. 66138, 2004.
[67] P. Liashchynskyi and P. Liashchynskyi, “Grid search, random search, genetic
algorithm: a big comparison for NAS,” arXiv Prepr. arXiv1912.06059, 2019.
[68] R. Kumar, R. Srivastava, and S. Srivastava, “Detection and classification of
cancer from microscopic biopsy images using clinically significant and biolo-
gically interpretable features,” J. Med. Eng., vol. 2015, Article no. 457906, 2015.
[69] P. Kumar, S. Srivastava, R. K. Mishra, and Y. P. Sai, “End-to-end improved
convolutional neural network model for breast cancer detection using mam-
mographic data,” J. Def. Model. Simul., vol. 19, no. 3, pp. 375–384, 2022.
[70] P. Kumar, A. Kumar, S. Srivastava, and Y. Padma Sai, “A novel bi-modal
extended Huber loss function based refined mask RCNN approach for
automatic multi instance detection and localization of breast cancer,” Proc.
Inst. Mech. Eng. Part H J. Eng. Med., vol. 236, no. 7, pp. 1036–1053, 2022.
[71] X.-S. Yang and X. He, “Nature-inspired optimization algorithms in engineer-
ing: overview and applications,” Nat.-Inspired Comput. Eng., pp. 1–20, 2016.
[72] A. M. Hemeida, S. Alkhalaf, A. Mady, E. A. Mahmoud, M. E. Hussein, and
A. M. B. Eldin, “Implementation of nature-inspired optimization algo-
rithms in some data mining tasks,” Ain Shams Eng. J., vol. 11, no. 2,
pp. 309–318, 2020.
[73] X.-S. Yang (Ed.), “Chapter 14 — How to deal with constraints,” in Nature-
Inspired Optimization Algorithms (Second Edition), Academic Press, 2021,
pp. 207–220. https://doi.org/10.1016/B978-0-12-821986-7.00021-4.
[74] J. Nocedal, “Theory of algorithms for unconstrained optimization,” Acta
Numer., vol. 1, pp. 199–242, 1992.
82 Nature-inspired optimization algorithms and soft computing

[75] M. J. Box, “A new method of constrained optimization and a comparison

with other methods,” Comput. J., vol. 8, no. 1, pp. 42–52, 1965.
[76] D. P. Bertsekas, Constrained Optimization and Lagrange Multiplier
Methods. Academic Press, 2014.
[77] K. Deb, “An efficient constraint handling method for genetic algorithms,”
Comput. Methods Appl. Mech. Eng., vol. 186, no. 2–4, pp. 311–338, 2000.
[78] P. E. Gill, W. Murray, and M. H. Wright, Practical Optimization. SIAM,
2019.
[79] E. Mezura-Montes and C. A. C. Coello, “Constraint-handling in nature-
inspired numerical optimization: past, present and future,” Swarm
Evol. Comput., vol. 1, no. 4, pp. 173–194, 2011.
[80] C. A. C. Coello, “Use of a self-adaptive penalty approach for engineering
optimization problems,” Comput. Ind., vol. 41, no. 2, pp. 113–127, 2000.
[81] L. Jacob, G. Obozinski, and J.-P. Vert, “Group lasso with overlap and graph
lasso,” in Proceedings of the 26th Annual International Conference on
Machine Learning, 2009, pp. 433–440.
[82] H. Zou and T. Hastie, “Regularization and variable selection via the elastic net,”
J. R. Stat. Soc. Ser. B (Statistical Methodol.), vol. 67, no. 2, pp. 301–320, 2005.
[83] A. B. Owen, “A robust hybrid of lasso and ridge regression,”
Contemp. Math., vol. 443, no. 7, pp. 59–72, 2007.
[84] A. R. Conn, N. I. M. Gould, and P. L. Toint, Trust Region Methods. SIAM, 2000.
[85] D. Bertsimas and J. N. Tsitsiklis, Introduction to Linear Optimization,
vol. 6. Athena Scientific Belmont, MA, 1997.
[86] W. I. Zangwill, “Non-linear programming via penalty functions,” Manage.
Sci., vol. 13, no. 5, pp. 344–358, 1967.
[87] C. A. Floudas and V. Visweswaran, “Quadratic optimization,” in
Handbook of Global Optimization. Springer, 1995, pp. 217–269.
[88] V. Kumar, “Algorithms for constraint-satisfaction problems: a survey,” AI
Mag., vol. 13, no. 1, pp. 32, 1992.
[89] F. Rossi, P. Van Beek, and T. Walsh, Handbook of Constraint
Programming. Elsevier, 2006.
[90] S. C. Brailsford, C. N. Potts, and B. M. Smith, “Constraint satisfaction
problems: algorithms and applications,” Eur. J. Oper. Res., vol. 119, no. 3,
pp. 557–581, 1999.
[91] P. T. Boggs and J. W. Tolle, “Sequential quadratic programming for large-
scale nonlinear optimization,” J. Comput. Appl. Math., vol. 124, no. 1–2,
pp. 123–137, 2000.
[92] S.-J. Kim, K. Koh, M. Lustig, S. Boyd, and D. Gorinevsky, “An interior-
point method for large-scale l1-regularized least squares,” IEEE J. Sel.
Top. Signal Process., vol. 1, no. 4, pp. 606–617, 2007.
[93] E. Wong, Active-set Methods for Quadratic Programming. San Diego:
University of California, 2011.
[94] Y. Wang, Z. Cai, Y. Zhou, and Z. Fan, “Constrained optimization based on
hybrid evolutionary algorithm and adaptive constraint-handling technique,”
Struct. Multidiscip. Optim., vol. 37, pp. 395–413, 2009.
 Application aspects of nature-inspired optimization algorithms 83

[95] C. A. C. Coello, “Theoretical and numerical constraint-handling techniques

used with evolutionary algorithms: a survey of the state of the art,” Comput.
Methods Appl. Mech. Eng., vol. 191, no. 11–12, pp. 1245–1287, 2002.
[96] A. Mani and C. Patvardhan, “A novel hybrid constraint handling technique
for evolutionary optimization,” in 2009 IEEE Congress on Evolutionary
Computation, 2009, pp. 2577–2583.
[97] H. Garg, “A hybrid PSO-GA algorithm for constrained optimization pro-
blems,” Appl. Math. Comput., vol. 274, pp. 292–305, 2016.
[98] B. Meyer, “Hybrids of constructive metaheuristics and constraint pro-
gramming: a case study with aco,” in Hybrid Metaheuristics: An Emerging
Approach to Optimization, 2008, pp. 151–183.
[99] G. Chandrashekar and F. Sahin, “A survey on feature selection methods,”
Comput. Electr. Eng., vol. 40, no. 1, pp. 16–28, 2014.
[100] B. Xue, M. Zhang, W. N. Browne, and X. Yao, “A survey on evolutionary
computation approaches to feature selection,” IEEE Trans. Evol. Comput.,
vol. 20, no. 4, pp. 606–626, 2015.
[101] C.-L. Huang and C.-J. Wang, “A GA-based feature selection and para-
meters optimization for support vector machines,” Expert Syst. Appl.,
vol. 31, no. 2, pp. 231–240, 2006.
[102] M. Rostami, S. Forouzandeh, K. Berahmand, and M. Soltani, “Integration
of multi-objective PSO based feature selection and node centrality for
medical datasets,” Genomics, vol. 112, no. 6, pp. 4370–4384, 2020.
[103] L. G. Fahad, S. F. Tahir, W. Shahzad, M. Hassan, H. Alquhayz, and R.
Hassan, “Ant colony optimization-based streaming feature selection: an
application to the medical image diagnosis,” Sci. Program., vol. 2020,
pp. 1–10, 2020.
[104] M. S. Uzer, N. Yilmaz, and O. Inan, “Feature selection method based on
artificial bee colony algorithm and support vector machines for medical
datasets classification,” Sci. World J., vol. 2013, Article ID 419187, 2013.
[105] M. N. Sudha and S. Selvarajan, “Feature selection based on enhanced
cuckoo search for breast cancer classification in mammogram image,”
Circuits Syst., vol. 7, no. 4, pp. 327–338, 2016.
[106] R. Sawhney, P. Mathur, and R. Shankar, “A firefly algorithm based
wrapper-penalty feature selection method for cancer diagnosis,” in
Computational Science and Its Applications–ICCSA 2018: 18th
International Conference, Melbourne, VIC, Australia, July 2–5, 2018,
Proceedings, Part I 18, 2018, pp. 438–449.
[107] J. Nayak, B. Naik, A. K. Das, and D. Pelusi, “Nature inspired optimization
and its application to engineering,” Evol. Intell., vol. 14, pp. 1–3, 2021.
This page intentionally left blank
 Chapter 4
Particle swarm optimization applications and
implications
Sapna Katiyar1 and Aakansha Garg2

4.1 Introduction to PSO

The particle swarm optimization (PSO) algorithm was developed by Kennedy and
Eberhart in 1995 [1]. PSO algorithm is swarm-based algorithm, which mimics the
social behavior of animals like flocks, birds, fish, and insect. Swarms tend to vary
their search pattern corresponding to their own learning experience or from others.
Therefore, they use collaborative approach to search food [2,3].
PSO algorithm design key idea is strongly associated with two studies:
● Evolutionary algorithm: Swarm approach used in PSO, facilitates searching
larger regions in solution space to get optimized objective function.
● Artificial life cycle: PSO analyses the artificial structure with life traits.
There are five fundamental principles, derived from the social animal beha-
viors including artificial life cycle. These principles are proximity, quality, stabi-
lity, diverse response, and adaptability. In PSO algorithm, particles have the
tendency to update their velocity and position as environment varies. The swarm
movement is not restricted; however, they constantly search for optimal solution in
search space [4].
In standard PSO model, every bird is characterized as a point in cartesian
coordinate system. Some initial velocity and position are arbitrary assigned to
each bird. Nearest proximity velocity match rule is applied to execute the pro-
gram and hence every entity has similar speed as its closest neighbor. With
multiple iterations, suddenly all points have equal velocity. To make the entire
simulation in real situation, a random variable is added up to the velocity element
in each iteration [2,5].

1
Impledge Technologies, India
2
ABES Institute of Technology – Ghaziabad, India
86 Nature-inspired optimization algorithms and soft computing

4.1.1 PSO elements

Standard PSO is implemented by considering many elements. Few key elements
are [3,6] the following:
● Particle: Describe particles in real numbers as Pi.
● Fitness Function: Generally, it is objective function and used to determine
fitness function.
● Local Best (pbest): It is particle best position among all positions visited by particle.
● Global Best (gbest): Particle visits many positions; among those positions
wherever best fitness is attained.
● Velocity Update: It is a vector, which determines the particle velocity and direction.
● Position Update: For optimal fitness, all particles tend to move for best
position. In PSO to get global optima, every particle updates their position.

4.1.2 PSO algorithm

Let us assume position and velocity coordinates of each bird as (x,y) and (Vx, Vy).
If the corn field position coordinate is (x0, y0), current position and velocity per-
formance are measured by the distance between corn field and current position
[7,8]. It is assumed that each bird can memorize the best position reached (pbest)
due to their memory. Other parameters are velocity adjusting constant (a) and
random number on [0,1] (rand). Set of rules are there to establish the change in
velocity item:
if x > pbestx, vx = vx
rand  a, otherwise, vx = vx + rand  a
if y > pbesty, vy = vy
rand  a, otherwise, vy = vy + rand  a
Another assumption about the swarm is that they can communicate. Every
particle memorizes their best location as gbest. If velocity adjusting constant is b,
then adjusted velocity item must be updated as per the mentioned rules:
if x > gbestx, vx = vx
rand  b, otherwise, vx = vx + rand  b
if y > gbesty, vy = vy
rand  b, otherwise, vy = vy + rand  b
In PSO algorithm, every particle is considered without mass and volume, so it
is known as particle swarm optimization algorithm [2,6]. Figure 4.1 illustrates the
flowchart of the PSO algorithm. From simulation result of PSO algorithm, it is
observed that if the value of
● a/b is small, particles will gather nearby corn field slowly and wobblingly.
● a/b is comparatively large, all particles will gather in corn field speedily.
For PSO in continuous D-dimension space coordinates, let us assume:
● Swarm size is N
● Position vector of individual particle is Xi = (xi1, xi2, . . . , xid, . . . , xiD)
● Velocity vector is Vi = (vi1, vi2, . . . , vid, . . . , viD)
● Individual particle optimal position is Pi = (pi1, pi2, . . . , pid, . . . , piD)
● Swarm’s optimal position is characterized as Pg = (pg1, pg2, . . . , pgd, . . . , pgD)
 Particle swarm optimization applications and implications 87

Figure 4.1 Flowchart of PSO algorithm

In the original PSO algorithm, individual particle optimal position is modified

by (4.1).
( d
xi;tþ1 ; if f ðXi;tþ1 Þ < f ðPi;t Þ
pi;tþ1 ¼
d
(4.1)
pdi;t ; otherwise

The optimal position of swarm is the overall distinct optimal solutions.

Velocity and position are updated by (4.2).

udi;tþ1 ¼ udi;t þ c1  rand  ðpdi;t
xdi;t Þ

þ c2  rand  ðpdg;t
xdi;t Þ (4.2)
xdi;tþ1 ¼ xdi;t þ udi;tþ1
88 Nature-inspired optimization algorithms and soft computing

For optimization problems, the original version of PSO is not very efficient.
Therefore, it is modified by introducing inertia weight in the velocity vector
updating equation. This updated algorithm is known as canonical PSO algorithm.
So now, (4.2) can be written as follows:

udi;tþ1 ¼ w  udi;t þ c1  rand  ðpdi;t
xdi;t Þ

(4.3)
þ c2  rand  ðpdg;t
xdi;t Þ

Further, PSO algorithm is modified by considering constriction factor (c). This

factor guarantees convergence and advance convergence rate. Therefore velocity
update equation can be written as follows:

udi;tþ1 ¼ cðudi;t þ f1  rand  ðpdi;t
xdi;t Þ

(4.4)
þ f2  rand  ðpdg;t
xdi;t Þ

In the above two equations, there is a very slight difference. Both equations
would be identical if suitable parameters are chosen.

4.1.3 Standard pseudo code

Researchers have made several attempts to modify the PSO algorithm to improve
the overall performance [3,9]. The pseudo code for standard PSO is “initialize
population at random.”
While (Size of Population)
{
Loop
Evaluate fitness
If fitness value is superior to pbest (best fitness value) in record, then
Update pbest with latest value of pbest
end loop
Choose best fitness value of particle among all gbest
While max. iterations or min. error criteria is not reached
{
Update particle velocity as per equation:

Vitþ1 ¼ W  Vit þ c1 U1t ðPtb1
Pti Þ þ c2 U2t ðgbt
Pti Þ (4.5)

Move the particles to the latest position

Ptþ1
i ¼ Pti þ vtþ1
i (4.6)
Next
}
}

4.1.4 PSO advantages and disadvantages

Since there are many optimization algorithms, particle swarm optimization algo-
rithm is relatively cheaper, fast, and efficient. Few parameters are involved in the
 Particle swarm optimization applications and implications 89

Figure 4.2 Advantages and disadvantages of PSO algorithm

algorithm, which makes it simple [1,5]. It is best fit for discrete, continuous, non-
linear, and non-convex kind of problems. Figure 4.2 illustrates various advantages
and disadvantages of PSO.

4.1.5 PSO applications

Table 4.1 summarizes the PSO applications in various fields for optimization. PSO has
the capability to simulate the motion of particle swarm, so it can also be applied for
visual effects such as specialized effects in Hollywood film [10,11]. Different fields
where PSO has successfully applied in real-world applications are industrial automation,
robotics, combinatorial problems, signal processing, image processing, healthcare, net-
working, biomedical, power generation, path planning and optimization, environmental
applications, smart cities, commercial applications, and prediction and forecasting [6].

4.2 Outline of swarm intelligence

Swarm intelligence (SI), which falls into the umbrella of Artificial Intelligence [50],
has always drawn the attention of researchers. In early 1960s itself many swarm-based
optimization algorithms have been proposed. In 1980, the word SI was envisioned as a
buss word and slowly it has covered broad category. A swarm can be observed as a
collection of several thousands of individuals, which have decided their own desire to
converge for a common goal. SI depicts the self-configured, coordinated, complex
clustered, flexible, and robust action of a group. SI systems comprises many entities
and the individuals are somewhat consistent. In such systems, interactions among the
entities are established on the basis of simple behavioral rules and overall system
behavior is the outcome of group behavior, i.e., individual interaction along with sur-
rounding environment. The two fundamentals and essential properties of SI are:
● Self-organization: It is characterized as the system ability to grow its com-
ponents or agents into appropriate manner without any outside support.
Table 4.1 Summary of PSO applications for different application areas

Reference Authors Application Area Strength Limitations

[12] Pashaei et al. Healthcare (medical image Better performance due to hybrid Computation time is not
segmentation and disease algorithm considered
[13] Zeng et al. diagnosis) Accomplished using SVM model Experimental analysis is done
and deep learning algorithm with one data set only
[14] Jain et al. Quicker and more reliable approach Computation time is not
with good local optimum solution in considered
standard BPSO
[15] Li et al. Improved performance with Restricted activity of image
superior contextual details segmentation problems
[16] Raj and Ray High accuracy, better sensitivity Lack of real-time analysis and
with improved productivity more computation time
[17] Srisukkham Expedited chaotic investigation. Only particle’s position is
et al. Modified BBPSO algorithm is updated. Velocity is not
applied considered here
[18] Kumar et al. Environmental applications Improved accuracy and overall Lack of investigation in real-
(agricultural and performance with the help of hybrid time environment
environmental forecasting, model
[19] Zarel et al. prediction, and flood routing) Hybrid evolutionary algorithm is Computation time can be
used, which gives high reliability reduced further, and monetary
and improved water release values cost can be estimated
[20] Chen et al. Hybrid forecasting model is used, Certain emergencies like major
which gives improved accuracy holidays and few influential
with less run time factors are missing
[21] Rahgoshay Uses support vector method. Better Very simple structure and
et al. accuracy and less computation time complex parameters are missing
[4] Kour and Hybrid PSO-SVM algorithm is used Plants having medicinal values
Arora for plants classification and are not considered, which have
segmentation. Higher accuracy with major significance
less computation time
[22] Cao et al. Modified BPSO is used to monitor Hyper-spectral sensors are used,
water quality. Fast convergence and which has low spectral
good stability resolution. Other lakes can also
be investigated with same
approach
[23] Ghorbani Hybrid ANN-QPSO gives better Doubtfulness in the forecasting
et al. accuracy in forecasting evaporation data
[24] Etheram Uses hybrid BA and PSO algorithm Creating Muskingum model with
et al. for flood routing. Best solution is few parameters only
obtained with reduced computation
time
[25] Camci et al. Achieved improved performance for Inefficient control of quadcopter
rice farm inspection using for natural landscapes
Quadcopter. Capable to handle
uncertainties and noise also
[26] Maiyar and Industrial applications Decision support tool helps to Perishable food grain products
Thakkar (nuclear power plants, wind reduce food grain wastage. have not taken into consideration
turbines, TV panels, food Economic system
[27] Alnaqi et al. transportation, airborne Hybrid algorithm is used for Consideration for scalability of
system platforms) prediction. Provides good such systems is missing
performance and reliability
[28] Wang et al. Hybrid algorithm and fault Real time data is not evaluated
diagnosis method gives better so that estimation is missing
accuracy with less training time and
noise
[29] Liu et al. Hybrid algorithm is used for design It is a multi-objective
optimization, which offers better optimization problem but only
platform stability and decreases three targets were considered
mechanical resonance
(Continues)
Table 4.1 (Continued)

Reference Authors Application Area Strength Limitations

[30] Ghorbani Hybrid algorithm is applied for PV- Overall system accuracy is not
et al. WT generation unit. It gives better evaluated
reliability with reduced cost
[31] Jiang et al. Decision support system works well Not capable to detect spam
for online customer reviews. It is review and decision can be
low cost and less time consuming biased by sentimental reviews
[32] Song et al. Proper positioning of wind turbines Optimization for tedious and
gives more production of power practical wind farms assessment
with less system cost is missing
[33] Qi et al. This approach offers better accuracy Long-term capability is not
with less cost and response time anticipated
[6] Lopes et al. Hybrid algorithm is developed for Investigation is done on tiny data
electric load distribution. It gives set. Computation time is not
good quality solutions and reduces calculated
industry production cost
[34] Jiao et al. Commercial applications Hybrid algorithm for location Concept of multi objective is not
estimation is deployed. It provides incorporated
better transportation facility with
reduced cost
[35] Tang et al. Building materials price forecasting Computation time evaluation is
is done via hybrid approach. Offers missing
fast convergence with better
stability
[36] Shen and Provides real time profit calculation Computational cost is not
Han for accounting information system considered in overall profit
calculation
[37] Yi et al. It offers many advantages so can be For such systems, scalability
successfully applied practically. must be computed
Accuracy is also improved
[7] Kumar et al. NN model provides statistically Overall system computation time
prominent results. Offers improved calculation is missing
accuracy
[38] Abid et al. Smart city (highways, smart Algorithm is used for energy Accuracy needs to be calculated
homes, residential areas) management. Offers low cost and and present techniques can be
low power consumption compared with others
[39] Zhang et al. Locations for electric vehicles Only two scenarios have been
charging stations are identified taken for testing, which can be
using PSO. Creates more revenue increased
and provides more coverage
[40] Li et al. Hybrid model-based algorithm is This procedure is very time
applied for traffic forecasting. It consuming and data set is also
provides good stability with very small
improved accuracy
[41] Jordehi Hybrid algorithm is used, which For such systems scalability can
reduces daily consumer bill be considered
[42] Le et al. Robust hybrid PSO algorithm is Cost calculation can be done,
applied for investigating building and accuracy can be further
heating load. It offers good improved
reliability
[43] Ma et al. Scheduling model is developed for Sometimes doubtful for
appliance, which provides good renewable energy generation
coverage and profits by selling
electricity
[5] Hu et al. Hybrid algorithm is applied to Cost and scalability are not
schedule traffic light in urban area estimated. Some case studies of
organized cities can also be
included
[44] Sato et al. Quality of solution is improved for Less stability and no cost
power grids in smart cities estimation. Robustness needs to
be proved
(Continues)
Table 4.1 (Continued)

Reference Authors Application Area Strength Limitations

[45] Zarrouk Scheduling PSO algorithm provides better Cost factor is not taken into
et al. quality solutions with lower CPU consideration, which is a very
time important factor in such type of
problems
[46] Mokhtari Jobs are done punctually without Problem size is small. Only
and Noroozi any delay commercial optimization is
considered, which can be
extended further
[47] Nouiri et al. Approach is very effective for real Every entity does not participate
time scenario. Decision making is in final decision. Very high
decentralized energy consumption
[48] Zhong et al. Path planning and PSO is implemented for Traveling Can be compared with another
optimization salesman problem. It offers good significant optimization
balance between distance travelled algorithm
and time
[49] Thabit and Path planning is done for multi It can be investigated for real
Mohades robots. This algorithm provides risk time scenario
free, short, and smooth paths
 Particle swarm optimization applications and implications 95

Scheduling Routing

Clustering Optimization
Swarm
Intelligence
Capabilities

Figure 4.3 Capabilities of SI

Self-organization always depends on positive and negative feedbacks, numer-

ous interactions, and variations.
● Division of Work: This property defines synchronized execution of multiple
viable and straight forward jobs by individuals. It allows swarms to deal with
complicated nature problem, where individuals are working collectively.
Depending upon the interesting properties of SI systems, many capabilities are
offered.
Figure 4.3 shows the following SI capabilities:
Clustering: It is the procedure of automatic grouping. In SI, cluster is the group of
like and unlike agents in another cluster.
Scheduling: The focus is on job’s relative position instead of its direct successor
or predecessor in the process of scheduling. Finally, summation assessment rule or
global pheromone assessment rule is applied.
Routing: It is meant for end to end deliver and enhanced connectivity. As in the
case of ants, backward moving ants utilize the information collected by forward
ants about their route from source to destination.
Optimization: It is defined as a process of any problem to determine the best
possible solution or minimum cost function among all feasible solutions.

4.2.1 General swarm principles

To frame the wide behavior arises from the swarm and to establish the relationship
between swarm and PSO, various principles of SI are presented in Figure 4.4:
● Proximity Principle: Every entity of swarm should react back to environ-
mental fluctuations initiated by agents’ interaction. Nest building and living
resource searching are the basic behavior of agents.
● Stability Principle: Every time when environment changes, mode of behavior
of population should not change.
● Adaptability Principle: Swarm is sensitive for environmental changes, which
may depict varied swarm behavior.
96 Nature-inspired optimization algorithms and soft computing

Proximity
Principle

Diverse Stability
Response
Principle Principle
General
Swarm
Principles

Quality Adaptability
Principle Principle

Figure 4.4 General principles of SI

● Quality Principle: Swarm should have the capability to react for quality fac-
tors like deciding safe location.
● Diverse Response Principle: Resources must note be accumulated into nar-
rowing area. Distribution must be planned so that each agent can be highly
protected against environmental fluctuations.
In 1989, the concept of SI was presented by Gerardo Beni and Jing Wang. It is
a rising field of biologically inspired artificial intelligence, which is based on the
behavior theory of social insects like bat, bees, ants, wasps etc. Advantages and
disadvantages of Swarm Intelligence is listed in Figure 4.5. Multiple SI-based
algorithms were developed and applied effectively multi-domain problems. There
are many ways to classify the Swarm based algorithm, but one significant cate-
gorization is based on insect, bacteria, bird, wild, and amphibious. Figure 4.6 shows
some well-known and proven optimization algorithms.
● Ant colony optimization (ACO) is truly inspired by the searching behavior of
actual ants. It is a population-based metaheuristic approach, and the idea was
proposed by Marco Dorigo in his PhD thesis in 1992 [51]. Several real-life
examples of ACO are job scheduling, travelling salesman problem, path opti-
mization, timetable scheduling, distribution planning, etc. [52].
● Cuckoo search (CS) algorithm was proposed in 2009 by Xin-She Yang and Deb. It is
nature-inspired metaheuristic algorithm, encouraged by the brood parasitism of certain
cuckoo varieties by leaving their eggs in the nest of host birds of other varieties [53].
● Teaching learning-based optimization (TLBO) algorithm was developed in
2011 by Rao et al. This algorithm mimics the classroom environment of
teacher and learners, for optimizing any given objective function. There are
two phases: teacher phase and learner phase [54,55].
● Particle swarm optimization (PSO) algorithm has many similarities with
genetic algorithm and evolutionary computation. It is an iterative approach; its
 Particle swarm optimization applications and implications 97

• Flexibility: Always respond to internal instabilities and external challenges

• Robustness: Tasks are accomplished even if some agents be unsuccessful
• Scalability: From a limited agents to many millions
• Decentralized: No centralized control in colony
Advantages • Self-organized: Evolving solutions rather than pre-defined
• Adaptation: The swarm structure is highly adaptable even for new stimuli
• Speed: Any changes in network can be communicated/propagated very quickly
• Modularity: No dependency of agents on other layers of network
• Parallelism: Inherently parallel operation capability of agents

• Behavior: It is very difficult to forecast the behavior from the distinct rules
• Information: The colony functionality can not be predicted with the information of
individual agent functioning
• Sensitivity: A minor change in simple rules may impact the behaviors of different group
Disadvantages level
• Act: Agent behavior may be seen as noise in action of choice is stochastic
• Non optimal: Because swarm systems are highly redundant and there is no central control
• Uncontrollable: Difficult to exercise control over swarm
• Unpredictable: Swarm system complexity leads to unforeseeable results
• Non immediate: Complex swarm systems with rich hierarchies take time

Figure 4.5 Advantages and disadvantages of SI

Ant Colony
Hunting Cuckoo
Optimization
Search Search
(ACO)
(HS) (CS)
Artificial
Immune Intelligent
System Water Drop
(AIS) (IWD)

Gravitational Paticle
Search Swarm
Algorithm Optimization
(GSA) Swarm Intelligence (PSO)
Algorithms
Orcas
Algorithm
(OA) Fish Swarm
Algorithm
(FSA)
Tunicate
Swarm Firefly
Algorithm Krill Herd Bat
Algorithm Algorithm
(TSA) Algorithm
(KH) (FA)
(BA)

Figure 4.6 Different types of SI algorithms

98 Nature-inspired optimization algorithms and soft computing

aim is to improve the candidate solution with respect to assigned quality

measures in search space [6,11].
● Bat algorithm (BA) is inspired by the behavior of bats, and the idea was pro-
posed in 2012 by Yang and Gandomi. Microbats have the tendency of echo-
location with changing loudness and pulse rate of radiation [24,56]. It is
somewhat similar with PSO and consist of position and velocity equations.
● Firefly algorithm (FA) was established by Xin-She Yang in 2008. It is created on
the flashing designs and behavior of tropical fireflies. Fireflies’ fitness is evaluated
by their blinking brightness [19,57]. FA is easy to implement and simple in nature.
● Fish swarm algorithm (FSA) was proposed by Li et al. in 2002. It is inspired by
the regular behavior of fish for food search. In view of optimization, fish
behavior is related to learning capacity, which directs fish swarm for searching
fresh food sources, i.e., design space. FSA can avoid local minimums to
accomplish global optimization [58,59].
● Krill Herd (KH) algorithm is best suitable for continuous optimization pro-
blems. In 2012, KH algorithm was introduced by Gandomi and Alavi for
global standard function [60,61]. This algorithm mimics the herding actions of
krill individuals. In comparison to many SI algorithms, it is robust, efficient, no
implementation complexity, and require few control parameters.
● Tunicate swarm algorithm (TSA) is anticipated by Kaur et al. in 2020. It is
encouraged by the swarm actions of tunicate to live effectively in ocean depth.
TSA uses population best solution and works effective for optimizing non-
linear constraint problems [62].
● Orcas algorithm (OA) is encouraged by the lifestyle of Killer Whale or Orca,
especially the hunting methods and echolocation actions [63]. It is developed
as artificial intelligence grounded on the practical actions for known an
ordinary adoption in association with the stages of selection and speeding.
● Gravitational search algorithm (GSA) is extensively applied in machine
learning and is proposed by Rashedi et al. in 2009. GSA is encouraged by the
theory of Newton’s Law of Gravity and motion. In this algorithm, the best
solution is the solution with the heavier mass [64,65].
● Artificial immune system (AIS) is the computationally intelligent algorithm,
resulting from the principles encouraged by human immune system [66,67].
AIS algorithm was proposed by Farmer et al. in 1986 and Bersini et al. in
1990. This algorithm is very sophisticated and adaptive in nature.
● Hunting search (HS) algorithm is stimulated by the group hunting actions of
predatory animals such as wolves, lions, and dolphins. All these hunters have
different way of hunting, but one thing is common, which is group prey. The
idea of HS algorithm is anticipated by Oftadeh et al. in 2010 [68].

4.3 PSO for single-objective problem

When optimization problem involves only one objective function, the task of
finding the optimal solution is known as single-objective optimization problem.
 Particle swarm optimization applications and implications 99

PSO is a metaheuristic technique, and at present, it is one of the highly defined and
most extensively used SI algorithm. Because of the simplicity of PSO, it has wide
range of applications in problems for optimizing single objective [2,4]. PSO is
problem independent algorithm, where merely single information is needed to run
the algorithm, which is the fitness calculation for candidate solution.
Standard form of PSO is altered and modified, many variants were proposed by
researchers to deal with distinct types of problems [5]. Some examples are travel-
ling salesman problem, task planning problem, optimizing logistic planning, UAV
mission planning, flood control and routing, and optimizing water harvesting.

4.4 PSO for multi-objective problem

In multi-objective optimization problems, numerous objectives are required to be
handed at the same time [9,26]. In many situations, optimization of two or more
than two parameters are required so PSO suits well for such multi optimization.
Additional situation is when two nondominant solutions are of similarly good
quality, then which one is to be chosen to use as best particle position. The range
difficulty can be handled by integrating an extra set, known as external repository.
This repository stores all nondominant solutions found throughout search. But the
choice of highly appropriate repository member depends upon the nature of
method. The restricted size of outer repository has certain constraints so that pre-
sent solutions can be replaced with the new one with the help of specific rules [29].
The common multi-objective PSO structure can be expressed with mentioned
pseudocode.
Start
Initialize parameters (swarm, velocities, and best positions)
Initialize external repository (initially it must be blank)
While (criteria to stop in not satisfied) Do
For every Particle
Choose a member of external repository (if required)
Update velocity & position
Calculate the fresh position
Updating the best position & external repository
End For
End While
End

4.5 Different approaches of multi-objective PSO

Many applications suffer from such problems where two or more than two-
objective functions need to be handled simultaneously. Nature of problem may
have multiple criteria’s; therefore, the concept of Pareto optimality has come
up. Many approaches are there for multi-objective PSO based on objective func-
tions and Pareto-based methods separately [31,39].
100 Nature-inspired optimization algorithms and soft computing

4.5.1 Objective function aggregation approach

This approach combines all the objective functions over weighted combination into
one.
X
k
FðxÞ ¼ wi fi ðxÞ (4.7)
i¼1

where wi is the nonnegative weights such that

X
k
wi ¼ 1
i¼1

Here, optimization is applied to F(x) like single-objective optimization. During

algorithm execution, if weights remain stable, then this approach is conventional
weighted aggregation (CWA) [69]. It is very simple approach but has many lim-
itations also. For a specific given weight setting when PSO is applied, it gives only
one solution. Therefore, this algorithm needs to be repeated and again for different
weight settings to have required solutions. CWA method cannot find solutions in
concave area of Pareto front. These limitations can be addressed by dealing
dynamic weight adjustment. Big-bang weighted aggregation (BWA) is one such
approach.

4.5.2 Objective function ordering approach

In this approach, first ranking of objective functions is to be determined. Then each
function is minimized separately in the order of priority. In 2002, Hu and Eberhart
suggested a scheme based on such ranking. Pareto front puts limits on fitness value
space and algorithm keeps easiest objective function fixed and minimizes the
remaining objective functions. In such approach, PSO variants with dynamic
neighborhoods is used, where controlled solutions are considered as particles best
position. Objective function ordering approach works well for two-objective
functions only, but its performance is not justified for more than two-objective
functions [70,71]. This method is modified by integrating some exterior memory to
store nondominated results and, hence decreasing the computational cost. This
approach received restricted acceptability and not capable to address the binary
string problem.

4.5.3 Non-Pareto, vector-evaluated approach

Vector-evaluated PSO (VEPSO) is a multi-swarm algorithm. It is established on the
concept of vector-evaluated genetic algorithm (VEGA). In VEPSO algorithm, one
swarm is dedicated to one-objective function and whole estimation is done for that
objective function. Then for updating swarm, one swarm best position is utilized
for velocity update of other swarm with different objective function [72].
If there are k swarms in a problem with k objectives. Let vi[s] indicates ith
particle velocity in the sth swarm, where s = 1, . . . , k; therefore, the velocity is
 Particle swarm optimization applications and implications 101

updated using the following equation:

½s ½s ½s ½s
vij ðt þ 1Þ ¼ w vij ðtÞ þ c1 r1 ðpij ðtÞ
xij ðtÞÞ þ c2 r2
½q ½s
(4.8)
ðpgj ðtÞ
xij ðtÞÞ;

where p[q]
g is the best position related to qth swarm, calculated with qth objective
function.
A modified version of VEPSO is employed, where one processor indicates
each swarm. The total number of swarms need not to be same as overall objective
functions. In this system, swarms can communicate with other swarms through
Island Migration approach. This algorithm is effectively applied for regulating
generator contributions for transmission systems and optimizing radiometer array
antenna [73].
One more comparable approach like VEPSO was suggested, known as multi-
species PSO. This algorithm was established and linked to robotics for self-
sufficient agent response knowledge framework. Subswarms were utilized, which
form groups, one is used for every objective function. Then again, every subswarm
is estimated with its specific objective function and greatest particle information is
conveyed to nearby subswarms for particles velocity update. Therefore, ith particle
velocity for sth swarm is modified as per equation:
½s ½s ½s ½s
vij ðt þ 1Þ ¼ vij ðtÞ þ a1 ðpij ðtÞ
xij ðtÞÞ þ a2
½s ½s
(4.9)
ðpgj ðtÞ
xij ðtÞÞÞ þ A;

where
X
Hs
½l ½s
A¼ ðpgj ðtÞ
xij ðtÞÞ
l¼1

Hs is the total number of swarms, which conveys with sth swarm. pg[l] is the lth
swarm’s best position.

4.5.4 Algorithms based on Pareto dominance

The approaches that fall into this category apply the idea of Pareto Dominance to
identify particle’s best position, which will further drive swarm in the course of
searching. Therefore, Pareto-based PSO methods turn out to be the budding
research area. Some substantial advancements in this field are listed here.
1. Multi-objective PSO (MOPSO) is the initial Pareto-based PSO method. In
MOPSO, repository is used to save nondominant results, which particles
investigate. Entire search space is split into hypercubes and each hypercube is
allocated a fitness value [74]. The fitness value is inversely proportional to the
quantity of particles in hypercube. Here Classical Roulette Wheel selection is
applied for picking a hypercube. The best particle position is updated after
every iteration depending upon current best and its new position. Therefore,
102 Nature-inspired optimization algorithms and soft computing

the ith particle velocity update in the form of best position (pi) and repository
leader (Rh) will be:

vijðt þ 1Þ ¼ w vijðtÞ þ c1 r1ðpijðtÞ
xijðtÞÞ þ c2 r2ðRh ðtÞ
xijðtÞÞ

(4.10)

Since the repository has restricted size so when it is full, new solutions can
be stored based on retention criteria.
2. Repository insufficiencies to store limited records is overcome in another
approach. It uses relatively complicated tree-type structure, known as domi-
nant tree, for addressing the unconstrained repository upkeep. Except for
repository upkeep, this algorithm acts like MOPSO. Inclusion of one more
feature into this algorithm makes it more efficient, which is mutation, also
known as craziness. Mutation acts for particle velocity, which conserves
diversity [75].
3. Another algorithm is based on the selection criteria of leaders. More precisely
every particle is evaluated for each objective function independently. Swarm
global best position is updated as per mean of the best particle for each func-
tion. Particle diversity is conserved over the distance calculation. Choice of
leader is inclined with respect to non-dominant solution, which promotes the
mitigation of particles gathering in groups. This algorithm is examined over the
number of problems to estimate efficiency [76].
4. Maximin PSO is one more approach, which utilized maximin fitness function.
Fitness function for a particular decision vector (x) with swarm size (N) and
number of objective functions (k) is given as follows:

max min ffi ðxÞ
fi ðyÞg (4.11)

j¼1;...;N ;x6¼y i¼1;...;k

Those decision vectors whose maximin function value is smaller than zero is
considered as nondominant solutions. Swarm diversity is encouraged by
maximin function because it corrects particles clusters [77]. There is one
debatable point about it that it encourages intermediate solutions in convex
faces and severe solutions in concave faces. This point can be handled by using
a sufficiently large group of swam.
5. Another MOPSO technique with crowding distance mechanism is developed.
It is beneficial for selecting globally the best particle and nondominant solu-
tions removal from outer archive. To maintain diversity, mutation is taken into
consideration. For every nondominant solution, crowding distance is calculated
individually [78].
Let us assume,
● R as outer archive
● f1, f2 . . . , fk are objective functions
● then the calculation of crowding distance of p is R
● q is the point on R, which instantly follows p in sorting
 Particle swarm optimization applications and implications 103

By sorting all points in R, w.r.t. their objective value, fi and taking,

CD fj ¼ fj ðqÞ
fj ðrÞ (4.12)
Therefore, the overall crowding distance of p is written as follows:
X
k
CD fj (4.13)
j¼1

Swarm leaders are selected from the fraction of nondominant point of R, which
are having maximum crowding distance. Mutation can also be applied upon par-
ticles to encourage swarm diversity.

4.6 Variants of PSO algorithm

Particle swarm optimization (PSO) was recommended by Eberhart and Kennedy in
1995. Statistically, it is proven but researchers kept trying efforts to enhance the
standard PSO. Therefore, many variants of PSO are there, which are differentiated
based on different parameters such as initialization, indolence weight, mutation
operator, and constriction factor [10,38,41]. Some types of PSO variants are shown
in Figure 4.7. The following are the different PSO variants.

4.6.1 Discrete PSO

The discrete PSO algorithm gives fast convergence and improved performance.
Here the position of the particle has a discrete value. This algorithm can be effi-
ciently applied to combinatorial and discrete optimization problems. Generally,
when integer solutions are required, an optimal solution can be generated by
rounding off the actual best value to the closest integer.

Immunity Discrete PSO

Enhance Binary PSO
PSO

Neighborhood Variants of PSO

Adaptive PSO
SPSO

Neighborhood Hybrid PSO

CGPSO

Figure 4.7 Different variants of PSO

104 Nature-inspired optimization algorithms and soft computing

4.6.2 Binary PSO

In the binary PSO algorithm, every particle position is represented by either 0 or 1
i.e., in binary value. Every particle in the population takes a binary decision in
terms of 1 (True/Yes) or 0 (False/No). Particles value keeps changing from 0 to 1 or
1 to 0. In this algorithm, velocity must be confined in the range of [0, 1].

4.6.3 Adaptive PSO

Adaptive PSO provides improved search efficiency than standard PSO. PSO
parameters like acceleration coefficients, inertia weight, etc. are adjusted in real
time for finer performance. Though it involves many new parameters, it does not
create any implementation complexity. Globally it is observed as the best particle
to come out as likely local optima.

4.6.4 Hybrid PSO

Fresh and more refined PSO variants are continuously innovated to achieve better
optimization performance. In research, there is always a trend for integrating the
capabilities of different evolutionary computational methods, so same can be
combined with PSO.

4.6.5 Neighborhood-guaranteed convergence PSO

This variant of PSO models the social networks structuring. Neighborhood guar-
anteed convergence PSO (GCPSO) permits particles to investigate bigger area of
search space. The objective of neighborhood is actually to conserve the hetero-
geneity within the swarm by impending the information flow through the network.
Velocity update must be done to implement neighborhood in the standard PSO.
Equation (4.1) is used to update velocity:
viðt þ 1Þ ¼ X ðviðtÞ þ c1r1ðpbestðtÞ
xiðtÞÞ þ c2r2ðgbestðtÞ
xiðtÞÞÞ
(4.14)

where gbest: best position found so far in the neighborhood of the ith particle.

4.6.6 Neighborhood search strategies PSO

This algorithm applies two global neighborhood searches and one local neighbor-
hood search. Two operations are there:
● For every particle, neighborhood search generates three trial particles.
● Among the three trial particles, the best is selected and that is taken as fresh
current particle.

4.6.7 Immunity-enhanced PSO

In this algorithm, a portion of particle population is sampled in workable space.
Thereafter the population executes algorithm and PSO variants by updating velo-
city and position. Then non-uniform mutation is executed in accordance with some
 Particle swarm optimization applications and implications 105

probability. So fresh generation is obtained with the help of selection operator

when particles fly.

4.7 PSO in hybrid environment

Hybridization is defined as a broad model, which integrates two or more algo-
rithms. The objective is to utilize their advantages and reduce their barriers.
Through algorithm hybridization, two important properties of algorithms i.e.
exploitation and exploration, can be entirely enhanced. Likewise, the PSO algo-
rithm can be combined with another optimization algorithms [18,20]. PSO in
hybridization helps to increase particles quality, algorithm efficiency, and effec-
tiveness. PSO algorithm tends to trap in local optima, so it is unable to discover the
entire region of search space. This difficulty can be overcome while combining
PSO with others and simultaneously defeating the integral deficiencies of other
algorithms [27]. Many hybrid algorithms have been proposed by researchers for
solving multiple nature of problems with increased performance. Some of such
algorithms have been discussed here.
(A) PSO with Genetic Algorithm (GA): Like PSO in case of GA, there is
population of possible candidate solutions [21,30]. Every population element
has chromosome, which gets mutated depending upon certain probability to
preserve population diversity level and solution enhancements. During search
process, PSO cannot avoid its trap for local optima though GA algorithm helps
to reduce this weakness along with its operator [79].
(B) PSO with Differential Evolution (DE): DE falls into the category of evolu-
tionary computation. Candidate solution is improved by iterative process and
hence it optimizes any problem. Though DE does not give guarantee optimal
solution, it has an excessive capability to preserve population diversity and
perform local searches in definite search space. Therefore, hybridization of
PSO with DE gives promising results [80,81].
(C) PSO with Artificial Neural Network (ANN): In such a hybridization, model
PSO can be used to train ANN product unit. When it is compared with other
algorithms, it offers less error with improved performance. In another hybrid
model, the data set is divided into three: training set, validation set, and testing
set. PSO algorithm is used to train the ANN architecture along with all nodes
produced when algorithm is initialized [23,82].
(D) PSO with Bat Algorithm (BA): Integration of PSO with BA offers great
convergence rate and avoid trapping into local minima [24,83].
(E) PSO with Firefly Algorithm (FA): Hybrid algorithm of PSO with FA deter-
mines global best fitness value so that it can determine when to initiate local
search [19,84]. Convergence accuracy is improved.
(F) PSO with Sine Cosine Algorithm (SCA): PSO parameters, pbest and gbest
are implanted into the traditional SCA algorithm to enhance finding possible
candidate solution [85]. Therefore, PSO swarm can exploit the possible
search space.
106 Nature-inspired optimization algorithms and soft computing

(G) PSO with Hybrid Whale (HW) Algorithm: Whale optimization algorithm is
popular due to its excellent exploration capability. When it is combined with
PSO, it overcomes the PSO phase limitations [86,87]. Forced whale works for
exploration and capping phenomenon for exploitation phase to converge glo-
bal optimum quickly.
(H) PSO with Many Other Algorithms: It has been observed that PSO algorithm
can be integrated with ant colony optimization (ACO), cuckoo search (CS),
artificial bee colony (ABC) algorithms, gray wolf optimization (GWO), cat
swarm optimization (CWO), salp swarm algorithm (SSA), etc. for enhanced
performance [88–91].

4.8 Computational experiments

Various numerical experiments were performed by integration PSO. Altered ver-

sion of PSO was used to control inaccurate requirements, operating cost, and data
capacity in layered production planning system. It resolved the limitations of pro-
duction plant prototype by minimizing operating cost and risk involved to attain
maximum profit. Such model can be efficiently applied for real-world production
scheduling and planning type of problems with poor defined data [92]. In another
actual machinery system scenario about cut depth, cutting speed, and feeding rate;
process parameters are examined with PSO algorithm for optimizing cutting
situations. Results can be observed with minimum considered standards and asso-
ciated optimal cutting situations [93]. An improved version of PSO algorithm is
utilized to resolve the squared error by using joint fitness function. This error is the
difference between calculated value and demonstrated value for identification
nature problems. Numerical experiments were carried out to verify algorithm per-
formance and authenticating results [94]. Supplementing further improved PSO
algorithm was proposed to verify convergence. To do this, very large dimension,
benchmark objective function is used. Numerical experiments results depict that
the proposed algorithm is very successful [95]. Wireless sensor networks perfor-
mance optimization is done with total number of active nodes, through improved
PSO algorithm. In traditional background, performance of this proposed algorithm
is validated [96].

4.9 Convergence

The word convergence can be defined in two distinct forms in relation with PSO
[9,10]:
● All particles must converge at a point in search space, which is the con-
vergence of series of solutions. Convergence may be optimum or may not be.
● Irrespective of swarm behavior, all particles personal best (p) and swarm’s
position best (g) approaches the problem’s local optimum. Here convergence is
for local optimum.
 Particle swarm optimization applications and implications 107

Series of solutions for convergence has been analyzed for PSO. These analyses
give the outcome in selecting PSO parameters, which are required for convergence
and avoid particle’s divergence. The analysis was criticized because it was too sim-
plified. Assumption was that there is only one particle, which does not use stochastic
variable. Reason for attraction was particles best-known position (p) and swarm’s
best-known position (g), remains unchanged throughout the entire optimization
process [97]. However, for swarm convergence, it has been observed that such
simplifications are not going to impact boundaries defined by this examination for
parameters. In recent years, significant efforts have been made for modifying the
modelling assumption in stability analysis of PSO [98]. Recently, more generalized
results were applied to many PSO variants. It has been observed about the necessary
minimum requirement about the modeling assumption. For local optimum, con-
vergence has been investigated for PSO [99,100]. Evidence shows that some mod-
ifications are required in PSO for achieving guarantee local optimum. Hence, it can
be concluded that PSO parameters and algorithm convergence potentiality depends
upon observational and experimental results. Orthogonal learning strategy can also
be employed for improving already existing information to achieve quicker global
convergence, excellent quality solution, and powerful robustness [101].
If there is a requirement of trade off in between divergence and convergence,
some adaptive kind of mechanism needs to be introduced. In comparison to regular
PSO, adaptive PSO offers improved search efficiency. Adaptive PSO (APSO) can
execute global search in whole search space along with improved convergence rate.
Search efficiency and effectiveness can be improved by controlling real time
algorithm parameters and acceleration coefficients [25]. Though APSO encloses
many new parameters, it does not require additional design and hence no imple-
mentation complexity.

4.10 PSO implications on image processing problems

Optimization algorithms are always essential for answering real-time problems,
particularly where tuning of parameters is required. PSO is one such algorithm, and
it is extensively used in the domain of image processing to determine optimal
solution in search space.
● Image Segmentation: Multiple approaches are there like thresholding, clus-
tering, edge detection, region growing, compression based, histogram based,
watershed algorithms, etc. Standard PSO algorithm and many of its variants
are used effectively with improved performance [4]. Incorporating PSO with
Markov random field (MRF) can additionally increase the image enhancement
performance. Some model of Fuzzy hybrid adaptive PSO along with ant col-
ony optimization is suggested. Identification of cluster partitioning can be done
successfully, which gives superior performance in terms of optimal value and
fast convergence, compared to existing techniques [102,103].
● Feature Selection: Numerous engineering field where image processing is
required, selection of suitable feature should be accomplished [5,87]. Feature
108 Nature-inspired optimization algorithms and soft computing

selection subfields where feature need to be extracted are text recognition, image
recognition, and pattern recognition. Features include contrast, object, size, color
change, texture change, cluster importance, etc. Standard techniques are used so
far, either for global or local selection. Therefore, feature selection is more of NP
hard problem. PSO capability is proven for resolving such problems [5,104]. For
clustering, PSO is used to search centroids of clusters, which offers better result.
● Edge Detection: Edges play an important role in images. In literature,
numerous traditional approaches have been defined, which relies on first order,
second order, Laplacian operator, or Canny edge detector, etc. PSO-based
approaches have proven their potential to overcome many challenges faced by
traditional methods. PSO gives best fitness curve of image, which represents
object boundaries. PSO integrated algorithms gives more precise results for
highly noisy images too [105,106].
● Edge Linking: Many image detection methods undergo for certain limitations
like false edge detection, missing actual edge, detecting thin lines, or produ-
cing appropriate thickness [89]. Connecting broken edges is also a tedious job.
PSO-based techniques are efficient for detecting edges continuous and clear
object images, even in the presence of noise also.
● Image Compression: The purpose behind the image compression is to remove
redundancies available in the image to save storage space and efficient trans-
mission bandwidth. PSO integrated algorithms helps to fulfil the said task and
compressed images benefits for time required to CPU execution [107,108].

4.11 PSO implications on optimum route-finding

problems
Route planning is a process of determining best route, which can be followed by
any object usually robot [24]. Objective is safe travel within two nodes indepen-
dently without no interventions or restricted interventions of humans. PSO sub-
stantially contributes for simple to complex nature route-finding problems [48,49].
Ahmed et al. [109] proposed a PSO-based approach for stochastic dynamic envir-
onment, where such problems can be carried by keeping a trade off in between
short and safest route. Two-objective functions have been introduced, which are
dependent on distance and risk.
● Nasrollahy et al. suggested PSO algorithm for route finding in dynamic
environment along with moving targets and obstacles. PSO tends to identify
collision-free path with minimum distance travelled though avoiding local
optimums. It is achieved by minimizing the cost function and outcome is
enhanced in comparison to other techniques [110].
● Ever et al. developed simplified swarm optimization (SSO) algorithm for route
planning. Outcomes indicate that it does not gather into obstacle space even it
automatically shifted toward obstacle free space [111].
● Li et al. recommended improved PSO, which incorporates several factors for
improved control of path planning. These factors are learning factor, uniform
 Particle swarm optimization applications and implications 109

distribution, exponential attenuation inertia weight, etc. to attain superior result

in lesser iterations [112].
● Standard PSO for path planning has challenges of early convergence, inadequate
global search capability, and dropping into local optimum. Cheng et al. established
integration of improved PSO with Gray wolf optimization (GWO) for route plan-
ning [91]. Proposed algorithm achieved the optimal path quickly without any chaos.
Global search capability is also enhanced along with particles diversity [113].
● Thammachantuek et al. proposed PSO algorithm for route planning with multi-
objective. Weight adjustment approach is used for the particle’s movement in
every iteration so that probability of finding the top fit solution is enhanced.
Suggested algorithm determined the shortest path in both static and dynamic
environment with lesser processing time [114].

4.12 Implementation and results

Among numerous optimization algorithms, PSO algorithm has obtained extensive
attention among researchers in several years. It is very easy and simple to implement.
This chapter supports the concept of Swarm Intelligence and particle swarm optimi-
zation. Moreover, multi-objective problems solution and various approached have been
discussed. To overcome the limitations of PSO, it can be combined with many other
algorithms so that better system efficiency, performance, quality, robustness and con-
vergence can be greatly enhanced. PSO algorithms have extensive applications in
almost all the fields, but this chapter explains its application for image processing and
route-finding problem. Still PSO algorithm continue with many unresolved challenges
of arbitrary convergence and convergence rate, determining easy and effective flexible
algorithm parameters, and stabilizing global exploration with local exploitation.

References
[1] J. Kennedy and R. C. Eberhart, “Particle swarm optimization”, In:
Proceedings of the IEEE International Conference on Neural Networks,
Perth, Australia, pp. 1942–1948, 1995.
[2] T. M. Blackwell, “Particle swarms and population diversity”, Soft
Computing, vol 9(11), pp. 793–802, 2005.
[3] Gad, A. G., “Particle swarm optimization algorithm and its applications: a
systematic review”, Archives of Computational Methods in Engineering,
vol 29, pp. 2531–2561, 2022.
[4] V. P. Kour and S. Arora, “Particle swarm optimization-based support
vector machine (P-SVM) for the segmentation and classification of plants”,
IEEE Access, vol 7, pp. 29374–29385, 2019.
[5] S. W. Lin, K. C. Ying, S. C. Chen, and Z. J. Lee, “Particle swarm optimi-
zation for parameter determination and feature selection of support vector
machines”, Expert Systems with Applications, vol 35, no. 4, pp. 1817–1824,
2008.
110 Nature-inspired optimization algorithms and soft computing

[6] R. F. Lopes, F. F. Costa, A. Oliveira, and A. C. de C. Lima, “Algorithm

based on particle swarm applied to electrical load scheduling in an indus-
trial setting”, Energy, vol 147, pp. 1007–1015, 2018.
[7] D. P. Kumar and V. Ravi, “Forecasting financial time series volatility using
particle swarm optimization trained quantile regression neural network,”
Applied Soft Computing, vol 58, pp. 35–52, 2017.
[8] W. Hu, H. Wang, Z. Qiu, C. Nie, and L. Yan, “A quantum particle swarm
optimization driven urban traffic light scheduling model”, Neural
Computing and Applications, vol 29, pp. 901–911, 2018.
[9] W. D. Chang, “A modified particle swarm optimization with multiple
subpopulations for multimodal function optimization problems”, Applied
Soft Computing, vol 33, pp. 170–182, 2015.
[10] W. Chen, J. Zhang, Y, Lin, et al., “Particle swarm optimization with an
aging leader and challenger”, IEEE Transaction of Evolutionary
Computing, vol 17(2), pp. 241–258, 2013.
[11] J. Geng, M. Li, Z. Dong, and Y. Liao, “Port throughput forecasting by
MARS-RSVR with chaotic simulated annealing particle swarm optimiza-
tion algorithm”, Neurocomputing, vol 147, pp. 239–250, 2014.
[12] E. Pashaei, E. Pashaei, and N. Aydin, “Gene selection using hybrid binary
black hole algorithm and modified binary particle swarm optimization”,
Genomics, vol 111(4), pp. 669–686, 2019.
[13] N. Zeng, H. Qiu, Z. Wang, W. Liu, H. Zhang, and Y. Li, “A new switching-
delayed-PSO-based optimized SVM algorithm for diagnosis of
Alzheimer’s disease”, Neurocomputing, vol 320, pp. 195–202, 2018.
[14] I. Jain, V. K. Jain, and R. Jain, “Correlation feature selection based
improved-binary particle swarm optimization for gene selection and cancer
classification”, Applied Soft Computing, vol 62, pp. 203–215, 2018.
[15] Y. Li, X. Bai, L. Jiao, and Y. Xue, “Partitioned-cooperative quantum-
behaved particle swarm optimization based on multilevel thresholding
applied to medical image segmentation”, Applied Soft Computing, vol 56,
pp. 345–356, 2017.
[16] S. Raj and K. C. Ray, “ECG signal analysis using DCT-based DOST and
PSO optimized SVM”, IEEE Transaction on Instruments and
Measurement, vol 66(3), pp. 470–478, 2017.
[17] W. Srisukkham, L. Zhang, S. C. Neoh, S. Todryk, and C. P. Lim,
“Intelligent leukaemia diagnosis with bare-bones PSO based feature opti-
mization”, Applied Soft Computing, vol 56, pp. 405–419, 2017.
[18] S. Kumar, S. K. Pal, R. Singh, “A novel hybrid model based on particle
swarm optimization and extreme learning machine for short-term tem-
perature prediction using ambient sensors”, Sustainable Cities and Society,
vol 49, 2019.
[19] X. S. Yang, “Firefly algorithms for multimodal optimization”, In: International
Symposium on Stochastic Algorithms, Springer, pp. 169–178, 2009.
[20] S. Chen, J. Q. Wang, and H. Y. Zhang, “A hybrid PSO-SVM model based
on clustering algorithm for short-term atmospheric pollutant concentration
 Particle swarm optimization applications and implications 111

forecasting”, Technology Forecasting and Social Change, Elsevier, vol

146, pp. 41–54, 2019.
[21] M. Rahgoshay, S. Feiznia, M. Arian, and S. A. A. Hashemi, “Simulation of
daily suspended sediment load using an improved model of support vector
machine and genetic algorithms and particle swarm”, Arabian Journal of
Geoscience, vol 12(9), 2019.
[22] Y. Cao, Y. Ye, H. Zhao, et al., “Remote sensing of water quality based on
HJ-1A HSI imagery with modified discrete binary particle swarm
optimization-partial least squares (MDBPSO-PLS) in inland waters: a case
in Weishan lake”, Ecological Informatics, vol 44, pp. 21–32, 2018.
[23] M. A. Ghorbani, R. Kazempour, K. W. Chau, S. Shamshirband, and P. T.
Ghazvinei, “Forecasting pan evaporation with an integrated artificial neural
network quantum-behaved particle swarm optimization model: a case study
in Talesh, Northern Iran”, Engineering Applications of Computational
Fluid Mechanics, vol 12(1), pp. 724–737, 2018.
[24] M. Ehteram, F. B. Othman, Z. M. Yaseen, et al., “Improving the
Muskingum flood routing method using a hybrid of particle swarm opti-
mization and bat algorithm”, Water, vol 10(6), pp. 807, 2018.
[25] E. Camci, D. V. Kripalani, L. Ma, E. Kayacan, and M. A. Khanesar, “An
aerial robot for rice farm quality inspection with type-2 fuzzy neural net-
works tuned by particle swarm optimization-sliding mode control hybrid
algorithm”, Swarm and Evolutionary Computation, vol 41, pp. 1–8, 2017.
[26] M. L. Maiyar and J. J. Thakkar, “Environmentally conscious logistics
planning for food grain industry considering wastages employing multi
objective hybrid particle swarm optimization”, Transportation Research
Part E: Logistics and Transportation Review, vol 127, pp. 220–248, 2019.
[27] A. Alnaqi, H. Moayedi, A. Shahsavar, and T. K. Nguyen, “Prediction of
energetic performance of a building integrated photovoltaic/thermal system
thorough artificial neural network and hybrid particle swarm optimization
models”, Energy Conversion and Management, vol 183, pp. 137–148, 2019.
[28] H. Wang, M. j Peng, J. W. Hines, G. Y. Zheng, Y. K. Liu, and B. R.
Upadhyaya, “A hybrid fault diagnosis methodology with support vector
machine and improved particle swarm optimization for nuclear power
plants”, ISA Transaction, vol 95, pp. 358–371, 2019.
[29] G. Liu, W. Chen, and H. Chen, “Quantum particle swarm with teamwork
evolutionary strategy for multi-objective optimization on electro-optical
platform”, IEEE Access, vol 7, pp. 41205–41219, 2019.
[30] N. Ghorbani, A. Kasaeian, A. Toopshekan, L. Bahrami, and A. Maghami,
“Optimizing a hybrid wind-PV-battery system using GA-PSO and MOPSO for
reducing cost and increasing reliability”, Energy, vol 154, pp. 581–591, 2018.
[31] H. Jiang, C. Kwong, W. Park, and K. Yu, “A multi-objective PSO approach
of mining association rules for affective design based on online customer
reviews”, Journal of Engineering Design, vol 29(7), pp. 381–403, 2018.
[32] M. Song, K. Chen, and J. Wang, “Three-dimensional wind turbine posi-
tioning using gaussian particle swarm optimization with differential
112 Nature-inspired optimization algorithms and soft computing

evolution”, Journal of Wind Engineering and Industrial Aerodynamics, vol

172, pp. 317–324, 2018.
[33] C. Qi, A. Fourie, and Q. Chen, “Neural network and particle swarm opti-
mization for predicting the unconfined compressive strength of cemented
paste backfill”, Construction and Building Materials, vol 159, pp. 473–478,
2018.
[34] R. Jiao, X. Huang, H. Ouyang, G. Li, Q. Zheng, and Z. Jiang, “Optimal
electric business centre location by centre-decentre quantum particle swarm
optimization”, System Science and Control Engineering, vol 7(1), pp. 222–233,
2019.
[35] B. Tang, J. Han, G. Guo, Y. Chen, and S. Zhang, “Building material prices
forecasting based on least square support vector machine and improved
particle swarm optimization”, Architectural Engineering and Design
Management, vol 15(3), pp.196–212, 2019.
[36] J. Shen and L. Han, “Design process optimization and profit calculation
module development simulation analysis of financial accounting informa-
tion system based on particle swarm optimization (PSO)”, Information
Systems and e-Business Management, vol 18, pp. 1–14, 2019.
[37] T. Yi, H. Zheng, Y. Tian, and J. p. Liu, “Intelligent prediction of trans-
mission line project cost based on least squares support vector machine
optimized by particle swarm optimization”, Mathematical Problems in
Engineering, vol 2018, Article ID 5458696, 2018.
[38] S. Abid, A. Zafar, R. Khalid, et al., “Managing energy in smart homes
using binary particle swarm optimization”, In: Conference on Complex,
Intelligent, and Software Intensive Systems, Springer, pp. 189–196, 2017.
[39] Y. Zhang, Q. Zhang, A. Farnoosh, S. Chen, and Y. Li, “GIS-based multi-
objective particle swarm optimization of charging stations for electric
vehicles”, Energy, vol 169, pp. 844–853, 2019.
[40] L. Li, L. Qin, X. Qu, J. Zhang, Y. Wang, and B. Ran, “Day-ahead traffic
flow forecasting based on a deep belief network optimized by the multi-
objective particle swarm algorithm”, Knowledge Based System, vol 172,
pp. 1–14, 2019.
[41] R. Jordehi, “Binary particle swarm optimization with quadratic transfer
function: a new binary optimization algorithm for optimal scheduling of
appliances in smart homes”, Applied Soft Computing, vol 78, pp. 465–480,
2019.
[42] L. T. Le, H. Nguyen, J. Zhou, J. Dou, and H. Moayedi, “Estimating the
heating load of buildings for smart city planning using a novel artificial
intelligence technique PSO-XGBoost”, Applied Sciences, vol 9(13),
p. 2714, 2019.
[43] K. Ma, S. Hu, J. Yang, X. Xu, and X. Guan, “Appliances scheduling via
cooperative multi-swarm PSO under day-ahead prices and photovoltaic
generation”, Applied Soft Computing, vol 62, pp. 504–513, 2018.
[44] M. Sato, Y. Fukuyama, T. Iizaka, and T. Matsui, “Total optimization of
energy networks in a smart city by multi-swarm differential evolutionary
 Particle swarm optimization applications and implications 113

particle swarm optimization”, IEEE Transactions on Sustainable Energy,

vol 99, p. 1, 2018.
[45] R. Zarrouk, I. E. Bennour, and A. Jemai, “A two-level particle swarm
optimization algorithm for the flexible job shop scheduling problem”,
Swarm Intelligence, vol 13, pp. 1–24, 2019.
[46] H. Mokhtari and A. Noroozi, “An efficient chaotic based PSO for earliness/
tardiness optimization in a batch processing flow shop scheduling pro-
blem”, Journal of Intelligent Manufacturingr, Springe, vol 29(5), pp. 1063–
1081, 2018.
[47] M. Nouiri, A. Bekrar, A. Jemai, S. Niar, and A. C. Ammari, “An effective
and distributed particle swarm optimization algorithm for flexible job-shop
scheduling problem”, Journal of Intelligent Manufacturing, vol 29(3),
pp. 603–615, 2018.
[48] Y. Zhong, J. Lin, L. Wang, and H. Zhang, “Discrete comprehensive
learning particle swarm optimization algorithm with metropolis acceptance
criterion for traveling salesman problem”, Swarm and Evolutionary
Computation, vol 42, pp. 77–88, 2018.
[49] S. Thabit and A. Mohades, “Multi-robot path planning based on multi-objective
particle swarm optimization”, IEEE Access, vol 7, pp. 2138–2147, 2018.
[50] S. Katiyar and A. Farhana, “Smart agriculture: the future of agriculture using
AI and IoT”, Journal of Computer Science, vol 17(10), pp. 984–999, 2021.
[51] M. Dorigo and T. Stützle, Ant Colony Optimization, MIT Press, Cambridge,
MA, 2004.
[52] Y. Tamura, T. Sakiyama, and I. Arizono, “Ant colony optimization using
common social information and self memory”, Complexity, vol 2021, pp.
1–7, 2021.
[53] A. S. Joshi, O. Kulkarni, G. M. Kakandikar, and V. M. Nandedkar,
“Cuckoo search optimization- a review”, Materials Today: Proceedings,
vol 4, pp. 7262–7269, 2017.
[54] Y. K. Chen, S. X. Weng, and T. P. Liu, “Teaching–Learning Based
Optimization (TLBO) with variable neighbourhood search to retail shelf-
space allocation”, Mathematics, vol 8(8), p. 1296, 2020.
[55] D. Wu, S. Wang, Q. Liu, L. Abualigah, and H. Jia, “An improved teaching-
learning-based optimization algorithm with reinforcement learning strategy
for solving optimization problems”, Computational Intelligence and
Neuroscience, vol 2022, Article ID 1535957, 2022.
[56] C. Enache, V. Sgarciu, and A. P. Nita, “Intelligent feature selection method
rooted in Binary Bat Algorithm for intrusion detection”, In: Proceedings of
the IEEE International Symposium on Applied Computational Intelligence
and Informatics, Romania, 2015.
[57] R. R. Chhikara, P. Sharma, and L. Singh, “An improved dynamic discrete
firefly algorithm for blind image steganalysis”, International Journal of
Machine Learning and Cybernetics, vol 9, pp. 821–835, 2018.
[58] K. C. Lin, S. Y. Chen, and J. C. Hung, “Feature selection and parameter
optimization of support vector machines based on modified artificial fish
114 Nature-inspired optimization algorithms and soft computing

swarm algorithms”, Mathematical Problem in Engineering, vol 58,

p. 604108, 2015.
[59] Y. Chen, Q. Zhu, and H. Xu, “Finding rough set reduces with fish swarm
algorithm”, Knowledge Based Systems, vol 81, pp. 22–29, 2015.
[60] A. H. Gandomi and A. H. alavi, “Krill Herd Algorithm: a new bio-inspired
optimization algorithm”, Communications in Nonlinear Science and
Numerical Simulation, vol 17, pp. 4831–4845, 2012.
[61] A. L. Bolaji, M. A. Al-Betar, M. A. Awadallah, A. T. Khader, and L. M.
Abualigah, “A comprehensive review: Krill Herd algorithm (KH) and its
applications”, Applied Soft Computing, vol 49, pp. 437–446, 2016.
[62] S. Kaur, L. K. Awasthi, A. L. Sangal, and G. Dhiman, “Tunicate Swarm
Algorithm: a new bio-inspired based metaheuristic paradigm for global
optimization”, Engineering Applications of Artificial Intelligence, vol 90,
p. 103541, 2020.
[63] Y. Jiang, Q. Wu, S. Zhu, and L. Zhang, “Orca predation algorithm: a novel
bio-inspired algorithm for global optimization problems”, Expert Systems
with Applications, vol 188, p. 116026, 2022.
[64] K. M. Htay, R. R. Othman, A. Amir, and J. M. H. Alkanaani, “Gravitational
search algorithm-based strategy for combinatorial t-way test suite generation”,
Journal of Computer and Information Sciences, vol 34(8), pp. 4860–4873, 2022.
[65] R. Shankar, N. Ganesh, R. Čep, R. C. Narayanan, S. Pal, and K. Kalita,
“Hybridized particle swarm—gravitational search algorithm for process
optimization”, Processes, vol 10(3), p. 616, 2022.
[66] N. Rashid, J. Iqbal, F. Mahmood, A. Abid, U. S. Khan, and M. I. Tiwana,
“Artificial immune system–Negative Selection Classification Algorithm
(NSCA) for four class Electroencephalogram (EEG) signals”, Frontier in
Human Neuroscience, vol 12, p. 439, 2018.
[67] S. Aldhaheri, D. Alghazzawi, L. Cheng, A. Barnawi, and B. A. Alzahrani,
“Artificial immune systems approaches to secure the internet of things: a
systematic review of the literature and recommendations for future research,”
Journal of Network and Computer Applications, vol 157, p. 102537, 2020.
[68] R. Oftadeh, M. J. Mahjoob, and M. Shariatpanahi, “A novel meta-heuristic
optimization algorithm inspired by group hunting of animals: hunting
search”, Computers & Mathematics with Applications, vol 60(7), pp. 2087–
2098, 2010.
[69] Y. Jin, M. Olhofer, and B. Sendhoff, “Evolutionary dynamic weighted
aggregation for multi-objective optimization: why does it work and how?”,
In: Proceedings of the GECCO Conference, San Francisco, CA,
pp. 1042–1049, 2001.
[70] X. Hu and R. Eberhart, “Multi-objective optimization using dynamic neigh-
bourhood particle swarm optimization”, In: Proceedings of the IEEE Congress
Evolutionary Computation, IEEE Service Centre, pp. 1677–1681, 2002.
[71] X. Hu, R. C. Eberhart, and Y. Shi, “Particle swarm with extended memory
for multi-objective optimization”, In: Proceedings of the IEEE Swarm
Intelligence Symposium, IEEE Service Centre, pp. 193–197, 2003.
 Particle swarm optimization applications and implications 115

[72] D. Gies and Y. R. Samii, “Vector evaluated particle swarm optimization

(VEPSO): optimization of a radiometer array antenna”, In: Proceedings of
International Symposium (Digest) of Antennas and Propagation Society,
vol 3, pp 2297–2300, 2004.
[73] J. G. Vlachogiannis and K. Y. Lee, “Determining generator contributions to
transmission system using parallel vector evaluated particle swarm optimiza-
tion”, IEEE Transactions on Power Systems, vol 20(4), pp. 1765–1774, 2005.
[74] C. A. Coello, G. T. Pulido, and M. S. Lechuga, “Handling multiple objec-
tives with particle swarm optimization”, IEEE Transaction of Evolutionary
Computing, vol 8(3), pp. 256–279, 2004.
[75] J. E. Fieldsend, R. M. Everson, and S. Singh, “Using unconstrained elite
archives for multi-objective optimization”, IEEE Transaction on
Evolutionary Computing, vol 7(3), pp. 305–323, 2003.
[76] X. H. Huo, L. C. Shen, and H. Y. Zhu, “A smart particle swarm optimi-
zation algorithm for multi-objective problems”, Lecture Notes in Computer
Science, vol 4115, pp. 72–80, 2006.
[77] X. Li, “Better spread and convergence: particle swarm multi-objective
optimization using the maximin fitness function”, Lecture Notes in
Computer Science, vol 3102, pp. 117–128, 2004.
[78] C. R. Raquel and P. C. Naval, “An effective use of crowding distance in
multi-objective particle swarm optimization”, In: Proceedings of the
GECCO, ACM Press, pp. 257–264, 2005.
[79] C. F. Juang, “A hybrid of genetic algorithm and particle swarm optimiza-
tion for recurrent network design”, IEEE Transaction on Systems, Man and
Cybernetics, vol 34, pp. 997–1006, 2004.
[80] L. Xiao and X. Zuo, “Multi-DEPSO: A DE and PSO based hybrid algo-
rithm in dynamic environments”, In: Proceedings of the IEEE Congress on
Evolutionary Computation, Australia, pp 1–7, 2012.
[81] M. G. Omran, A. P. Engelbrecht, and A. Salman, “Bare bones differential
evolution”, European Journal of Operation Research, vol 196, pp. 128–
139, 2009.
[82] C. Zhang, H. Shao, and Y. Li, “Particle swarm optimisation for evolving
artificial neural network”, In: Proceedings of the IEEE International
Conference on Systems, Man and Cybernetics (SMC), USA, vol 4,
pp. 2487–2490, 2000.
[83] A. Ferdowsi, S. Farzin, S. F. Mousavi, and H. Karami, “Hybrid bat &
particle swarm algorithm for optimization of labyrinth spillway based on
half & quarter round crest shapes”, Flow Measurement and
Instrumentation, vol 66, pp. 209–217, 2019.
[84] I. B. Aydilek, “A hybrid firefly and particle swarm optimization algorithm
for computationally expensive numerical problems”, Applied Soft
Computing, vol 66, pp. 232–249, 2018.
[85] H. Nenavath, R. K. Jatoth, and S. Das, “A synergy of the sine cosine algo-
rithm and particle swarm optimizer for improved global optimization and
object tracking”, Swarm Evolutionary Computing, vol 43, pp. 1–30, 2018.
116 Nature-inspired optimization algorithms and soft computing

[86] N. M. Laskar, K. Guha, I. Chatterjee, S. Chanda, K. L. Baishnab, and P. K.

Paul, “HWPSO: a new hybrid whale-particle swarm optimization algorithm
and its application in electronic design optimization problems”, Applied
Intelligence, vol 49(1), pp. 265–291, 2019.
[87] M. M. Mafarja and S. Mirjalili, “Hybrid Whale Optimization Algorithm
with simulated annealing for feature selection”, Neurocomputing, vol 260,
pp. 302–312, 2017.
[88] S. Katiyar, R. Khan, and S. Kumar, “Artificial Bee Colony algorithm for
fresh food distribution without quality loss by delivery route optimization”,
Journal of Food Quality, vol 2021, pp. 1–9, 2021.
[89] M. Schiezaro and H. Pedrini, “Data feature selection based on Artificial
Bee Colony algorithm”, EURASIP Journal of Image and Video Processing,
vol 47, pp. 1–8, 2013.
[90] K. C. Lin, K. Y. Zhang, Y. H. Huang, J. C. Hung, and N. Yen, “Feature
selection based on an improved cat swarm optimization algorithm for big data
classification”, Journal of Supercomputing, vol 72, pp. 3210–3221, 2016.
[91] E. Emary, H. M. Zawbaa, and C. Grosan, “Experienced gray wolf optimi-
zation through reinforcement learning and neural networks”, IEEE
Transaction of Neural Networks and Learning Systems, vol 29, pp. 1–14,
2017.
[92] A. Carlisle and G. Dozier, “An off-the-shelf PSO”, In: Proceedings of the
Workshop on Particle Swarm Optimization, Indiana, USA, 2001.
[93] M. R. Ganesh, R. Krishna, K. Manikantan, and S. Ramachandran, “Entropy
based binary particle swarm optimization and classification for ear detec-
tion”, Engineering Applications of Artificial Intelligence, vol 27, pp. 115–
128, 2014.
[94] J. Lu, H. Hu, and Y. Bai, “Generalized radial basis function neural network
based on an improved dynamic particle swarm optimization and ADA-
boost algorithm,” Neurocomputing, vol 152, pp. 305–315, 2015.
[95] M. R. Sierra and C. A. C. Coello, “Improving PSO-based multi-objective
optimization using crowding, mutation and epsilon-dominance”, Lecture
Notes on Computer Science, vol 3410, pp. 505–519, 2005.
[96] S. Salehian and S. K. Subraminiam, “Unequal clustering by improved
particle swarm optimization in wireless sensor network”, Procedia
Computer Science, vol 62, pp. 403–409, 2015.
[97] M. E. H. Pedersen and A. J. Chipperfield, “Simplifying particle swarm
optimization”, Applied Soft Computing, vol 10(2), pp. 618–628, 2010.
[98] W. C. Cleghorn, “Particle swarm convergence: standardized analysis and
topological influence”, Lecture Notes in Computer Science, vol 8667,
pp. 134–145, 2014.
[99] F. V. D. Bergh and A. P. Engelbrecht, “A convergence proof for the par-
ticle swarm optimiser”, Fundamenta Informatics, vol 105, pp. 1–47, 2010.
[100] M. R. Bonyadi and Z. Michalewicz, “A locally convergent rotationally
invariant particle swarm optimization algorithm”, Swarm Intelligence, vol
8(3), pp. 159–198, 2014.
 Particle swarm optimization applications and implications 117

[101] Z. H. Zhan, J. Zhang, Y. Li, and Y. H. Shi, “Orthogonal learning particle

swarm optimization”, IEEE Transactions on Evolutionary Computation,
vol 15(6), pp. 832–847, 2011.
[102] M. Maitra and A. Chatterjee, “A hybrid cooperative comprehensive learn-
ing based PSO algorithm for image segmentation using multilevel thresh-
olding”, Expert Systems with Applications, vol 34(2), pp. 1341–1350, 2008.
[103] A. Chander, A. Chatterjee, and P. Siarry, “A new social and momentum
component adaptive PSO algorithm for image segmentation”, Expert
Systems with Applications, vol 38, pp. 4998–5004, 2011.
[104] Z. Xinchao, “A perturbed particle swarm algorithm for numerical optimi-
zation”, Applied Soft Computing, vol 10(1), pp. 119–124, 2010.
[105] N. S. Dagar and P. K. Dahiya, “Edge detection technique using binary
particle swarm optimization”, Procedia Computer Science, vol 67,
pp. 1421–1436, 2020.
[106] C. Dongyue, T. Zhou, and X. Yu, “A new method of edge detection based
on PSO”, In Advances in Neural Networks, pp 239–246, 2012.
[107] S. M. Ibrahim, E. A. Zanaty, and H. Alkinani, “Medical image compression
based on wavelets and particle swarm optimization”, Computers, Materials
and Continua, vol 67(2), pp. 1577–1593, 2021.
[108] D. E. Touil and N. Terki, “Quality preserved colour image compression
using particle swarm optimization algorithm”, In: International Symposium
on Modelling and Implementation of Complex Systems, vol 156,
pp. 175–187, 2020.
[109] A. A. Hilli, M. A. Ibadi, A. M. Alfadhel, S. H. Abdulshaheed, and A. H.
Hadi, “Optimal path finding in stochastic quasi-dynamic environments
using particle swarm optimization”, Expert Systems with Applications, vol
186, p. 115706, 2021.
[110] A. Z. Nasrollahy and H. H. S. Javadi, “Using particle swarm optimization
for robot path planning in dynamic environments with moving obstacles
and target,” In: UKSim European Symposium on Computer Modelling and
Simulation, pp 60–65, 2009.
[111] Y. K. Ever, “Using simplified swarm optimization on path planning for
intelligent mobile robot”, Procedia Computer Science, vol 120, pp. 83–90,
2017.
[112] X. Li, D. Wu, J. He, M. Bashir, and M. Liping, “An improved method of
particle swarm optimization for path planning of mobile robot”, Journal of
Computer Engineering, vol 2020, Article ID 3857894, 2020.
[113] X. Cheng, J. Li, C. Zheng, J. Zhang, and M. Zhao, “An improved PSO-
GWO algorithm with chaos and adaptive inertial weight for robot path
planning”, Frontiers in Neurorobotics, vol 15, 770361, 2021.
[114] I. Thammachantuek and M. Ketcham, “Path planning for autonomous
mobile robots using multi-objective evolutionary particle swarm optimi-
zation”, PLoS One, vol 17(8), e0271924, 2022.
This page intentionally left blank
 Chapter 5
Advanced optimization by nature-inspired
algorithm
Altaf Q.H. Badar1

5.1 Introduction
The human race is currently witnessing an era where the world is changing at a very
fast pace, technologically. The innovations and inventions have been taking place
on a daily basis. Also, it is not only engineering field that is experiencing this fast
cycle of new incoming technologies but also various fields like medical, social
engineering, etc. are also accepting emerging techniques.
One of the most influential reasons for these changes being visible is the
advent of Artificial Intelligence (AI). AI is applied and has affected all aspects of
life. Some of the examples are presented below:
● AI used to write research papers [1]
● AI’s advent into legal services [2]
● AI in medical field [3]
● AI applications in literature [4]
● AI in robots [5]
● AI used in drone swarms [6,7]
The above list is not even the tip of the iceberg, considering the contributions
of AI in today’s world. Remember that this list does not feature the detailed list of
engineering explorations brought about by AI.
Evolutionary optimization algorithms are one of the important components of
AI. These optimization techniques are used to find optimal solutions for complex
search problems. Another application of the optimization techniques is to obtain
feasible solutions for problems, where finding a solution itself is the objective.
Such problems have very less number of solutions that are acceptable and thus such
a problem is similar to finding a needle in a haystack.
Evolutionary optimization algorithms are usually based on methodologies that
are derived from natural processes. These methods are very robust and easily adapt

1
National Institute of Technology and Science – Warangal, India
120 Nature-inspired optimization algorithms and soft computing

to different multidimensional complex problems. Each algorithm has its own merits
and demerits.
The initial optimization method, i.e., genetic algorithm was derived from
Darwin’s theory of “Survival of the fittest.” In this theory, the individuals with a
better set of genes are able to survive and reproduce. Hence the new generations,
have the genes of those individuals from the previous generation that were able to
better adapt themselves to the environment. The process repeats and those genes,
which are required to survive are passed on and may be modified as per the
adaptations.
Evolutionary optimization algorithms were introduced as it was observed that
the existing conventional methods applied for solving optimization problems were
not capable to obtain optimal solutions. These algorithms are fast, robust, less
probable to get trapped in local minima, adaptable, etc. A number of evolutionary
optimization algorithms have been introduced over a period of time. Some of these
methods have been modified from the base version while some have been combined
with other optimization methods to form a hybrid method. Different applications of
optimization techniques have been discussed in detail in [8]. Also, step-by-step
evaluation of some of these optimization techniques has been presented in [9].
In this chapter, an exhaustive list of evolutionary optimization algorithms is
presented first. In the second part of the chapter, some of the advanced evolutionary
optimization techniques are discussed.

5.2 List of nature-inspired algorithms

In this section, a table containing a list of more than 200 optimization algorithms is
enlisted. Table 5.1 contains the name of the method, author, year of publication,
and the reference for each optimization technique. However, for some of the old
optimization techniques, there was not a single publication, which introduced the
method but a number of publications. For such optimization techniques, one of
these publications is listed in the Table 5.1 or the most relevant reference has
been cited.

Table 5.1 List of nature-inspired algorithms

Method Author, Year Ref.

African Buffalo Optimization Algorithm (ABO) Odili and Karhar, 2015 [10]
African Vulture Optimization Algorithm Abdallahzadeh et al., 2021 [11]
(AVOA)
Animal Migration Optimization Li et al., 2014 [12]
Ant Colony Optimization (ACO) Marco Dorigo, 1996 [13]
Ant Lion Optimizer (ALO) Mirjalili, 2015 [14]
Aquila Optimizer (AO) Abualigah et al., 2021 [15]
Archimedes Optimization Algorithm Hashim et al., 2021 [16]
Arithmetic Optimization Algorithm (AOA) Laith et al., 2021 [17]
(Continues)
 Advanced optimization by nature-inspired algorithm 121

Table 5.1 (Continued)

Method Author, Year Ref.

Artificial Algae Algorithm (AAA) Uymaz et al., 2015 [18]
Artificial Bee Colony (ABC) Dervis Karaboga, 2005 [19]
Artificial Ecosystem-based Optimization (AEO) Zhao et al., 2020 [20]
Artificial Flora (AF) Optimization Algorithm Cheng et al., 2021 [21]
Artificial Immune Algorithm (AIA) Bagheri et al., 2010 [22]
Backtracking Search Optimization Algorithm Pinar Civicioglu, 2013 [23]
(BSA)
Bacterial Foraging Optimization (BFO) Passino, 2002 [24]
Bald Eagle Search (BES) Optimization Alsattar et al., 2020 [25]
Algorithm
Bat Algorithm (BA) Xin-She Yang, 2010 [26]
Battle Royale Optimization Algorithm Rahkar Farshi, 2021 [27]
Bee Colony Optimization Algorithm (BCO) Teodorovic et al., 2005 [28]
Bee Swarm Optimization Habiba et al., 2005 [29]
Bees Algorithm Pham et al., 2006 [30]
Beetle Swarm Optimization Algorithm Wang and Yang, 2018 [31]
Beluga Whale Optimization (BWO) Changting Zhong, 2022 [32]
Big Bang Big Crunch (BB-BC) Erol and Ersking, 2006 [33]
Biogeography-based Optimization (BBO) Simon, 2008 [34]
Bird Mating Optimizer Askarzadeh, 2014 [35]
Black Hole Algorithm (BH) Abdolreza Hatamlou, 2013 [36]
Black Widow Optimization Algorithm Hayyolalam and Kazem, [37]
2020
Blind Naked Mole Rats Algorithm (BNMR) Mohammad, 2013 [38]
Bonobo Optimizer (BO) Amit and Dilip Kumar, 2019 [39]
Brain Storm Optimization Algorithm (BSO) Yuhui Shi, 2011 [40]
Bumble Bees Mating Optimization (BBMO) Marinakis et al., 2010 [41]
Butterfly Mating Optimization Algorithm Jada et al., 2016 [42]
(BBMO)
Butterfly Optimization Algorithm (BOA) Arora et al., 2019 [43]
Cat Swarm Optimization (CSO) Chu et al., 2006 [44]
Central Force Optimization (CFO) Formato, 2008 [45]
Centroid Mutation-Based Search and Rescue Houssein et al., 2022 [46]
(cmSAR) Optimization Algorithm
Chameleon Swarm Algorithm (CSA) Braik, 2021 [47]
Chaos Optimization Algorithm Jeng et al., 2017 [48]
Chaos Optimization Method (COM) Jiang, 1998 [49]
Charged System Search (CSS) Kaveh and Talatahari, 2010 [50]
Chicken Swarm Optimization Meng et al., 2014 [51]
Chimp Optimization Algorithm (ChOA) Khishe and Mosavi, 2020 [52]
Class Topper Optimization Algorithm Das et al., 2018 [53]
Cognitive Behavior Optimization Algorithm Li et al., 2016 [54]
Collective Decision Optimization Algorithm Zhang et al., 2017 [55]
Colliding Bodies Optimization (CBO) Kaveh and Mahdavi, 2014 [56]
Competition over Resources Mohseni et al., 2014 [57]
Competitive Optimization Algorithm (COOA) Sharafi et al., 2016 [58]
Coot Bird Search Algorithm Mamarzadeh and Keynia, [59]
2021
(Continues)
122 Nature-inspired optimization algorithms and soft computing

Table 5.1 (Continued)

Method Author, Year Ref.

Coral Reef Optimization Algorithm Sanz et al., 2014 [60]
Covariance Matrix Adaptation Evolution Strategy Hansen et al., 2003 [61]
(CMA-ES)
Coyote Optimization Algorithm (COA) Pierezen and Coelho, 2018 [62]
Crisscross Optimization Algorithm Meng et al., 2016 [63]
Crow Search Algorithm (CSA) Askarzadeh, 2016 [64]
Cuckoo Search (CS) Yang and Deb, 2009 [65]
Cuttle Fish Algorithm (CFA) Adel Sabry Eesa et al., 2013 [66]
Deer Hunting Optimization Algorithm Deer Hunting Optimization, [67]
2019
Differential Evolution (DE) Storn and Price, 1997 [68]
Dolphin Echolocation (DE) Kaveh and Farhaundi, 2013 [69]
Dragonfly Algorithm (DA) Mirjalili, 2016 [70]
Dwarf Mongoose Optimization (DMO) Agushaka et al., 2022 [71]
Algorithm
Ebola Optimization Search Algorithm (EOSA) Oyelade et al., 2022 [72]
Egret Swarm Optimization Algorithm Chen et al., 2022 [73]
Electromagnetic Field Optimization (EFO) Hosein et al., 2016 [74]
Elephant Herding Optimization Wang et al., 2015 [75]
Emperor Penguin Optimizer (EPO) Harifi et al., 2019 [76]
Equilibrium Optimizer (EO) Faramarzi et al., 2020 [77]
Firefly Algorithm (FA) Xin-She Yang, 2009 [78]
Fireworks Algorithm (FWA) Tan and Zhu, 2010 [79]
Flower Pollination Algorithm (FPA) Xin-She Yang, 2012 [80]
Forest Optimization Algorithm Ghaemi and Feizi-Derakhshi, [81]
2014
Fruit Fly Optimization Algorithm Xing and Gao, 2014 [82]
Galactic Swarm Optimization Muthiah and Mathew, 2016 [83]
Gases Brownian Motion Optimization (GBMO) Marjan et al., 2013 [84]
Genetic Algorithms (GA) John Holland, 1970s [85,86]
Genetic Programming (GP) John Koza, 1990 [87]
Glowworm Swarm Optimization (GSO) Krishnanad and Ghose, 2009 [88]
Golden Eagles Optimization (GEO) Abdolkarim et al., 2021 [89]
Gradient-Based Optimizer (GBO) Ahmadianfar et al., 2020 [90]
Grass Fibrous Root Optimization Algorithm Akkar and Mahdi, 2017 [91]
Grasshopper Optimization Algorithm (GOA) Mirjalili, 2018 [92]
Gravitational Search Algorithm (GSA) Rashedi et al., 2009 [93]
Greedy Randomized Adaptive Search Procedure Resende et al., 1994 [94,95]
(GRASP)
Grey Wolf Optimizer (GWO) Mirjalili et al., 2014 [96]
Group Teaching Optimization Algorithm Zhang and Jin, 2020 [97]
Harmony Search (HS) Loganathan et al., 2001 [98]
Harris Hawk Optimization (HHO) Heidari et al., 2019 [99]
Heart Heat Transfer Relation-Based Optimization Asef et al., 2021 [100,101]
Algorithm (HOTA)
Heat Transfer Search (HTS) Patel and Savsani, 2015 [102]
Henry Gas Solubility Optimization Hashim et al., 2019 [103]
Honey Badger Algorithm (HBA) Hashim et al., 2022 [104]
(Continues)
 Advanced optimization by nature-inspired algorithm 123

Table 5.1 (Continued)

Method Author, Year Ref.

Horse Herd Optimization Algorithm Miar Naeimi et al., 2021 [105]
Human Learning Optimization Algorithm Wang et al., 2014 [106]
Human Mental Search Mousavirad et al., 2017 [107]
Hunger Games Search (HGS) Yang et al., 2021 [108]
Hurricane-Based Optimization Algorithm Rbouh and El Imrani, 2014 [109]
Imperialist Competitive Algorithm (ICA) Atashpaz-Gargari and Lucas, [110]
2007
Inclined Planes System Optimization (IPO) Mozaffari et al., 2016 [111]
Algorithm
Integrated Optimization Algorithm (IOA) Chen Li et al., 2022 [112]
Interactive Autodidactic School (IAS) Jahangiri et al., 2020 [113]
Invasive Tumor Growth Optimization (ITGO) Tang et al., 2015 [114]
Algorithm
Invasive Weed Optimization Rad and Lucas, 2007 [115]
Ions Motion Algorithm Javidy et al., 2015 [116]
Iterated Local Search (ILS) Katayama and Narihisa, 1999 [117]
Jaya Optimization Algorithm (JOA) Venkata Rao, 2016 [118]
Kinetic Gas Molecule Optimization (KGMO) Moein and Logeswaran, 2014 [119]
Krill Herd Algorithm (KHA) Gandomi and Alavi, 2012 [120]
Lichtenberg Optimization Algorithm Pereira et al., 2021 [121]
Lightning Attachment Procedure Optimization Nematollahi et al., 2019 [122]
Algorithm
Lion Optimization Algorithm (LOA) Yazdani and Jolani, 2016 [123]
Locust Swarm Optimization Algorithm Cuevas et al., 2020 [124]
Magnetic Optimization Algorithm Kushwaha et al., 2018 [125]
Magnetic-Inspired Optimization Algorithm Tayarani and Akbarzadeh, [126]
2014
Manta Ray Foraging Optimization (MRFO) Weiguo Zhao et al., 2020 [127]
Marine Predators Algorithm (MPA) Faramarzi et al., 2020 [128]
Marriage in Honey Bees Optimization (HBO) Abbass, 2001 [129]
Mayfly Optimization Algorithm Zervoudakis and Safarakis, [130]
2020
Mine Blast Algorithm (MBA) Ali Sadollah et al., 2013 [131]
Monarch Butterfly Optimization (MBO) Wang et al., 2015 [132]
Monkey King Evolutionary Algorithm Meng and Pen, 2016 [133]
Monkey Search Algorithm Seref and Akcali, 2002 [134]
Most Valuable Player Algorithm (MVPA) Bouchekara, 2020 [135]
Moth-Flame Optimization Algorithm (MFO) Mirjalili, 2015 [136]
MultiVerse Optimizer (MVO) Azizi et al., 2019 [137]
Optics Inspired Optimization (OIO) Kashan, 2015 [138]
Parliamentary Optimization Algorithm (POA) Borji, 2007 [139]
Particle Swarm Optimization (PSO) Eberhart and Kennedy, 1995 [140]
Pathfinder Algorithm (PFA) Yapici and Cetinkayab, 2019 [141]
Pelican Optimization Algorithm (POA) Trojovsky and Dehghani, [142]
2022
Pigeon-Inspired Optimization Duan and Qiao, 2014 [143]
Polar Bear Optimization Algorithm Polap and Wo’zniak, 2017 [144]
Political Optimizer Qamar Askari et al., 2020 [145]
(Continues)
124 Nature-inspired optimization algorithms and soft computing

Table 5.1 (Continued)

Method Author, Year Ref.

Poor and Rich Optimization Algorithm Moosavi and Bardsiri, 2019 [146]
Puzzle Optimization Algorithm (POA) Zeidabadi and Dehghani, [147]
2022
Quantum Approximate Optimization Algorithm Farhi et al., 2014 [148]
Queuing Search Algorithm (QSA) Zhang et al., 2018 [149]
Raccoon Optimization Algorithm Koohi et al., 2018 [150]
Rain Optimization Algorithm (ROA) Moazzeni and Khamehchi, [151]
2020
Rain-Fall Optimization Algorithm Kaboli et al., 2017 [152]
Rao Optimization Algorithms (ROA) Venkata Rao and Pawar, [153]
2020
Raven Roosting Optimization Algorithm Brabazon et al., 2016 [154]
Ray Optimization (RO) Kaveh Khayatazad, 2012 [155]
Red Deer Algorithm (RDA) Fard and Hajiaghaei- [156]
Keshteli, 2016
Red Fox Optimization Algorithm (RFO) Polap and Wozniak, 2021 [157]
Reducing Variable Trend Search (RVTS) Badar et al., 2014 [158]
Remora Optimization Algorithm Jia et al., 2021 [159]
Reptile Search Algorithm (RSA) Laith Abualigah et al., 2022 [160]
Rooted Tree Optimization Algorithm Labbi et al., 2016 [161]
RUNge Kutta optimizer (RUN) Ahmadianfar et al., 2021 [162]
Salp Swarm Algorithm Mirjalili et al., 2017 [163]
Sandpiper Optimization Algorithm (SOA) Kaur et al., 2020 [164]
Satin Bowerbird Optimizer Moosavi and Bardsiri, 2017 [165]
Scatter Search (SS) Fred Glover, 1977 [166]
Seagull Optimization Algorithm (SOA) Dhiman and Kumar, 2019 [167]
Sea Lion Optimization Algorithm Masadeh et al., 2019 [168]
Search and Rescue Optimization Algorithm Shabani et al., 2020 [169]
Seasons Optimization Algorithm Emami, 2020 [170]
Seeker Optimization Algorithm Dai et al., 2006 [171]
Selfish Herd Optimizer (SHO) Fausto et al., 2017 [172]
Shark Smell Optimization (SSO) Oveis Abdinia et al., 2016 [173]
Shuffled Complex Optimizers (SCO) Duan et al., 1993 [174]
Shuffled Frog-Leaping Algorithm (SFLA) Eusuff et al., 2006 [175]
Simulated Annealing (SA) Kirkpatrick et al., 1983 [176]
Sine Cosine Algorithm (SCA) Seyedali Mirjalili, 2016 [177]
Slime Mold Algorithm (SMA) Li et al., 2020 [178]
Snake Optimizer Hashim and Hussien, 2022 [179]
Social Mimic Optimization Algorithm Bolachian and Baloochian, [180]
2019
Social Spider Optimization (SSO) Cuevas et al., 2013
James and Victor, 2015 [181,182]
Socio Evolution and Learning Optimization Kumar et al., 2018 [183]
Algorithm
Sooty Tern Optimization Algorithm (STOA) Dhiman and Kaur, 2019 [184]
Sparrow Search Algorithm (SSA) Xue and Shen, 2020 [185]
Sperm Swarm Optimization Algorithm Hisham et al., 2018 [186]
Spider Monkey Optimization Algorithm Bansal et al., 2014 [187]
(Continues)
 Advanced optimization by nature-inspired algorithm 125

Table 5.1 (Continued)

Method Author, Year Ref.

Spotted Hyena Optimizer (SHO) Dhiman et al., 2017 [188]
Squirrel Search Algorithm (SSA) Jain, 2019 [189]
States of Matter Search (SMS) Cuevas et al., 2013 [190]
Sunflower Optimization Algorithm (SFO) Gomes et al., 2019 [191]
Supply–Demand-Based Optimization Zhao et al., 2019 [192]
Symbiotic Organism Search (SOS) Cheng and Prayogo, 2014 [193]
Tabu Search (TS) Fred Glover, 1989 [194,195]
Teaching-Learning-Based Optimization (TLBO) Venkata Rao, 2011 [196]
Teamwork Optimization Algorithm Dehghani and Trojovsky, [197]
2021
Thermal Exchange Optimization (TEO) Kaveh and Dadras, 2017 [198]
Transient Search Optimization Qais et al., 2020 [199]
Variable Neighbourhood Search (VNS) Mladenovi’c and Hansen, [200]
1997
Virtual Bee Algorithm Yang, 2005 [201]
Virulence Optimization Algorithm Jaderyan and Khotanlou, [202]
2016
Virus Optimization Algorithm Liang and Cuevas Juarez, [203]
2016
Vortex Search (VS) Algorithm Berat and Tamer, 2015 [204]
War Strategy Optimization Algorithm Tummala et al., 2022 [205]
Water Cycle Algorithm (WCA) Eskanda et al., 2012 [206]
Water Evaporation Optimization (WEO) Kaveh and Bakshpoori, 2016 [207]
Water Wave Optimization (WWO) Zheng, 2015 [208]
Weed Optimization Algorithm Asgari et al., 2016 [209]
Weighted mean of vectors (INFO) Ahmandianfar et al., 2022 [210]
Whale Optimization Algorithm (WOA) Mirjalili and Lewis, 2016 [211]
Wind Driven Optimization Bayraktar et al., 2010 [212]
Wingsuit Flying Search Covic and Lacevic, 2020 [213]
World Cup Optimization Algorithm Shahrezaee, 2017 [214]
Yin-Yang-Pair Optimization Punnathanam and Kotecha, [215]
2016

5.3 Optimization techniques

In this section, some of the recent optimization techniques have been discussed. All
the necessary equations required to realize these optimization techniques are also
presented.

5.3.1 Anarchic society optimization (ASO)

ASO is inspired by the anarchic behavior of society members to improve their
situation. Anarchic behavior means that the behavior of the members does not
follow any controlling rules or principles. In ASO, the members can be fickle
minded, i.e., (i) one whose loyalties and affections can change frequently, (ii)
126 Nature-inspired optimization algorithms and soft computing

adventurous, and (iii) unstable. This kind of behavior of the members can lead to
irrational movements which could lead a member to an inferior position. Such
anarchic movement of members increases as the difference between them increases
or the situation worsens.
The members of the society have the knowledge about the best position
attained by any other member of the society till the current iteration is considered.
The members are also aware of the member who is occupying the current best
position in the society. There are three policies on which a member shall move.
1. The first policy for movement is based on the current position of a member.
This movement is dependent on Fickleness Index (FI). It represents a mem-
ber’s dissatisfaction with its current position in comparison with other mem-
bers. The FI can be obtained through one of the following equations:
f ðXik ðkÞÞ f ðPiðkÞÞ
FIi ðkÞ ¼ 1  ai  ð1  ai Þi
f ðXi ðkÞÞ f ðXi ðkÞÞ

f ðGðkÞÞ f ðPiðkÞÞ
FIi ðkÞ ¼ 1  ai  ð1  ai Þ
f ðXi ðkÞÞ f ðXi ðkÞÞ
where a is a number between 0 and 1; “k” is the current iteration; “I” is the
member of the society; Xi(k) is the position of ith member for kth iteration; Pi
(k) is the personal best position achieved by ith member; G(k) is the global best
position; ik* is the member holding the best position in the society in the kth
iteration; and f() represents the objective function value.
2. In the second policy, the movement of the member is dependent on other
members of the society. The movement under this policy is controlled by
external irregularity index (EI). EI is proposed to be evaluated as:

EIi ðkÞ ¼ 1  ediDðkÞ

where di is a positive number. D(k) is a suitable dispersion measure like
coefficient of variation.
3. In the third and last policy of movement, the past positions are considered. It is
not necessary that the best position in the past shall be considered. In the
movement, internal irregularity index (II) is utilized.
These movement policies are finally combined through the elitism or sequential
combination rule.

5.3.2 Antlion optimizer (ALO)

Antlions are also called doodlebugs. The method of foraging used by antlion larvae
involves enticing its prey into a cone-shaped pit. The larvae position itself at the
bottom of the circular pit and wait for the prey i.e., ant. The cone-shaped pit is so
designed that the prey falls into the pit when it crosses the boundary. As the prey
falls into the pit, the antlion grabs it and consumes the ant. In case the prey tries to
escape, the antlion throws sand on the insect which drags it down to the bottom of
 Advanced optimization by nature-inspired algorithm 127

the pit. The pit size is observed to be dependent on the amount of hunger and the
moon size.
Ants, i.e., the prey, move in a stochastic way in search of food. Its movement is
modeled as:
X ðtÞ ¼ ½0; csð2rðt1Þ  1Þ; csð2rðt2Þ  1Þ; ::::csð2rðtnÞ  1
where cs is the cumulative sum; r(t) is evaluated as follows:
rðtÞ ¼ 1 if rand > 0:5
0 if rand  0:5
“n” is the maximum iteration count and “t” is the random walk step.
A pattern of such a random walk is represented through a graph in the refer-
ence publication.
The positions of the ants are saved in a matrix called Mant, which is given as:
2 3
A1;1 A1;2 . . . . . . A1;d
6 7
6 A2;1 A2;2 . . . . . . A2;d 7
6 7
6 . .. .. 7
6 .. .   . 7
4 5
An;1 An;2       An;d
where “d” is the number of dimensions; “n” is the number of ants; and Aij repre-
sents a value for ith and for jth dimension.
The ant in the ALO is similar to wolves in Grey Wolf Optimizer or a particle
in PSO.
There is another matrix MOA which stores the objective function value of each
ant. This matrix will be a column matrix of size “n”.
Similar to Mant, another matrix is formed which is termed as Mantlion. It is used
to store the position of each antlion. Also, corresponding to MOA, another matrix is
used for storing the objective function value of antlions and is given by MOAL.
The movement of the ants should be within the search space. Also, in case the
ant gets fitter than the antlion then it means that it has been caught by that antlion.
The random walk is represented as:
ðXit  ai Þ  ðdi  cti Þ
Xit ¼ þ ci
ðdit  ai Þ
where “i” represents the variable/dimension; “t” is the iteration; “a” is the mini-
mum random walk for the variable “i”; “c” and “d” are the minimum and maximum
values of ith variable in the tth iteration.
The effect of antlion traps on the random walk of ants is given as:
cti ¼ Antliontj þ ct
dit ¼ Antliontj þ d t

where Antlionjt gives the position of the jth antlion in the tth iteration.
128 Nature-inspired optimization algorithms and soft computing

The sliding of ants toward an antlion is modeled as:

ct
ct ¼
I
t
d
dt ¼
I
where I represents a ratio.
The catching of an ant by the antlion is given by the next equation:
Antliontj ¼ Antit if f ðAntit Þ > f ðAntliontj Þ

The ant is supposed to walk randomly around a particular antlion, selected

through the roulette wheel process, and also toward the best antlion. This is given as:
RtA þ RtE
Antit ¼
2
where “R” is the random walk; “t” is the iteration; “A” is the antlion selected
through the roulette wheel; “E” is the global best antlion.

5.3.3 Cat swarm optimization

As the name of the optimization is, this algorithm is based on cat behavior. The cat
behavior is classified into (i) seeking and (ii) tracking mode. In cat swarm opti-
mization (CSO), the cats play the role of particles in PSO or chromosomes in GA.
Seeking mode represents the lazy mode of cat behavior in which it rests or
looks around. This mode introduces four factors:
1. Seeking Memory Pool (SMP): It helps the cat pick a point from the memory.
2. Selected Dimension (SRD): It represents a mutative ratio.
3. Counts of Dimension to Change (CDC): This factor represents the number of
dimensions that shall be changed.
4. Self-Position Considering (SPC): This is a Boolean variable that represents
whether the cat will move or not.
In CSO, initially, a number of copies are created from the current position of
the catk. Then based on the factors presented above the variations are introduced in
them. The objective function value is evaluated for each new candidate position.
Any candidate can be selected based on the probability which is obtained as:
jFSi  FSb j
Pi ¼
FSmax  FSmin
where FSb = FSmax for the minimization problem and FSb = FSmin for the max-
imization problem.
The new position is randomly picked and catk has a new position.
In the tracing mode of search, the cat is able to move in all dimensions. The
equations applied for the movement of the cat is quite similar to the equations
applied in PSO except that the individual best term is not considered.
 Advanced optimization by nature-inspired algorithm 129

5.3.4 Crow search algorithm

Crow search algorithm (CSA) exploits the behavior of crows to develop a meta-
heuristics technique. The distinct features of crows are:
1. They live in flocks.
2. They can memorize their hiding places.
3. Thievery is prevalent in the flock.
4. They try to protect their catch.
In CSA, the movement of a crow “I” is influenced by the movement of another
crow “j,” toward its individual best, which it might have achieved in previous
iterations. The effect of global best is neglected in the search process, while the
movement of a crow is dependent on the individual best of another crow. The
movement of a crow is given as:
 i;it
x þ ri  fli;it  ðmj;it  xi;it Þ rj  APj;it
xi;itþ1 ¼
a random position otherwise

where “I” is a particular crow in the flock; “it” is the iteration; “x” represents the
position of a crow in the search space; “r” is a random number between 0 and 1;
“fl” signifies the flight length. Indirectly, it is a factor which helps control the
search process and leads it into an exploration or an exploitation phase; “m”
represents the individual best position; “AP” gives awareness probability of
a crow.
In CSA, it is required to store the individual best positions of each crow and
update them in every iteration.

5.3.5 Cuckoo search

Cuckoos have a special strategy for reproduction. They remove the eggs of other
birds and instead put their own eggs in their nest. The cuckoo’s chick outgrows the
chicks of the host bird and therefore is able to survive. The cuckoo search is based
on three rules:
● One egg per cuckoo is laid in a random nest
● Nests with better quality eggs will be considered in the search process
● In case the host is able to discover the cuckoo’s egg, it can either discard the
egg or build a new nest.
Levy flight which deals with the flight behavior of many animals is also
included in this search process.
The equation used in the cuckoo search is:
ðtþ1Þ ðtÞ
xi ¼ xi þ a LevyðlÞ

where “t” is the iteration; “I” is the cuckoo number; a is a constant defined as a > 0
or preferably taken as a = 1; represents entry-wise multiplication.
130 Nature-inspired optimization algorithms and soft computing

5.3.6 Mine blast algorithm

Mine blast algorithm is based on the process of mine bomb explosion. The shrapnel
from the bomb is spread in all directions. This shrapnel then triggers other mine
bombs with which they collide. The process represents a minefield being cleared.
The aim is to find the mine with the highest explosive capacity and this mine is the
optimal solution being searched through the process.
The casualties caused due to a particular bomb is proportional to the objec-
tive function value of that mine. Also, a high explosion in a particular area
indicates the presence of other mines with high capacity in proximity. In mine
blast algorithm, the process starts with one mine which then expands to more
number of mines.
The equation that governs the mine blast algorithm is given as:
0 v ffiffiffiffiffiffiffiffiffiffi
u f 1
u m
f
Xnþ1 f
¼ Xeðnþ1Þ þ exp@t fnþ1 AXnf ; n ¼ 0; 1; 2; 3; ::::
dnþ1

where “X” represents a solution; “f” represents first shot points; “e” represents the
explosion point; “m” represents the direction of shrapnel; “d” represents the dis-
tance that this shrapnel will cover.
The position of the exploding mine is defined as:
f f
Xeðnþ1Þ ¼ dnþ1  cos ðqÞ; n ¼ 0; 1; 2; :::

where rand is a random number; q is angle at which the shrapnel will travel and it is
given as 360/Ns, where Ns is the number of shrapnel pieces.
The distance and the direction of the shrapnel are given as:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
f f
dnþ1 ¼ ðXnþ1  Xnf Þ2 þ ðFnþ1 f
 Fnf Þ2 ; n ¼ 0; 1; 2; 3; :::
f
Fnþ1  Fnf
mfnþ1 ¼ f
; n ¼ 0; 1; 2; 3; :::
Xnþ1  Xnf

where “F” is the objective function value.

To promote exploration in the search process, two parameters are considered:
(i) exploration factor (m) and (ii) iteration number index (k). The relationship
between these two parameters decides if the exploration process is to be initiated.
The equations used for exploration are:
f
dnþ1 ¼ dnf  ðj randn jÞ2 ; n ¼ 0; 1; 2; :::
f f
Xeðnþ1Þ ¼ dnþ1  cos ðqÞ; n ¼ 0; 1; 2; :::

where randn is a random number which is normally distributed as in other algo-

rithms, the initial phase of the search process concentrates on exploration, and the
 Advanced optimization by nature-inspired algorithm 131

later part is directed toward exploitation. Thus, the distance of shrapnel is con-
trolled as:
f
dn1
dnf ¼ ; n ¼ 1; 2; 3; :::
expðk=aÞ
where a is the reduction constant. The value of alpha should be such that the
shrapnel distance should reduce with every iteration. This reduction should lead to
no distance traveled by the shrapnel in the last iteration.

5.3.7 Water cycle algorithm

A complete water cycle includes the process of evaporation, condensation, tran-
spiration, precipitation, etc. Water cycle algorithm (WCA) is based on the natural
process of the formation of streams/rivers as water flows down from hills toward
the sea as a part of the water cycle. The small rivulets combine to form a big river
as they approach the sea. The population in WCA is made up of raindrops, and the
best raindrop is considered as the sea.
The total population is made up of sea, rivers, and raindrops. There can be a single
sea in the population, which is represented by the raindrop having the best solution. Nsr
is used to represent a subset of a population consisting of rivers and sea. NSn is used to
represent the number of streams that flow into the rivers or sea. It is given as:
  !
 Cost 
 n 
NSn ¼ round PNsr   NRaindrops
 i¼1 Costi 

where Costn is the objective function value of nth raindrops; NRaindrops is given as
NPOP – Nsr, where Npop is the population size.
Streams can combine to form a river or can directly flow into the sea. The
distance of travel by a stream can be randomly chosen.
The new position of the rivers and streams is represented as:
iþ1
XStream ¼ XStream
i
þ rand  C  ðXRiver
i
 XStream
i
Þ
iþ1
XRiver ¼ XRiver
i
þ rand  C  ðXSea
i
 XRiver
i
Þ
where rand is a random number between 0 and 1; “X” is the position.
In case the objective function value of a stream is better than the connecting
river then their positions are interchanged. A similar operation can happen between
a sea and a river.
In case, the distance between the river and sea is smaller than dmax, then the
process of evaporation happens after which the raining process will start which is
similar to the generation of random solutions within the search space. The value of
dmax is proposed to increase the search process near the sea. The value of dmax can
also be made linearly dependent on iterations as:
di
iþ1
dmax ¼ dmax
i
 max
maxiter
132 Nature-inspired optimization algorithms and soft computing

The raining process can be used to generate random solutions near the sea for
better exploitation in that region.

5.4 Conclusion
The chapter is used to present a vast collection of nature-inspired algorithms along
with their references. This information should be very useful for researchers who
can access the details of approximately 200 algorithms in a single place.
The chapter also introduces seven recently proposed evolutionary algorithms
along with the necessary equations of these methods.

References

[1] Trends M. AI Writes an Academic Paper About Itself and Researchers

Try to Publish It!; 2022. https://www.analyticsinsight.net/ai-writes-an-
academic-paper-about-itself-and-researchers-try-to-publish-it/.
[2] Belton P. Would You Let a Robot Lawyer Defend You?; 2021. https://
www.bbc.com/news/ business-58158820.
[3] Basu K, Sinha R, Ong A, et al. Artificial intelligence: How is it changing
medical sciences and its future? Indian Journal of Dermatology. 2020;
65(5):365.
[4] Hart M. Google’s New AI Helps You Write Poetry Like POE; 2020. https://
nerdist. com/article/google-ai-writes-poetry-like-legendary-poets/#: :text=
Google’s%20new%20AI%20tool%2C%20Verse,reading%20the%20poets’
%20respective%20oeuvres.
[5] Moravec H. Rise of the robots – the future of artificial intelligence. In:
Scientific American. Springer; 2009. p. 23.
[6] Awasthi S, Balusamy B, and Porkodi V. Artificial intelligence supervised
swarm UAVs for reconnaissance. In: International Conference on Recent
Developments in Science, Engineering and Technology. Springer; 2019.
p. 375–388.
[7] Lomonaco V, Trotta A, Ziosi M, et al. Intelligent drone swarm for search
and rescue operations at sea, 2018. arXiv preprint arXiv:181105291.
[8] Badar AQH. In: Mercangöz BA, editor. Different Applications of PSO.
Cham: Springer International Publishing; 2021. p. 191–208. https://doi.org/
10.1007/978-3-030-70281-6 11.
[9] Badar AQ. Evolutionary Optimization Algorithms. CRC Press; 2021.
[10] Odili JB and Kahar MNM. Numerical function optimization solutions using
the African buffalo optimization algorithm (ABO). British Journal of
Mathematics & Computer Science. 2015;10(1):1–12.
[11] Abdollahzadeh B, Gharehchopogh FS, and Mirjalili S. African vultures
optimization algorithm: a new nature-inspired metaheuristic algorithm for
global optimization problems. Computers & Industrial Engineering. 2021;
158:107408.
 Advanced optimization by nature-inspired algorithm 133

[12] Li X, Zhang J, and Yin M. Animal migration optimization: an optimization

algorithm inspired by animal migration behavior. Neural Computing and
Applications. 2014;24(7):1867–1877.
[13] Dorigo M, Birattari M, and Stutzle T. Ant colony optimization. IEEE
Computational Intelligence Magazine. 2006;1(4):28–39.
[14] Mirjalili S. The ant lion optimizer. Advances in Engineering Software.
2015;83:80–98.
[15] Abualigah L, Yousri D, Abd Elaziz M, et al. Aquila optimizer: a novel
meta-heuristic optimization algorithm. Computers & Industrial
Engineering. 2021;157:107250.
[16] Hashim FA, Hussain K, Houssein EH, et al. Archimedes optimization
algorithm: a new metaheuristic algorithm for solving optimization pro-
blems. Applied Intelligence. 2021;51(3):1531–1551.
[17] Abualigah L, Diabat A, Mirjalili S, et al. The arithmetic optimization
algorithm. Computer Methods in Applied Mechanics and Engineering.
2021;376:113609.
[18] Uymaz SA, Tezel G, and Yel E. Artificial algae algorithm (AAA) for
nonlinear global optimization. Applied Soft Computing. 2015;31:153–171.
[19] Karaboga D and Basturk B.. An idea based on honey bee swarm for
numerical optimization. Technical Report – tr06, Erciyes University,
Engineering Faculty; 2005.
[20] Zhao W, Wang L, and Zhang Z. Artificial ecosystem-based optimization: a
novel nature-inspired meta-heuristic algorithm. Neural Computing and
Applications. 2020;32(13):9383–9425.
[21] Cheng L, Wu Xh, and Wang Y. Artificial flora (AF) optimization algo-
rithm. Applied Sciences. 2018;8(3):329.
[22] Bagheri A, Zandieh M, Mahdavi I, et al. An artificial immune algorithm for
the flexible job-shop scheduling problem. Future Generation Computer
Systems. 2010;26(4):533–541.
[23] Civicioglu P. Backtracking search optimization algorithm for numerical
optimization problems. Applied Mathematics and Computation. 2013;219
(15):8121–8144.
[24] Passino KM. Biomimicry of bacterial foraging for distributed optimization
and control. IEEE Control Systems Magazine. 2002;22(3):52–67.
[25] Alsattar HA, Zaidan A, and Zaidan B. Novel meta-heuristic bald eagle
search optimisation algorithm. Artificial Intelligence Review. 2020;53
(3):2237–2264.
[26] Yang XS. A new metaheuristic bat-inspired algorithm. In: Nature Inspired
Cooperative Strategies for Optimization (NICSO2010). Springer; 2010.
p. 65–74.
[27] Rahkar Farshi T. Battle Royale optimization algorithm. Neural Computing
and Applications. 2021;33(4):1139–1157.
[28] Teodorovic D and Dell’Orco M. Bee colony optimization – a cooperative
learning approach to complex transportation problems. Advanced OR and
AI Methods in Transportation. 2005;51:60.
134 Nature-inspired optimization algorithms and soft computing

[29] Drias H, Sadeg S, and Yahi S. Cooperative bees swarm for solving the
maximum weighted satisfiability problem. In: International Work-
Conference on Artificial Neural Networks. Springer; 2005. p. 318–325.
[30] Pham DT, Ghanbarzadeh A, Ko̧c E, et al. The bees algorithm—a novel tool
for complex optimisation problems. In: Intelligent Production Machines
and Systems. Elsevier; 2006. p. 454–459.
[31] Wang T, Yang L. Beetle swarm optimization algorithm: theory and appli-
cation; 2018. arXiv preprint arXiv:180800206.
[32] Zhong C, Li G, and Meng Z. Beluga whale optimization: a novel nature-
inspired metaheuristic algorithm. Knowledge-Based Systems. 2022;251:
109215.
[33] Erol OK and Eksin I. A new optimization method: big bang–big crunch.
Advances in Engineering Software. 2006;37(2):106–111.
[34] Simon D. Biogeography-based optimization. IEEE Transactions on
Evolutionary Computation. 2008;12(6):702–713.
[35] Askarzadeh A. Bird mating optimizer: an optimization algorithm inspired
by bird mating strategies. Communications in Nonlinear Science and
Numerical Simulation. 2014;19(4):1213–1228.
[36] Hatamlou A. Black hole: a new heuristic optimization approach for data
clustering. Information Sciences. 2013; 222:175–184.
[37] Hayyolalam V and Kazem AAP. Black widow optimization algorithm: a
novel meta-heuristic approach for solving engineering optimization pro-
blems. Engineering Applications of Artificial Intelligence. 2020;87:
103249.
[38] Taherdangkoo M, Shirzadi MH, Yazdi M, et al. A robust clustering method
based on blind, naked mole-rats (BNMR) algorithm. Swarm and
Evolutionary Computation. 2013;10:1–11.
[39] Das AK and Pratihar DK. A new bonobo optimizer (BO) for real-parameter
optimization. In: 2019 IEEE Region 10 Symposium (TENSYMP). IEEE;
2019. p. 108–113.
[40] Shi Y. Brain storm optimization algorithm. In: International Conference in
Swarm Intelligence. Springer; 2011. p. 303–309.
[41] Marinakis Y, Marinaki M, and Matsatsinis N. A bumble bees mating
optimization algorithm for global unconstrained optimization problems. In:
Nature Inspired Cooperative Strategies for Optimization (NICSO2010).
Springer; 2010.p. 305–318.
[42] Jada C, Vadathya AK, Shaik A, et al. Butterfly mating optimization. In:
Intelligent Systems Technologies and Applications. Springer; 2016. p. 3–15.
[43] Arora S and Singh S. Butterfly optimization algorithm: a novel approach
for global optimization. Soft Computing. 2019;23(3):715–734.
[44] Chu SC, Tsai PW, and Pan JS. Cat swarm optimization. In: Pacific
Rim International Conference on Artificial Intelligence. Springer; 2006.
p. 854–858.
[45] Formato RA. Central force optimization: a new nature inspired computa-
tional framework for multidimensional search and optimization. In: Nature
 Advanced optimization by nature-inspired algorithm 135

Inspired Cooperative Strategies for Optimization (NICSO 2007). Springer;

2008. p. 221–238.
[46] Houssein EH, Saber E, Ali AA, et al. Centroid mutation-based Search and
Rescue optimization algorithm for feature selection and classification.
Expert Systems with Applications. 2022;191:116235.
[47] Braik MS. Chameleon Swarm Algorithm: a bio-inspired optimizer for
solving engineering design problems. Expert Systems with Applications.
2021;174:114685.
[48] Feng J, Zhang J, Zhu X, et al. A novel chaos optimization algorithm.
Multimedia Tools and Applications. 2017;76(16):17405–17436.
[49] Jiang BLW. Optimizing complex functions by chaos search. Cybernetics &
Systems. 1998;29(4):409–419.
[50] Kaveh A and Talatahari S. A novel heuristic optimization method: charged
system search. Acta Mechanica. 2010;213(3):267–289.
[51] Meng X, Liu Y, Gao X, et al. A new bio-inspired algorithm: chicken swarm
optimization. In: International Conference in Swarm Intelligence. Springer;
2014. p. 86–94.
[52] Khishe M and Mosavi MR. Chimp optimization algorithm. Expert Systems
with Applications. 2020;149:113338.
[53] Das P, Das DK, and Dey S. A new class topper optimization algorithm with
an application to data clustering. IEEE Transactions on Emerging Topics in
Computing. 2018;8(4):948–959.
[54] Li M, Zhao H, Weng X, et al. Cognitive behavior optimization algorithm for
solving optimization problems. Applied Soft Computing. 2016;39:199–222.
[55] Zhang Q, Wang R, Yang J, et al. Collective decision optimization algo-
rithm: a new heuristic optimization method. Neurocomputing. 2017;221:
123–137.
[56] Kaveh A and Mahdavi VR. Colliding bodies optimization: a novel meta-
heuristic method. Computers & Structures. 2014;139:18–27.
[57] Mohseni S, Gholami R, Zarei N, et al. Competition over resources: a new
optimization algorithm based on animals behavioral ecology. In: 2014
International Conference on Intelligent Networking and Collaborative
Systems. IEEE; 2014. p. 311–315.
[58] Sharafi Y, Khanesar MA, and Teshnehlab M. COOA: competitive optimi-
zation algorithm. Swarm and Evolutionary Computation. 2016;30:39–63.
[59] Memarzadeh G and Keynia F. A new optimal energy storage system model
for wind power producers based on long short term memory and Coot Bird
Search Algorithm. Journal of Energy Storage. 2021;44:103401.
[60] Salcedo-Sanz S, Del Ser J, Landa-Torres I, et al. The coral reefs optimi-
zation algorithm: a novel metaheuristic for efficiently solving optimization
problems. The Scientific World Journal. 2014;2014:739768.
[61] Hansen N, Müller SD, and Koumoutsakos P. Reducing the time complexity
of the derandomized evolution strategy with covariance matrix adaptation
(CMA-ES). Evolutionary Computation. 2003;11(1):1–18.
136 Nature-inspired optimization algorithms and soft computing

[62] Pierezan J and Coelho LDS. Coyote optimization algorithm: a new meta-
heuristic for global optimization problems. In: 2018 IEEE Congress on
Evolutionary Computation (CEC). IEEE; 2018. p. 1–8.
[63] Meng A, Ge J, Yin H, et al. Wind speed forecasting based on wavelet
packet decomposition and artificial neural networks trained by crisscross
optimization algorithm. Energy Conversion and Management. 2016;
114:75–88.
[64] Askarzadeh A. A novel metaheuristic method for solving constrained
engineering optimization problems: crow search algorithm. Computers &
Structures. 2016;169:1–12.
[65] Yang XS and Deb S. Cuckoo search via Lévy flights. In: 2009 World
Congress on Nature & Biologically Inspired Computing (NaBIC). IEEE;
2009. p. 210–214.
[66] Eesa AS, Brifcani AMA, and Orman Z. Cuttlefish algorithm – a novel bio-
inspired optimization algorithm. International Journal of Scientific &
Engineering Research. 2013;4(9):1978–1986.
[67] Brammya G, Praveena S, Ninu Preetha N, et al. Deer hunting optimization
algorithm: a new nature-inspired meta-heuristic paradigm. The Computer
Journal. 2019. https://doi.org/10.1093/comjnl/bxy133.
[68] Storn R and Price K. Differential evolution – a simple and efficient heur-
istic for global optimization over continuous spaces. Journal of Global
Optimization. 1997;11(4):341–359.
[69] Kaveh A and Farhoudi N. A new optimization method: Dolphin echolo-
cation. Advances in Engineering Software. 2013;59:53–70.
[70] Mirjalili S. Dragonfly algorithm: a new meta-heuristic optimization tech-
nique for solving single-objective, discrete, and multi-objective problems.
Neural Computing and Applications. 2016;27(4):1053–1073.
[71] Agushaka JO, Ezugwu AE, and Abualigah L. Dwarf mongoose optimiza-
tion algorithm. Computer Methods in Applied Mechanics and Engineering.
2022;391:114570.
[72] Oyelade ON, Ezugwu AES, Mohamed TI, et al. Ebola optimization search
algorithm: a new nature-inspired metaheuristic optimization algorithm.
IEEE Access. 2022;10:16150–16177.
[73] Chen Z, Francis A, Li S, et al. Egret swarm optimization algorithm:
an evolutionary computation approach for model free optimization.
Biomimetics. 2022;7(4):144.
[74] Abedinpourshotorban H, Shamsuddin SM, Beheshti Z, et al.
Electromagnetic field optimization: a physics-inspired metaheuristic opti-
mization algorithm. Swarm and Evolutionary Computation. 2016;26:8–22.
[75] Wang GG, Deb S, and Coelho LdS. Elephant herding optimization. In:
2015 3rd International Symposium on Computational and Business
Intelligence (ISCBI). IEEE; 2015. p. 1–5.
[76] Harifi S, Khalilian M, Mohammadzadeh J, et al. Emperor Penguins
Colony: a new metaheuristic algorithm for optimization. Evolutionary
Intelligence. 2019;12(2):211–226.
 Advanced optimization by nature-inspired algorithm 137

[77] Faramarzi A, Heidarinejad M, Stephens B, et al. Equilibrium optimizer: a

novel optimization algorithm. Knowledge-Based Systems. 2020;191:105190.
[78] Yang XS. Firefly algorithms for multimodal optimization. In: International
Symposium on Stochastic Algorithms. Springer; 2009. p. 169–178.
[79] Tan Y and Zhu Y. Fireworks algorithm for optimization. In: International
Conference in Swarm Intelligence. Springer; 2010. p. 355–364.
[80] Yang XS. Flower pollination algorithm for global optimization. In:
International Conference on Unconventional Computing and Natural
Computation. Springer; 2012. p. 240–249.
[81] Ghaemi M and Feizi-Derakhshi MR. Forest optimization algorithm. Expert
Systems with Applications. 2014;41(15):6676–6687.
[82] Xing B and Gao WJ. Fruit fly optimization algorithm. In: Innovative
Computational Intelligence: a Rough Guide to 134 Clever Algorithms.
Springer; 2014. p. 167–170.
[83] Muthiah-Nakarajan V and Noel MM. Galactic swarm optimization: a new
global optimization metaheuristic inspired by galactic motion. Applied Soft
Computing. 2016;38:771–787.
[84] Abdechiri M, Meybodi MR, and Bahrami H. Gases Brownian motion
optimization: an algorithm for optimization (GBMO). Applied Soft
Computing. 2013;13(5):2932–2946.
[85] Holland JH. Robust algorithms for adaptation set in a general formal fra-
mework. In: 1970 IEEE Symposium on Adaptive Processes (9th) Decision
and Control; 1970. p. 175–175.
[86] Goldberg DE and Holland JH. Genetic algorithms and machine learning.
In: Machine Learning, Springer; 1988.
[87] Koza JR. Genetic Programming: A Paradigm for Genetically Breeding
Populations of Computer Programs to Solve Problems. vol. 34. Stanford
University, Department of Computer Science Stanford, CA; 1990.
[88] Krishnanand K and Ghose D. Glowworm swarm optimization for simul-
taneous capture of multiple local optima of multi- modal functions. Swarm
Intelligence. 2009;3(2):87–124.
[89] Mohammadi-Balani A, Nayeri MD, Azar A, et al. Golden eagle optimizer:
a nature-inspired metaheuristic algorithm. Computers & Industrial
Engineering. 2021;152:107050.
[90] Ahmadianfar I, Bozorg-Haddad O, and Chu X. Gradient-based optimizer: a
new metaheuristic optimization algorithm. Information Sciences.
2020;540:131–159.
[91] Akkar HA and Mahdi FR. Grass fibrous root optimization algorithm.
International Journal of Intelligent Systems and Applications. 2017;
11(6):15.
[92] Mirjalili SZ, Mirjalili S, Saremi S, et al. Grasshopper optimization algo-
rithm for multi-objective optimization problems. Applied Intelligence.
2018;48(4):805–820.
[93] Rashedi E, Nezamabadi-Pour H, and Saryazdi S. GSA: a gravitational
search algorithm. Information Sciences. 2009;179(13):2232–2248.
138 Nature-inspired optimization algorithms and soft computing

[94] Li Y, Pardalos PM, and Resende MG. A greedy randomized adaptive

search procedure for the quadratic assignment. In: Quadratic Assignment
and Related Problems: DIMACS Workshop, May 20–21, 1993. vol. 16.
American Mathematical Soc.; 1994. p. 237.
[95] Feo TA, Resende MG, and Smith SH. A greedy randomized adaptive
search procedure for maximum independent set. Operations Research.
1994;42(5):860–878.
[96] Mirjalili S, Mirjalili SM, and Lewis A. Grey wolf optimizer. Advances in
Engineering Software. 2014;69:46–61.
[97] Zhang Y and Jin Z. Group teaching optimization algorithm: a novel
metaheuristic method for solving global optimization problems. Expert
Systems with Applications. 2020;148:113246.
[98] Geem ZW, Kim JH, and Loganathan GV. A new heuristic optimization
algorithm: harmony search. Simulation. 2001;76(2):60–68.
[99] Heidari AA, Mirjalili S, Faris H, et al. Harris hawks optimization: algorithm
and applications. Future Generation Computer Systems. 2019;97:849–872.
[100] Hatamlou A. Heart: a novel optimization algorithm for cluster analysis.
Progress in Artificial Intelligence. 2014;2(2):167–173.
[101] Asef F, Majidnezhad V, Feizi-Derakhshi MR, et al. Heat transfer relation-based
optimization algorithm (HTOA). Soft Computing. 2021;25(13):8129–8158.
[102] Patel VK and Savsani VJ. Heat transfer search (HTS): a novel optimization
algorithm. Information Sciences. 2015;324:217–246.
[103] Hashim FA, Houssein EH, Mabrouk MS, et al. Henry gas solubility opti-
mization: a novel physics-based algorithm. Future Generation Computer
Systems. 2019;101:646–667.
[104] Hashim FA, Houssein EH, Hussain K, et al. Honey Badger algorithm: new
metaheuristic algorithm for solving optimization problems. Mathematics
and Computers in Simulation. 2022;192:84–110.
[105] MiarNaeimi F, Azizyan G, and Rashki M. Horse herd optimization algo-
rithm: a nature-inspired algorithm for high-dimensional optimization pro-
blems. Knowledge-Based Systems. 2021;213:106711.
[106] Wang L, Ni H, Yang R, et al. A simple human learning optimization
algorithm. In: Computational Intelligence, Networked Systems and Their
Applications. Springer; 2014, p. 56–65.
[107] Mousavirad SJ and Ebrahimpour-Komleh H. Human mental search: a
new population-based metaheuristic optimization algorithm. Applied
Intelligence. 2017;47(3):850–887.
[108] Yang Y, Chen H, Heidari AA, et al. Hunger games search: visions, con-
ception, implementation, deep analysis, perspectives, and towards perfor-
mance shifts. Expert Systems with Applications. 2021;177:114864.
[109] Rbouh I and El Imrani AA. Hurricane-based optimization algorithm.
AASRI Procedia. 2014;6:26–33.
[110] Khabbazi A, Atashpaz-Gargari E, and Lucas C. Imperialist competitive
algorithm for minimum bit error rate beam- forming. International Journal
of Bio-Inspired Computation. 2009;1(1–2):125–133.
 Advanced optimization by nature-inspired algorithm 139

[111] Mozaffari MH, Abdy H, and Zahiri SH. IPO: an inclined planes system
optimization algorithm. Computing and Informatics. 2016;35(1):222–240.
[112] Li C, Chen G, Liang G, et al. Integrated optimization algorithm: a meta-
heuristic approach for complicated optimization. Information Sciences.
2022;586:424–449.
[113] Jahangiri M, Hadianfard MA, Najafgholipour MA, et al. Interactive auto-
didactic school: a new metaheuristic optimization algorithm for solving
mathematical and structural design optimization problems. Computers &
Structures. 2020;235:106268.
[114] Tang D, Dong S, Jiang Y, et al. ITGO: invasive tumor growth optimization
algorithm. Applied Soft Computing. 2015;36:670–698.
[115] Rad HS and Lucas C. A recommender system based on invasive weed
optimization algorithm. In: 2007 IEEE Congress on Evolutionary
Computation. IEEE; 2007. p. 4297–4304.
[116] Javidy B, Hatamlou A, and Mirjalili S. Ions motion algorithm for solving
optimization problems. Applied Soft Computing. 2015;32:72–79.
[117] Katayama K and Narihisa H. Iterated local search approach using genetic
transformation to the traveling salesman problem. In: Proceedings of the
1st Annual Conference on Genetic and Evolutionary Computation, vol. 1;
1999. p. 321–328.
[118] Rao R. Jaya: a simple and new optimization algorithm for solving con-
strained and unconstrained optimization problems. International Journal of
Industrial Engineering Computations. 2016;7(1):19–34.
[119] Moein S and Logeswaran R. KGMO: a swarm optimization algorithm based on
the kinetic energy of gas molecules. Information Sciences. 2014;275:127–144.
[120] Gandomi AH and Alavi AH. Krill herd: a new bio-inspired optimization
algorithm. Communications in Nonlinear Science and Numerical
Simulation. 2012;17(12):4831–4845.
[121] Pereira JLJ, Francisco MB, da Cunha Jr SS, et al. A powerful Lichtenberg
Optimization Algorithm: a damage identification case study. Engineering
Applications of Artificial Intelligence. 2021;97:104055.
[122] Nematollahi AF, Rahiminejad A, and Vahidi B. A novel multi-objective
optimization algorithm based on Lightning Attachment Procedure
Optimization algorithm. Applied Soft Computing. 2019;75:404–427.
[123] Yazdani M and Jolai F. Lion optimization algorithm (LOA): a nature-
inspired metaheuristic algorithm. Journal of Computational Design and
Engineering. 2016;3(1):24–36.
[124] Cuevas E, Fausto F, and González A. The locust swarm optimization
algorithm. In: New Advancements in Swarm Algorithms: Operators and
Applications. Springer; 2020. p. 139–159.
[125] Kushwaha N, Pant M, Kant S, et al. Magnetic optimization algorithm for
data clustering. Pattern Recognition Letters. 2018;115:59–65.
[126] Tayarani-N MH and Akbarzadeh-T MR. Magnetic-inspired optimization
algorithms: operators and structures. Swarm and Evolutionary
Computation. 2014;19:82–101.
140 Nature-inspired optimization algorithms and soft computing

[127] Zhao W, Zhang Z, and Wang L. Manta ray foraging optimization: an

effective bio-inspired optimizer for engineering applications. Engineering
Applications of Artificial Intelligence. 2020;87:103300. https://www. science
direct.com/science/article/pii/S0952197619302593.
[128] Faramarzi A, Heidarinejad M, Mirjalili S, et al. Marine Predators Algorithm:
a nature-inspired metaheuristic. Expert Systems with Applications. 2020;152:
113377.
[129] Abbass HA. MBO: marriage in honey bees optimization – a haplometrosis
polygynous swarming approach. In: Proceedings of the 2001 Congress on
Evolutionary Computation (IEEE Cat. No. 01TH8546). vol. 1. IEEE; 2001.
p. 207–214.
[130] Zervoudakis K and Tsafarakis S. A mayfly optimization algorithm.
Computers & Industrial Engineering. 2020;145:106559.
[131] Sadollah A, Bahreininejad A, Eskandar H, et al. Mine blast algorithm: a
new population-based algorithm for solving constrained engineering opti-
mization problems. Applied Soft Computing. 2013;13(5):2592–2612.
[132] Wang GG, Zhao X, and Deb S. A novel monarch butterfly optimization with
greedy strategy and self-adaptive. In: 2015 Second International Conference
on Soft Computing and Machine Intelligence (ISCMI). IEEE; 2015. p. 45–50.
[133] Meng Z and Pan JS. Monkey king evolution: a new memetic evolutionary
algorithm and its application in vehicle fuel consumption optimization.
Knowledge-Based Systems. 2016;97:144–157.
[134] Seref O and Akcali E. Monkey search: a new meta-heuristic approach. In:
INFORMS Annual Meeting, San Jose, CA; 2002.
[135] Bouchekara H. Most Valuable Player Algorithm: a novel optimization
algorithm inspired from sport. Operational Research. 2020;20(1):139–195.
[136] Mirjalili S. Moth-flame optimization algorithm: a novel nature-inspired
heuristic paradigm. Knowledge-Based Systems. 2015;89:228–249.
[137] Azizi M, Ghasemi SAM, Ejlali RG, et al. Optimal tuning of fuzzy para-
meters for structural motion control using multiverse optimizer. The
Structural Design of Tall and Special Buildings. 2019;28(13):e1652.
[138] Kashan AH. A new metaheuristic for optimization: optics inspired opti-
mization (OIO). Computers & Operations Research. 2015;55:99–125.
[139] Borji A. A New Global Optimization Algorithm inspired by parliamentary
political competitions. In: Gelbukh A, Kuri Morales ÁF, editors. MICAI
2007: Advances in Artificial Intelligence. Berlin, Heidelberg: Springer
Berlin Heidelberg; 2007. p. 61–71.
[140] Kennedy J and Eberhart R. Particle swarm optimization. In: Proceedings of
ICNN’95-International Conference on Neural Networks. vol. 4. IEEE;
1995. p. 1942–1948.
[141] Yapici H and Cetinkaya N. A new meta-heuristic optimizer: pathfinder
algorithm. Applied Soft Computing. 2019;78:545–568.
[142] Trojovský P and Dehghani M. Pelican optimization algorithm: a novel
nature-inspired algorithm for engineering applications. Sensors. 2022;22
(3):855.
 Advanced optimization by nature-inspired algorithm 141

[143] Duan H and Qiao P. Pigeon-inspired optimization: a new swarm intelli-

gence optimizer for air robot path planning. International Journal of
Intelligent Computing and Cybernetics. 2014;7(1):24–37.
[144] Po-lap D and Wo´zniak M. Polar bear optimization algorithm: meta-
heuristic with fast population movement and dynamic birth and death
mechanism. Symmetry. 2017;9(10):203.
[145] Askari Q, Younas I, and Saeed M. Political optimizer: a novel socio-
inspired meta-heuristic for global optimization. Knowledge-Based Systems.
2020;195:105709. https://www.sciencedirect.com/science/article/pii/ S09507
05120301350.
[146] Moosavi SHS and Bardsiri VK. Poor and rich optimization algorithm: a
new human-based and multi populations algorithm. Engineering
Applications of Artificial Intelligence. 2019;86:165–181.
[147] Zeidabadi FA and Dehghani M. Poa: puzzle optimization algorithm.
International Journal of Intelligent Systems. 2022;15:273–281.
[148] Farhi E, Goldstone J, and Gutmann S. A quantum approximate optimiza-
tion algorithm. arXiv preprint arXiv:14114028. 2014.
[149] Zhang J, Xiao M, Gao L, et al. Queuing search algorithm: a novel meta-
heuristic algorithm for solving engineering optimization problems. Applied
Mathematical Modelling. 2018;63:464–490.
[150] Koohi SZ, NAWA H, Othman M, et al. Raccoon optimization algorithm.
IEEE Access. 2018;7:5383–5399.
[151] Moazzeni AR and Khamehchi E. Rain optimization algorithm (ROA): a
new metaheuristic method for drilling optimization solutions. Journal of
Petroleum Science and Engineering. 2020;195:107512.
[152] Kaboli SHA, Selvaraj J, and Rahim N. Rain-fall optimization algorithm: a
population based algorithm for solving constrained optimization problems.
Journal of Computational Science. 2017;19:31–42.
[153] Rao RV and Pawar RB. Self-adaptive multi-population Rao Algorithms for
engineering design optimization. Applied Artificial Intelligence. 2020;34
(3):187–250. https://doi.org/10.1080/08839514.2020.1712789.
[154] Brabazon A, Cui W, and O’Neill M. The raven roosting optimisation
algorithm. Soft Computing. 2016;20(2):525–545.
[155] Kaveh A and Khayatazad M. A new meta-heuristic method: ray optimiza-
tion. Computers & Structures. 2012;112:283–294.
[156] Fard A and Hajiaghaei-Keshteli M. Red Deer Algorithm (RDA);
a new optimization algorithm inspired by Red Deers’ mating. In:
International Conference on Industrial Engineering, IEEE. vol. 12; 2016.
p. 331–342.
[157] Po-lap D and Wo´zniak M. Red fox optimization algorithm. Expert Systems
with Applications. 2021;166:114107.
[158] Badar AQH, Umre BS, and Junghare AS. Reducing Variable Trend Search
algorithm for optimizing non linear multidimensional space search. In:
2014 International Conference on Advances in Engineering Technology
Research (ICAETR-2014); 2014. p. 1–4.
142 Nature-inspired optimization algorithms and soft computing

[159] Jia H, Peng X, and Lang C. Remora optimization algorithm. Expert Systems
with Applications. 2021;185:115665.
[160] Abualigah L, Abd Elaziz M, Sumari P, et al. Reptile Search Algorithm
(RSA): a nature-inspired meta-heuristic optimizer. Expert Systems with
Applications. 2022;191:116158.
[161] Labbi Y, Attous DB, Gabbar HA, et al. A new rooted tree optimization
algorithm for economic dispatch with valve-point effect. International
Journal of Electrical Power & Energy Systems. 2016;79:298–311.
[162] Ahmadianfar I, Heidari AA, Gandomi AH, et al. RUN beyond the meta-
phor: an efficient optimization algorithm based on Runge Kutta method.
Expert Systems with Applications. 2021;181:115079.
[163] Mirjalili S, Gandomi AH, Mirjalili SZ, et al. Salp Swarm Algorithm: a bio-
inspired optimizer for engineering design problems. Advances in
Engineering Software. 2017;114:163–191.
[164] Kaur A, Jain S, and Goel S. Sandpiper optimization algorithm: a novel
approach for solving real-life engineering problems. Applied Intelligence.
2020;50(2):582–619.
[165] Moosavi SHS and Bardsiri VK. Satin bowerbird optimizer: a new optimi-
zation algorithm to optimize ANFIS for software development effort esti-
mation. Engineering Applications of Artificial Intelligence. 2017;60:1–15.
[166] Glover F. Heuristics for integer programming using surrogate constraints.
Decision Sciences. 1977;8(1):156–166.
[167] Dhiman G and Kumar V. Seagull optimization algorithm: theory and its
applications for large-scale industrial engineering problems. Knowledge-
Based Systems. 2019;165:169–196.
[168] Masadeh R, Mahafzah BA, and Sharieh A. Sea lion optimization algorithm.
International Journal of Advanced Computer Science and Applications.
2019;10(5).
[169] Shabani A, Asgarian B, Salido M, et al. Search and rescue optimization
algorithm: a new optimization method for solving constrained engineering
optimization problems. Expert Systems with Applications. 2020;161:
113698.
[170] Emami H. Seasons optimization algorithm. Engineering with Computers.
2020;38:1–21.
[171] Dai C, Zhu Y, Chen W. Seeker optimization algorithm. In: International
Conference on Computational and Information Science. Springer; 2006.
p. 167–176.
[172] Fausto F, Cuevas E, Valdivia A, et al. A global optimization algorithm
inspired in the behavior of selfish herds. Biosystems. 2017;160:
39–55.
[173] Abedinia O, Amjady N, and Ghasemi A. A new metaheuristic algorithm
based on shark smell optimization. Complexity. 2016;21(5):97–116.
[174] Duan Q, Gupta VK, and Sorooshian S. Shuffled complex evolution
approach for effective and efficient global minimization. Journal of
Optimization Theory and Applications. 1993;76(3):501–521.
 Advanced optimization by nature-inspired algorithm 143

[175] Eusuff M, Lansey K, and Pasha F. Shuffled frog-leaping algorithm: a

memetic meta-heuristic for discrete optimization. Engineering Optimization.
2006;38(2):129–154.
[176] Kirkpatrick S, Gelatt Jr CD, and Vecchi MP. Optimization by simulated
annealing. Science. 1983;220(4598):671–680.
[177] Mirjalili S. SCA: a sine cosine algorithm for solving optimization pro-
blems. Knowledge-Based Systems. 2016;96:120–133.
[178] Li S, Chen H, Wang M, et al. Slime mould algorithm: a new method for sto-
chastic optimization. Future Generation Computer Systems. 2020;111:300–323.
[179] Hashim FA and Hussien AG. Snake optimizer: a novel meta-heuristic
optimization algorithm. Knowledge-Based Systems. 2022;242:108320.
[180] Balochian S and Baloochian H. Social mimic optimization algorithm and
engineering applications. Expert Systems with Applications. 2019;134:178–191.
[181] Cuevas E, Cienfuegos M, Zaldı́var D, et al. A swarm optimization algo-
rithm inspired in the behavior of the social-spider. Expert Systems with
Applications. 2013;40(16):6374–6384.
[182] James J and Li VO. A social spider algorithm for global optimization.
Applied Soft Computing. 2015;30:614–627.
[183] Kumar M, Kulkarni AJ, and Satapathy SC. Socio evolution and learning
optimization algorithm: a socio-inspired optimization methodology. Future
Generation Computer Systems. 2018;81:252–272.
[184] Dhiman G and Kaur A. STOA: a bio-inspired based optimization algorithm
for industrial engineering problems. Engineering Applications of Artificial
Intelligence. 2019;82:148–174.
[185] Xue J and Shen B. A novel swarm intelligence optimization approach:
sparrow search algorithm. Systems Science & Control Engineering. 2020;8
(1):22–34.
[186] Shehadeh HA, Ahmedy I, and Idris MYI. Sperm swarm optimization
algorithm for optimizing wireless sensor network challenges. In:
Proceedings of the 6th International Conference on Communications and
Broadband Networking; 2018. p. 53–59.
[187] Bansal JC, Sharma H, Jadon SS, et al. Spider monkey optimization algo-
rithm for numerical optimization. Memetic Computing. 2014;6(1):31–47.
[188] Dhiman G and Kumar V. Spotted hyena optimizer: a novel bio-inspired
based metaheuristic technique for engineering applications. Advances in
Engineering Software. 2017;114:48–70.
[189] Jain M, Singh V, and Rani A. A novel nature-inspired algorithm for opti-
mization: squirrel search algorithm. Swarm and Evolutionary Computation.
2019;44:148–175.
[190] Cuevas E, Echavarrı́a A, Zaldı́var D, et al. A novel evolutionary algorithm
inspired by the states of matter for template matching. Expert Systems with
Applications. 2013;40(16):6359–6373.
[191] Gomes GF, da Cunha SS, and Ancelotti AC. A sunflower optimization
(SFO) algorithm applied to damage identification on laminated composite
plates. Engineering with Computers. 2019;35(2):619–626.
144 Nature-inspired optimization algorithms and soft computing

[192] Zhao W, Wang L, and Zhang Z. Supply-demand-based optimization: a

novel economics-inspired algorithm for global optimization. IEEE Access.
2019;7:73182–73206.
[193] Cheng MY and Prayogo D. Symbiotic organisms search: a new metaheur-
istic optimization algorithm. Computers & Structures. 2014;139:98–112.
[194] Glover F. Tabu search—Part I. ORSA Journal on Computing. 1989;1
(3):190–206.
[195] Glover F. Tabu search—Part II. ORSA Journal on Computing. 1990;2
(1):4–32.
[196] Rao RV, Savsani VJ, and Vakharia D. Teaching – learning-based optimi-
zation: a novel method for constrained mechanical design optimization
problems. Computer-Aided Design. 2011;43(3):303–315.
[197] Dehghani M and Trojovsky´ P. Teamwork optimization algorithm: a new
optimization approach for function minimization/maximization. Sensors.
2021;21(13):4567.
[198] Kaveh A and Dadras A. A novel meta-heuristic optimization algorithm:
thermal exchange optimization. Advances in Engineering Software.
2017;110:69–84.
[199] Qais MH, Hasanien HM, and Alghuwainem S. Transient search optimiza-
tion: a new meta-heuristic optimization algorithm. Applied Intelligence.
2020;50(11):3926–3941.
[200] Mladenovi´c N and Hansen P. Variable neighborhood search. Computers &
Operations Research. 1997;24(11):1097–1100.
[201] Yang XS. Engineering optimizations via nature-inspired virtual bee algo-
rithms. In: International Work-Conference on the Interplay Between
Natural and Artificial Computation. Springer; 2005. p. 317–323.
[202] Jaderyan M and Khotanlou H. Virulence optimization algorithm. Applied
Soft Computing. 2016;43:596–618.
[203] Liang YC and Cuevas Juarez JR. A novel metaheuristic for continuous
optimization problems: virus optimization algorithm. Engineering
Optimization. 2016;48(1):73–93.
[204] Doğan B and Ölmez T. A new metaheuristic for numerical function optimi-
zation: Vortex Search algorithm. Information Sciences. 2015;293:125–145.
[205] Ayyarao TS, RamaKrishna N, Elavarasan RM, et al. War strategy optimi-
zation algorithm: a new effective meta-heuristic algorithm for global opti-
mization. IEEE Access. 2022;10:25073–25105.
[206] Eskandar H, Sadollah A, Bahreininejad A, et al. Water cycle algorithm–A
novel metaheuristic optimization method for solving constrained
engineering optimization problems. Computers & Structures. 2012;110:
151–166.
[207] Kaveh A and Bakhshpoori T. Water evaporation optimization: a novel
physically inspired optimization algorithm. Computers & Structures.
2016;167:69–85.
[208] Zheng YJ. Water wave optimization: a new nature-inspired metaheuristic.
Computers & Operations Research. 2015;55:1–11.
 Advanced optimization by nature-inspired algorithm 145

[209] Asgari HR, Bozorg Haddad O, Pazoki M, et al. Weed optimization algo-
rithm for optimal reservoir operation. Journal of Irrigation and Drainage
Engineering. 2016;142(2):04015055.
[210] Ahmadianfar I, Heidari AA, Noshadian S, et al. INFO: an efficient opti-
mization algorithm based on weighted mean of vectors. Expert Systems
with Applications. 2022;195:116516.
[211] Mirjalili S and Lewis A. The whale optimization algorithm. Advances in
Engineering Software. 2016;95:51–67.
[212] Bayraktar Z, Komurcu M, and Werner DH. Wind Driven Optimization
(WDO): a novel nature-inspired optimization algorithm and its application
to electromagnetics. In: 2010 IEEE Antennas and Propagation Society
International Symposium. IEEE; 2010. p. 1–4.
[213] Covic N and Lacevic B. Wingsuit flying search—a novel global optimi-
zation algorithm. IEEE Access. 2020;8:53883–53900.
[214] Shahrezaee M. Image segmentation based on world cup optimization
algorithm. Majlesi Journal of Electrical Engineering. 2017;11(2).
[215] Punnathanam V and Kotecha P. Yin-Yang-pair optimization: a novel
lightweight optimization algorithm. Engineering Applications of Artificial
Intelligence. 2016;54:62–79.
This page intentionally left blank
 Chapter 6
Application and challenges of optimization in
Internet of Things (IoT)
Hemlata Sharma1, Shilpa Srivastava2 and Varuna Gupta2

Optimization is a key part of the Internet of Things (IoT) ecosystem. It is used to

make IoT devices, networks, and applications as efficient and effective as possible.
In this chapter, a number of optimization algorithms of various technologies have
been discussed in IoT like network, nature-inspired, evolutionary, and bio-inspired
optimization. In network optimization, IoT devices rely on wireless communication
protocols to transmit data, and the optimization of the network can help to reduce
latency, improve reliability, and minimize power consumption. Nature-inspired
optimization has shown great promise in optimizing various aspects of IoT systems,
and their use is likely to become more widespread as IoT continues to grow and
evolve. Evolutionary optimization can be used in the IoT to optimize various
aspects of the system, including resource allocation, routing, and energy manage-
ment. Bio-inspired optimization algorithms are designed to solve complex pro-
blems by simulating the behavior of natural systems, such as biological organisms,
swarms of insects, or ecological systems. In a comparative analysis of various
optimization algorithms, it can be safely concluded that optimization plays a vital
role in IoT, but it requires careful planning and execution to overcome the asso-
ciated challenges and realize the benefits of this technology.

6.1 Introduction
Optimization is crucial in the Internet of Things (IoT) to maximize the efficiency of
various IoT systems and devices. Some applications of optimization in the IoT
include resource optimization, energy optimization, network optimization, and
routing optimization. Despite the potential benefits of optimization in IoT, there are
also several challenges that need to be addressed like scalability, real-time pro-
cessing, security, and interoperability.
Overall, optimization is a critical component of IoT systems, and overcoming
these challenges will be key to realizing the full potential of this emerging

1
Sheffield Hallam University, UK
2
Christ (Deemed to be University) – Delhi, India
148 Nature-inspired optimization algorithms and soft computing

technology. To overcome the challenges, a number of optimization algorithms have

been proposed in the chapter.
Natural-inspired optimization algorithms have demonstrated considerable
promise for enhancing the effectiveness and performance of IoT systems, and their
use is anticipated to extend as the IoT develops and expands.
IoT systems that use bio-inspired heuristic algorithms have a number of benefits that
make them a desirable option for many optimization issues. In the context of IoT, they
provide a reliable, effective, and adaptable method to solve various real-life problems.[1].
Overall, the many evolutionary algorithm types use the principles and proce-
dures of natural selection and evolution to offer adaptable and effective solutions to
a variety of optimization challenges in the IoT. These techniques may be used to
optimize IoT systems for increased performance, scalability, and reliability.
In general, network optimization is a key component of IoT systems, and the
choice of the most suited algorithms relies on the particular needs of the network
and the devices involved.
These methods are based on the ideas of genetic algorithms, swarm intelli-
gence, and natural selection. By optimizing numerous factors including energy
usage, network traffic, and data processing, they may be utilized to increase the
effectiveness and performance of IoT systems.
Comparative examination of several optimization algorithms reveals that
optimization plays a crucial role in the IoT, but overcoming the related hurdles and
realizing the benefits of this technology needs careful preparation and execution.
This chapter is divided into six sections. Several applications of optimization in
IoT are explored in the first section, followed by network optimization methods in
the second section. In Sections 3–5, nature-inspired, evolutionary, and bio-inspired
algorithms are discussed. Finally, a comparison of all algorithms based on their
usefulness has been made.

6.2 Application of optimization in the IoT

Optimization techniques play a crucial role in IoT to ensure efficient resource
utilization and enhance system performance [2]. These are some of the key areas
where optimization is applied in IoT to ensure efficient, secure, and cost-effective
operations [3–5].
Some of the key applications of optimization in IoT are:
Network resource allocation: Optimization algorithms are used to allocate net-
work resources such as bandwidth and power to IoT devices, to ensure effective
utilization of these resources.
Data collection: Optimization techniques are applied to minimize data transmis-
sion costs in IoT systems by selecting the most cost-effective data collection
method and scheduling data transmission to reduce network congestion [6].
Edge computing: Optimization algorithms help in optimizing the distribution of
computing tasks between IoT devices and edge servers, reducing the load on the
network, and ensuring effective use of computing resources.
 Application and challenges of optimization in IoT 149

Sensor placement: Optimization algorithms are used to determine the optimal

placement of sensors in IoT systems, taking into account the trade-off between
sensing accuracy and energy consumption.
Security: Optimization algorithms are used to determine the optimal security measures to
protect IoT devices and networks, considering the trade-off between security and cost.

6.2.1 Challenges of optimization in IoT

Despite the potential benefits of optimization, there are several challenges asso-
ciated with implementing it in IoT:
Heterogeneous devices: The wide range of devices and technologies used in IoT
presents a significant challenge for optimization algorithms, as they must be able to
work effectively with different types of devices and data [7,8].
Network constraints: The limited resources and connectivity issues in IoT net-
works pose a challenge for optimization algorithms, as they must be able to operate
effectively within these constraints.
Real-time processing: IoT systems often require real-time processing of data,
which poses a significant challenge for optimization algorithms, as they must be
able to make decisions quickly and effectively.
Data privacy and security: Optimization algorithms must be designed to respect
the privacy and security of IoT data, as well as the security of the devices and
networks themselves.
Complex interactions: The interactions between devices and systems in IoT can
be complex and difficult to predict, presenting a challenge for optimization algo-
rithms, as they must be able to make decisions based on these interactions [8].
Overall, optimization in IoT is a crucial area for improving system perfor-
mance and efficiency, but it also presents significant technical and logistical chal-
lenges that must be overcome to achieve these goals.

6.3 Network optimization in IoT

Network optimization in IoT refers to the process of improving the performance,

scalability, security, and cost-effectiveness of the communication networks that
connect IoT devices and systems [9]. This involves selecting and configuring net-
work technologies, protocols, and services that are suitable for the specific needs
and constraints of IoT applications and services. Network optimization can also
involve the use of tools and techniques such as traffic management, network vir-
tualization, and network slicing to improve network utilization and reduce network
congestion and latency. Overall, the goal of network optimization in IoT is to
enable efficient and reliable communication between IoT devices, applications, and
services, while minimizing the costs and risks associated with network operations
and maintenance. IoT devices rely on wireless communication protocols to transmit
data, and the optimization of the network can help to reduce latency, improve
reliability, and minimize power consumption.
150 Nature-inspired optimization algorithms and soft computing

6.3.1 Types of network optimization in IOT

Here are some strategies for optimizing networks in IoT:
1. Protocol selection: IoT devices support various wireless communication pro-
tocols, such as Bluetooth, Wi-Fi, ZigBee, and LoRaWAN. Protocol selection
depends on the specific use case and the requirements of the application. For
example, if the application requires low power consumption, then ZigBee or
LoRaWAN would be a better choice.
2. Traffic shaping and management: This involves controlling and managing
the flow of data and traffic within the IoT network to reduce congestion,
minimize latency, and improve reliability.
3. Network virtualization: This involves dividing a physical network into mul-
tiple virtual networks that can run on the same physical infrastructure, allowing
for efficient resource utilization and flexible network configuration.
4. Network slicing: This involves creating multiple isolated and dedicated net-
work slices within a common physical network infrastructure, each designed to
serve a specific IoT application or use case with guaranteed network perfor-
mance and quality of service (QoS).
5. Security optimization: This involves implementing measures to secure IoT
devices, networks, and data from threats such as hacking, unauthorized access,
and data breaches.
6. Energy optimization: This involves reducing the energy consumption of IoT
devices and networks to extend battery life, reduce operational costs, and
reduce the environmental impact of IoT.
7. Cost optimization: This involves reducing the costs associated with deploy-
ing, operating, and maintaining IoT networks by using cost-effective network
technologies, infrastructure, and services.
8. Bandwidth optimization: IoT devices transmit small amounts of data at reg-
ular intervals. By optimizing the bandwidth, we can reduce the amount of data
transmitted and extend the battery life of IoT devices. Data compression,
traffic shaping, and QoS mechanisms can be used to optimize bandwidth.
9. Routing optimization: Routing optimization can be used to minimize the
latency and improve the reliability of IoT networks. This can be achieved
through the use of efficient routing algorithms, such as shortest path, load
balancing, and multipath routing.
In summary, network optimization is a critical aspect of IoT systems, and it is
essential to consider the specific requirements of the application and the network
to develop an effective optimization strategy (Figure 6.1). By optimizing the
network, we can improve the performance, reliability, and security of IoT devices
and applications [10].

6.3.2 Algorithms of network optimization in IoT

Network optimization is an important aspect of designing and deploying IoT sys-
tems. IoT networks often involve a large number of devices communicating with
 Application and challenges of optimization in IoT 151

Routing algorithms

Data aggregation algorithms

Security algorithms
Resource allocation algorithms

Predictive analytics algorithms

Figure 6.1 Network optimization algorithms in IOT

each other and with the cloud, and optimizing the network can improve efficiency,
reduce latency, and increase reliability [11]. Here are some common algorithms
used for network optimization in IoT:
● Routing algorithms: These algorithms determine the optimal path for data to
travel between devices in the network. This can involve selecting the shortest
or most reliable path, or balancing the load across different paths [12].
Examples of routing algorithms include Dijkstra’s algorithm, A* search, and
genetic algorithms.
● Resource allocation algorithms: These algorithms manage the allocation of
network resources such as bandwidth, power, and memory. They can be used
to balance the resources available to different devices in the network and
prevent overload or congestion. Examples of resource allocation algorithms
include time-division multiple access (TDMA), frequency-division multiple
access (FDMA), and code-division multiple access (CDMA) [13].
● Data aggregation algorithms: These algorithms reduce the amount of data
transmitted across the network by aggregating data from multiple devices
before sending it to the cloud. This can reduce network traffic and improve
efficiency. Examples of data aggregation algorithms include hierarchical
clustering, centroid-based clustering, and k-means clustering.
● Predictive analytics algorithms: These algorithms analyze data from the
network to predict future trends and patterns. This can be used to optimize the
network by anticipating changes in usage and adjusting network resources
accordingly. Examples of predictive analytics algorithms include decision
trees, random forests, and neural networks.
● Security algorithms: These algorithms protect the network from unauthorized
access, data breaches, and other security threats [14]. They can be used to
encrypt data, authenticate devices, and monitor network activity for anomalies.
Examples of security algorithms include AES, RSA, and SHA-256.
Overall, network optimization is a critical component of IoT systems, and the
selection of appropriate algorithms depends on the specific requirements of the
network and the devices involved.
152 Nature-inspired optimization algorithms and soft computing

6.3.3 Advantages of network optimization in IoT

There are several advantages to network optimization in IoT, including:
● Improved performance: Network optimization can improve the speed, relia-
bility, and responsiveness of IoT networks, which can lead to better user
experience and increased efficiency of IoT applications and services.
● Increased scalability: Network optimization can help IoT networks scale up or
down as needed to accommodate changes in demand and traffic patterns,
allowing for flexible and efficient resource utilization [15].
● Enhanced security: Network optimization can help secure IoT networks and
devices against cyber threats and attacks, protecting sensitive data and intel-
lectual property.
● Reduced costs: Network optimization can help reduce the costs associated
with deploying, operating, and maintaining IoT networks by using cost-
effective network technologies, infrastructure, and services.
● Increased energy efficiency: Network optimization can help reduce the energy
consumption of IoT networks, extending battery life and reducing the envir-
onmental impact of IoT.
● Improved quality of service: Network optimization can help ensure consistent
and reliable network performance, reducing network downtime and improving
the quality of service for IoT applications and services.
● Better network utilization: Network optimization can help improve the utili-
zation of network resources, reducing congestion and latency, and ensuring
efficient use of network infrastructure and bandwidth.

6.3.4 Disadvantages of network optimization in IoT

While network optimization in IoT has many benefits, there are also some potential
disadvantages that should be considered. Here are some of the key disadvantages of
network optimization in IoT:
● Complexity: Optimizing IoT networks can be complex, especially when
dealing with large numbers of devices and complex data flows. This can lead
to increased costs and longer development times.
● Limited bandwidth: IoT networks often have limited bandwidth, which can
be a major constraint when optimizing the network. This can result in slower
transmission speeds and longer processing times.
● Security concerns: Optimizing IoT networks can introduce security risks,
especially if the network is not properly secured. This can lead to data brea-
ches, unauthorized access, and other security issues.
● Interoperability issues: IoT devices often come from different vendors and
may use different protocols, which can create interoperability issues when
optimizing the network. This can lead to compatibility issues, increased com-
plexity, and higher costs.
● Reliability issues: Optimizing IoT networks can sometimes reduce reliability,
especially if the network is highly complex or if new technologies are
 Application and challenges of optimization in IoT 153

introduced. This can result in downtime, lost data, and other issues that can
impact the overall performance of the network.
● Privacy concerns: IoT networks often collect and transmit large amounts of
data, which can raise privacy concerns for users. This can result in decreased
trust and increased regulatory scrutiny.
Overall, while network optimization in IoT has many benefits, it is important
to carefully consider the potential disadvantages before implementing optimization
strategies. This will help to ensure that the benefits outweigh the risks and that the
network operates effectively and securely.

6.4 Nature-inspired optimization in IoT

Nature-inspired optimization refers to a class of optimization algorithms that are
inspired by the mechanisms and processes found in nature, such as evolution,
migration, and natural selection. In IoT, nature-inspired optimization techniques,
also known as metaheuristics, are a class of algorithms that mimic the behavior of
natural systems to solve complex optimization problems [16]. These techniques
have been applied to a wide range of fields, including the IoT, to improve perfor-
mance, reduce power consumption, and enhance overall system efficiency. Some
examples of nature-inspired optimization algorithms that are used in IoT include:

6.4.1 Algorithms of nature-inspired optimization in IoT

These nature-inspired optimization algorithms (Figure 6.2) can provide effective and
efficient solutions to optimize the deployment of IoT devices, the placement of sensors
in a wireless sensor network, the energy consumption of IoT devices, etc. [17].
● Genetic algorithms: It is a type of optimization algorithm that is inspired by
the process of natural selection and evolution. It uses principles such as
inheritance, mutation, and recombination to find the best solution to a problem.

Genetic algorithms Bee algorithm

Particle swarm optimization

Ant colony optimization

Figure 6.2 Nature-inspired algorithms in IoT

154 Nature-inspired optimization algorithms and soft computing

In a genetic algorithm, a set of candidate solutions to a problem, called

individuals or chromosomes, are represented as strings of bits or numbers.
Each individual represents a potential solution to the problem and is assigned a
fitness value that represents how well the solution performs compared to other
solutions [18].
The genetic algorithm then iteratively applies genetic operations such as
selection, crossover, and mutation to the population of individuals to produce
new generations of solutions. The selection operation is used to choose the
most fit individuals to survive and propagate to the next generation. The
crossover operation combines the genetic information of two individuals to
produce a new offspring. The mutation operation randomly changes the
genetic information of an individual to introduce diversity into the population.
The process continues until a stopping criterion is met, such as a maximum
number of generations or a solution with a fitness value that meets a certain
threshold. The best solution found by the genetic algorithm is then returned as
the final answer to the problem. Genetic algorithms have been applied to
various problems in IoT, including network topology design, resource alloca-
tion, and traffic management. They offer a flexible and efficient way to find
the best solution to optimization problems in IoT, by leveraging the mechan-
isms and processes found in nature.
● Ant colony optimization: It is a type of optimization algorithm that is inspired
by the behavior of ant colonies. It uses principles such as cooperation, com-
munication, and self-organization to solve optimization problems.
In ACO, a set of artificial ants are used to explore the solution space of a
problem and find the best solution. Each ant constructs a solution by visiting a
set of nodes in a graph and leaving a trail of pheromones, which are used to
communicate information about the quality of solutions to other ants. The
pheromones represent the strength of the connection between nodes and help
guide the ants toward promising areas of the solution space.
The optimization process consists of multiple iterations, during which the
ants construct solutions, update the pheromone trails, and adjust the para-
meters of the algorithm based on the quality of the solutions found. The
process continues until a stopping criterion is met, such as a maximum
number of iterations or a solution with a fitness value that meets a certain
threshold.
ACO has been applied to various problems in IoT, including network
topology design, resource allocation, and traffic management. It offers a flex-
ible and efficient way to find the best solution to optimization problems in IoT,
by leveraging the mechanisms of cooperation and communication found in ant
colonies.
● Particle swarm optimization: It is a heuristic optimization algorithm inspired
by the social behavior of birds or other animals in a flock or swarm. It is used
to find the global optimum of a function in a search space. In particle swarm
optimization (PSO), each particle represents a candidate solution and moves
through the search space guided by its own experience and the experiences of
 Application and challenges of optimization in IoT 155

other particles in the swarm. The position and velocity of each particle are
updated at each iteration based on its own best solution found so far (pbest) and
the best solution found by any particle in the swarm (gbest). PSO has been
applied to a wide range of optimization problems and is known for its sim-
plicity and effective convergence characteristics.
● Bee Algorithm: It is a metaheuristic optimization algorithm that is inspired by
the foraging behavior of honeybees. It is used to find the global optimum of a
function in a search space. In BA, the swarm of bees is divided into employed
bees, onlooker bees, and scout bees. The employed bees search for nectar
sources in their assigned search space and share the information with the
onlooker bees. The onlooker bees then choose the best solution based on the
information shared by the employed bees. The scout bees, on the other hand,
explore new search spaces if the solution quality degrades. The best solution
found so far by the swarm is updated at each iteration. BA has been applied to
various optimization problems and is known for its fast convergence, robust-
ness, and ability to handle multimodal and complex functions.

6.4.2 Role of nature-inspired algorithms in IoT

Nature-inspired optimization plays a crucial role in the development of IoT systems
by providing efficient and effective solutions to various optimization problems in
IoT [18]. Some of the key roles of nature-inspired optimization in IoT are:

● Optimization: Nature-inspired algorithms can be used to optimize various

aspects of IoT systems, including resource allocation, energy management, and
routing. For example, algorithms such as genetic algorithms, particle swarm
optimization, and ant colony optimization can be used to find the optimal
configuration of IoT devices and networks to improve performance and reduce
energy consumption [19].
● Adaptation: IoT systems often face dynamic and uncertain environments, and
nature-inspired algorithms can be used to help these systems adapt to changing
conditions. For example, algorithms such as evolutionary programming and
artificial immune systems can be used to optimize the behavior of IoT devices
in response to changes in the environment.
● Fault tolerance: Nature-inspired algorithms can be used to improve fault
tolerance in IoT systems. For example, algorithms such as artificial bee colony
optimization can be used to improve the robustness of wireless sensor net-
works against node failures.
● Security: Nature-inspired algorithms can also be used to enhance the security
of IoT systems. For example, algorithms such as cuckoo search and firefly
algorithms can be used to optimize the encryption and decryption of data
transmitted by IoT devices.
Overall, nature-inspired algorithms can play a significant role in optimizing,
adapting, and securing IoT systems, and their use is likely to become more wide-
spread as IoT continues to grow and evolve.
156 Nature-inspired optimization algorithms and soft computing

6.4.3 Advantages of nature-inspired optimization in IoT

Nature-inspired optimization algorithms have several advantages in IoT applications:

● Flexibility: Nature-inspired optimization algorithms can be applied to a wide

range of optimization problems in IoT, such as network design, resource
allocation, traffic management, and security, among others. They can also be
easily adapted to handle new problems and changing requirements.
● Global optimization: Nature-inspired optimization algorithms are capable
of finding the global optimum solution to an optimization problem,
rather than just a local optimum. This is important in IoT applications
where a globally optimal solution is desired to ensure optimal system per-
formance [20].
● Efficient exploration: Nature-inspired optimization algorithms can efficiently
explore the solution space of a problem, by using mechanisms such as coop-
eration, communication, and self-organization. This allows them to quickly
find good solutions and avoid getting stuck in local optima [21].
● Scalability: Nature-inspired optimization algorithms can be easily scaled to
handle large and complex optimization problems in IoT, such as those invol-
ving large numbers of nodes and variables. This makes them well-suited to
handle the growing complexity of IoT systems.
● Robustness: Nature-inspired optimization algorithms are robust to changes in
the problem, such as fluctuations in the environment, and can effectively
handle uncertainty and incomplete information. This makes them suitable for
use in dynamic and unpredictable IoT environments.

Overall, the advantages of nature-inspired optimization algorithms have shown

great promise in optimizing various aspects of IoT systems, and their use is likely
to become more widespread as IoT continues to grow and evolve [19].

6.4.4 Disadvantages of nature-inspired optimization in IoT

Nature-inspired optimization techniques are increasingly being used in IoT to
address the complex optimization problems that arise in these systems. While these
techniques have many benefits, there are also some potential disadvantages that
should be considered [22]. Here are some of the key disadvantages of nature-
inspired optimization in IoT:

● Computational complexity: Nature-inspired optimization algorithms can be

computationally intensive, especially when dealing with large numbers of
devices and complex data flows. This can lead to longer processing times and
increased costs.
● Difficulty of tuning parameters: Nature-inspired optimization algorithms
typically have many parameters that need to be tuned for optimal performance.
This can be a complex and time-consuming task, requiring expertise and
specialized tools.
 Application and challenges of optimization in IoT 157

● Lack of transparency: Nature-inspired optimization algorithms are often

black-box models, which can make it difficult to understand how they work
and why they make certain decisions. This can lead to reduced trust in the
system and difficulty in identifying and fixing problems.
● Sensitivity to initial conditions: Nature-inspired optimization algorithms can
be sensitive to the initial conditions and the quality of the data used to train
them. This can lead to suboptimal results and decreased reliability.
● Lack of generalizability: Nature-inspired optimization algorithms are often
designed for specific types of problems and may not be easily generalizable to
other types of problems. This can limit their usefulness in diverse IoT
applications.
● Ethical considerations: Some nature-inspired optimization techniques are
based on evolutionary principles and may raise ethical concerns related to their
impact on natural systems and their potential to reinforce biases or create
unintended consequences.

Overall, while nature-inspired optimization techniques have many benefits, it

is important to carefully consider the potential disadvantages before implementing
them in IoT systems. This will help to ensure that the benefits outweigh the risks
and that the system operates effectively and ethically.

6.5 Evolutionary algorithms in IoT

Evolutionary algorithms (EAs) are a class of optimization algorithms that are
inspired by the process of natural selection and evolution. EAs are a class of meta-
heuristic optimization algorithms inspired by the principles of biological evolution.
They can be used in the IoT to optimize various aspects of the system, including
resource allocation, routing, and energy management.
In an evolutionary algorithm, a population of candidate solutions is evolved
over time by applying genetic operations, such as selection, crossover, and muta-
tion. The best solutions are selected for survival and reproduction, and the popu-
lation is updated with the new offspring. The process continues until a stopping
criterion is met, such as a maximum number of iterations or a solution with a fitness
value that meets a certain threshold [23].
The key advantage of EAs in IoT is their ability to find near-optimal solutions
to complex and multi-objective optimization problems. EAs can efficiently explore
the solution space, handle uncertainty and incomplete information, and adapt to
changing environments. They are also well-suited to handle large and complex
problems, such as those involving large numbers of nodes and variables [24].
Overall, EAs have shown great promise in optimizing various aspects of IoT
systems, and their use is likely to become more widespread as IoT continues to
grow and evolve. By leveraging the mechanisms and processes of natural selection
and evolution, EAs can help improve the performance, scalability, and robustness
of IoT systems.
158 Nature-inspired optimization algorithms and soft computing

6.5.1 Algorithms of evolutionary optimization in the IoT

There are several types of EAs that are used in the IoT applications:
● Genetic algorithm (GA): GA is a type of EA that is inspired by the process of
natural selection and evolution. It is based on a population of candidate solu-
tions that are evolved over time through genetic operations, such as selection,
crossover, and mutation [25].
● Differential evolution (DE): DE is a type of EA that is based on the principle
of simulating the process of evolution through the exchange of information
between candidate solutions. It is well-suited to problems with many variables
and large solution spaces.
● Particle swarm optimization (PSO): PSO is a type of EA that is inspired by
the behavior of bird flocks and fish schools [26]. It is based on a population of
candidate solutions that move in the solution space, guided by the best solu-
tions found so far [27].
● Artificial bee colony (ABC): ABC is a type of EA that is inspired by the behavior
of bee colonies. It is based on a population of candidate solutions, called bees, that
search for food sources, guided by the best solutions found so far.
● Ant colony optimization (ACO): ACO is a type of EA that is inspired by the
behavior of ant colonies. It is based on a population of candidate solutions,
called ants, that search for solutions, guided by a trail of pheromones that are
left by previous ants.
Each of these types of EAs has its own strengths and weaknesses, and the best
type to use in a specific IoT application depends on the problem at hand, the
requirements of the system, and the available computational resources. Overall, the
various types of EAs provide flexible and efficient solutions to various optimization
problems in IoT, leveraging the mechanisms and processes of natural selection and
evolution. By using these algorithms (Figure 6.3), IoT systems can be optimized for
improved performance, scalability, and robustness.

Ant Colony
Particle Swarm Optimization
Optimization

Genetic
Algorithm

Differential Artificial Bee

Evolution Colony

Figure 6.3 Evolutionary optimization algorithms in IoT

 Application and challenges of optimization in IoT 159

6.5.2 Role of evolutionary algorithms in IoT

The role of EAs in the IoT can be significant in various applications. Some of the
applications of EAs in IoT are:
● Network optimization: EAs can be used to optimize network parameters,
such as node placement, resource allocation, and routing, to improve network
performance, reduce energy consumption, and increase network lifetime.
● Machine learning: EAs can be used to train machine learning models in IoT devi-
ces, such as sensor nodes, to make predictions, classify data, and perform other tasks.
● Parameter tuning: EAs can be used to automatically tune the parameters of
IoT devices, such as sensors, to optimize their performance and reduce the
need for manual tuning.
● Sensor placement: EAs can be used to determine the optimal placement of
sensors in IoT networks to improve coverage, reduce energy consumption, and
increase network lifetime.
● Resource allocation: EAs can be used to optimize the allocation of resources,
such as bandwidth and power, in IoT networks to improve performance, reduce
energy consumption, and increase network lifetime [28].
Overall, the use of EAs in IoT can lead to more efficient, effective, and reliable
IoT systems [29].

6.5.3 Advantages of evolutionary optimization in IoT

The main advantages of using EAs in the IoT applications are:
● Flexibility: EAs are flexible algorithms that can be applied to a wide range of
optimization problems in IoT, from network design to resource allocation and
traffic management to security. They can handle complex problems with many
variables and large solution spaces.
● Near-optimal solutions: EAs can efficiently explore the solution space and
find near-optimal solutions to complex and multi-objective optimization pro-
blems in IoT. They can handle uncertainty and incomplete information and
adapt to changing environments.
● Robustness: EAs are robust algorithms that can handle large and complex
problems, such as those involving many nodes and variables in IoT systems.
They can also handle noise and non-linear relationships between variables.
● Scalability: EAs can be easily parallelized and scaled to handle large-scale
optimization problems in IoT systems. This makes them well-suited to the
requirements of large and complex IoT systems.
● Global optimization: EAs are capable of global optimization, meaning they
can find the best solutions across the entire solution space. This is particularly
useful in IoT, where many optimization problems involve global solutions,
such as network design and resource allocation.
Overall, EAs provide flexible, efficient, and robust solutions to various opti-
mization problems in IoT. By leveraging the mechanisms and processes of natural
160 Nature-inspired optimization algorithms and soft computing

selection and evolution, EAs can help improve the performance, scalability, and
robustness of IoT systems.

6.5.4 Disadvantages of evolutionary optimization in IoT

Evolutionary algorithms are optimization methods inspired by the process of nat-
ural selection and are widely used in various domains, including the IoT. However,
despite their advantages, EAs have several limitations in the context of IoT,
including:
● Scalability: EAs can be computationally intensive and may struggle with
large-scale problems, making it difficult to apply them to the huge amounts of
data generated by IoT devices.
● Convergence rate: EAs can be slow to converge to optimal solutions, espe-
cially in complex problems with many variables, which can limit their
usefulness in time-sensitive IoT applications.
● Premature convergence: EAs can sometimes converge prematurely to sub-
optimal solutions, which can limit their ability to find globally optimal
solutions to IoT problems.
● Lack of interpretability: EAs can be difficult to understand and interpret,
making it challenging for practitioners to debug and improve them.
● Data privacy and security: EAs can be vulnerable to attacks and may not be
able to effectively protect the privacy and security of sensitive IoT data.
In conclusion, while EAs have shown great potential in many areas, they have
several limitations in the context of IoT and additional research is needed to
overcome these challenges and make them more suitable for IoT applications.

6.6 Bio-inspired heuristic algorithms in IoT

Bio-inspired optimization, also known as biologically inspired optimization, is a

field of study that draws inspiration from biological systems and processes to
develop optimization algorithms. Bio-inspired optimization algorithms are
designed to solve complex problems by simulating the behavior of natural systems,
such as biological organisms, swarms of insects, or ecological systems [30].
The increasing complexity of real-world problems motivates researchers to
search for efficient methods. Divide and conquer techniques are the one way to
solve large and complex problems which has been a practice in research for a
long time.
Swarms have relatively simple behaviors individually, but with the amazing
capability of coordinating and organizing their actions, they represent a complex
and highly structured social organization [31]. Examples are bee colonies, ant
colonies, mosquito swarms, fish schools, birds, flies, and particle swarms.
Swarm Intelligence is a branch of biologically inspired algorithms, which is
focused on the collective behavior of swarms in order to develop some meta-
heuristics that mimic the swarm’s problem-solution abilities. Taking the inspiration
 Application and challenges of optimization in IoT 161

from success and efficiency of the distributed, coordinated, and collective behavior
of swarms in the real world, researchers have tried to develop sophisticated meth-
ods and systems that make use of the techniques of the swarms to find solutions to
complex optimization problems.
Bio-inspired algorithms are a type of heuristic algorithms that are inspired by
nature and biological systems. In the context of the IoT, bio-inspired algorithms can
be used to solve various optimization problems and improve the efficiency of IoT
systems.
For example, genetic algorithms (GAs) are popular bio-inspired algorithms
that can be used for optimizing complex problems in IoT. In a GA, a population
of candidate solutions is evolved over multiple generations using genetic opera-
tions such as selection, crossover, and mutation. The solutions are evaluated
based on their fitness and the best solutions are selected for the next generation.
This process is repeated until a satisfactory solution is found or a stopping
criterion is met. Another example is the use of ACO algorithms in IoT. ACO
algorithms are inspired by the behavior of ant colonies, where ants work together
to find the shortest path between their colony and a food source. In IoT, ACO
algorithms can be used to find the shortest path for data transmission or to solve
routing problems.
In addition to GAs and ACO, there are other bio-inspired algorithms such as
PSO and ABC that can be applied to various optimization problems in IoT.
Overall, the use of bio-inspired algorithms in IoT can help improve the effi-
ciency and performance of IoT systems by providing better solutions to complex
problems.

6.6.1 Types of bio-inspired heuristic algorithm in IoT

There are several types of bio-inspired heuristic algorithms that can be used in the
context of the IoT:
● GAs: They are a type of optimization algorithm inspired by the principles of
evolution and natural selection. They can be used in IoT for problems such as
network optimization, device placement, and data analysis.
● ACO: They are inspired by the behavior of ant colonies and their ability to
find the shortest path between their colony and a food source. In IoT, ACO
algorithms can be used for routing and data transmission problems.
● PSO: It is a heuristic optimization algorithm inspired by the behavior of birds
flocking and fish schooling. In IoT, PSO can be used for problems such as
device placement, network optimization, and data analysis.
● ABC: It is a bio-inspired optimization algorithm inspired by the behavior of
honeybees in searching for food. In IoT, ABC can be used for problems such as
network optimization, device placement, and data analysis [32].
● Firefly algorithm (FA): It is a bio-inspired optimization algorithm inspired by
the behavior of fireflies and their ability to communicate and find mates. In
IoT, FA can be used for problems such as network optimization, device pla-
cement, and data analysis [33].
162 Nature-inspired optimization algorithms and soft computing

Firefly Algorithm

Particle Swarm Optimization

Artificial Bee Colony

Ant Colony Optimization Genetic Algorithms

Bacterial Foraging Optimization

Figure 6.4 Bio-inspired optimization algorithms in IoT

● Bacterial foraging optimization (BFO): It is a bio-inspired optimization

algorithm inspired by the behavior of bacteria in searching for food. In IoT,
BFO can be used for problems such as network optimization, device place-
ment, and data analysis.
These are some of the most commonly used bio-inspired heuristic algorithms
in IoT (Figure 6.4). The choice of a specific algorithm will depend on the specific
problem being solved and the desired performance characteristics.
In the context of the IoT, bio-inspired optimization algorithms can be used to
optimize various aspects of the system, including resource allocation, routing, and
energy management.

6.6.2 Role of bio-inspired heuristic algorithm in IoT

The role of bio-inspired heuristic algorithms in the IoT is to provide efficient and
effective solutions to complex optimization problems that arise in the context of
IoT systems. Some of the key roles of these algorithms in IoT include:
● Optimization: Bio-inspired algorithms can be used to optimize network
performance in IoT systems by finding the optimal placement of devices,
optimizing the routing of data, and reducing energy consumption [34].
● Device placement: Bio-inspired algorithms can be used to determine the
optimal placement of devices in an IoT system to maximize performance and
reduce energy consumption.
● Data analysis: Bio-inspired algorithms can be used to analyze large amounts
of data generated by IoT devices to identify patterns, relationships, and trends,
and to make predictions about future behavior.
● Resource allocation: Bio-inspired algorithms can be used to allocate resour-
ces in IoT systems such as bandwidth, processing power, and storage, in an
efficient and effective manner.
 Application and challenges of optimization in IoT 163

● Decision making: Bio-inspired algorithms can be used to make decisions in

IoT systems, such as determining the best course of action in response to a
change in the environment.
Overall, the role of bio-inspired heuristic algorithms in IoT is to provide effi-
cient and effective solutions to complex problems, while reducing energy con-
sumption and improving performance. They can help optimize the performance of
IoT systems, making them more scalable, reliable, and efficient.

6.6.3 Advantages of bio-inspired heuristic algorithm in IoT

The use of bio-inspired heuristic algorithms in the IoT systems provides several
advantages, including:
● Efficiency: Bio-inspired algorithms are designed to solve complex optimiza-
tion problems quickly and efficiently. They can provide solutions in a fraction
of the time required by traditional algorithms [35].
● Robustness: Bio-inspired algorithms are designed to be robust and to provide
reliable solutions even in the presence of noise and uncertainty. They are well
suited to the dynamic and unpredictable nature of IoT systems.
● Scalability: Bio-inspired algorithms can handle large-scale problems and can
easily be scaled to meet the needs of growing IoT systems.
● Adaptability: Bio-inspired algorithms are designed to be adaptive and can
automatically adjust to changing conditions in IoT systems. This allows
them to provide optimal solutions even in the face of changing
requirements.
● Global optimization: Bio-inspired algorithms can find the global optimal
solution to a problem, rather than being limited to finding a locally optimal
solution.
● Low complexity: Bio-inspired algorithms are generally simpler and easier to
implement than traditional algorithms, making them a cost-effective solution
for many optimization problems in IoT systems.
● Natural inspiration: The inspiration from natural systems provides a power-
ful and intuitive understanding of the optimization process, which can help
with problem-solving and design.
Overall, the use of bio-inspired heuristic algorithms in IoT systems provides
several advantages that make them an attractive solution for various optimization
problems. They offer a robust, efficient, and adaptive approach to problem-solving
in the context of IoT [1].

6.6.4 Limitations of bio-inspired heuristic algorithm in IoT

Bio-inspired heuristic algorithms are inspired by biological systems and try to
mimic their behaviors to solve problems in computer science. However, despite
their success in many applications, they have several limitations in the context of
the IoT. Some of these limitations include:
164 Nature-inspired optimization algorithms and soft computing

● Scalability: Bio-inspired algorithms often struggle with large-scale problems and

may not be efficient in handling the huge amounts of data generated by IoT devices.
● Complexity: Bio-inspired algorithms can be complex and may require sig-
nificant computational resources to run, which may not be feasible for
resource-constrained IoT devices.
● Lack of interpretability: Bio-inspired algorithms can be difficult to under-
stand and interpret, making it challenging for practitioners to debug and
improve them.
● Lack of adaptivity: Bio-inspired algorithms may not be able to adapt quickly
enough to changing conditions in an IoT environment, which can limit their
effectiveness in dynamic scenarios.
● Data privacy and security: Bio-inspired algorithms can be vulnerable to
attacks and may not be able to effectively protect the privacy and security of
sensitive IoT data.

In conclusion, while bio-inspired heuristic algorithms have shown great pro-

mise in many areas, they have significant limitations in the context of IoT, and
additional research is needed to overcome these challenges.

6.7 Load optimization in cognitive IoT

Load optimization in cognitive IoT refers to the process of optimizing the use of
resources, such as computing power, bandwidth, and storage, in a cognitive IoT
system [36]. Load optimization is important in cognitive IoT to ensure that the
system is operating efficiently and effectively while avoiding overloading or
underutilizing any of its resources [37].
There are several techniques for optimizing load in cognitive IoT [38],
including:
● Resource allocation: Allocating resources in an efficient manner to ensure
that they are used optimally. This can involve using algorithms such as load
balancing or scheduling to determine how resources should be assigned to
different tasks.
● Data reduction: Reducing the amount of data that needs to be processed by
the cognitive IoT system, for example, by compressing or aggregating data.
● Dynamic resource allocation: Dynamically allocating resources based on the
current load and demand, so that resources can be redirected as needed to meet
changing requirements.
● Edge computing: Moving processing and storage closer to the source of data,
which can reduce the amount of data that needs to be transmitted and pro-
cessed by the central system.
● Distributed computing: Distributing processing and storage across multiple
devices, which can improve the overall efficiency of the system.
In conclusion, load optimization is a crucial aspect of cognitive IoT, as it
helps to ensure that the system is operating efficiently and effectively while
 Application and challenges of optimization in IoT 165

avoiding overloading or underutilizing any of its resources [39]. By implementing

techniques such as resource allocation, data reduction, dynamic resource alloca-
tion, edge computing, and distributed computing, it is possible to achieve optimal
load optimization in cognitive IoT.

6.7.1 Uses of load optimization in cognitive IoT

Load optimization in cognitive IoT is used to improve the performance and effi-
ciency of IoT systems by optimizing the allocation of computing and network
resources [40]. Here are some of the key uses of load optimization in cognitive IoT:
● Improved system performance: Load optimization can help to improve the
overall performance of IoT systems by ensuring that tasks are allocated to the
most suitable devices and networks and that the workload is distributed evenly
across the system.
● Efficient use of resources: Load optimization can help to ensure that resour-
ces are used efficiently in IoT systems, by minimizing waste and reducing the
need for additional hardware and infrastructure.
● Increased scalability: Load optimization can help to increase the scalability of
IoT systems, by enabling them to handle a larger number of devices and data
flows without becoming overloaded.
● Reduced energy consumption: Load optimization can help to reduce the
energy consumption of IoT systems, by minimizing the need for devices to
operate at full capacity and reducing the overall load on the system.
● Enhanced reliability: Load optimization can help to improve the reliability of
IoT systems, by reducing the risk of system failures and ensuring that tasks are
completed in a timely and efficient manner.
Overall, load optimization is an essential aspect of cognitive IoT, which can
help to improve the performance, efficiency, and reliability of these systems, while
also reducing costs and energy consumption [41].

6.7.2 Advantages of load optimization in cognitive IoT

Load optimization in cognitive IoT can bring several advantages to the system,
including:
● Improved resource utilization: Load optimization helps to ensure that all
resources in the cognitive IoT system, such as computing power, bandwidth,
and storage, are used effectively and efficiently, leading to better utilization of
these resources.
● Better performance: Load optimization helps to optimize the performance of
the cognitive IoT system, ensuring that it is operating at optimal levels and
providing the best possible outcomes.
● Increased efficiency: Load optimization helps to reduce the amount of data
that needs to be processed and transmitted, reducing the overall demand on the
system and increasing its efficiency.
166 Nature-inspired optimization algorithms and soft computing

● Increased reliability: Load optimization helps to ensure that the cognitive IoT
system is operating within its capacity and avoiding overloading, which can
increase the reliability and stability of the system.
● Cost savings: By optimizing the use of resources and reducing the demand on the
system, load optimization can help to reduce the overall cost of operating the cognitive
IoT system.
In conclusion, load optimization plays a crucial role in ensuring the efficient
and effective operation of the cognitive IoT system. By optimizing the use of
resources and reducing the demand on the system, load optimization can bring
several advantages, including improved resource utilization, better performance,
increased efficiency, increased reliability, and cost savings [42].

6.7.3 Disadvantages of load optimization in cognitive IoT

Despite the many advantages that load optimization can bring to cognitive IoT sys-
tems, there are also some potential disadvantages that need to be considered, including:
● Complexity: Load optimization can be a complex and challenging task,
especially in large and dynamic cognitive IoT systems. Implementing load
optimization algorithms can require significant expertise and may be difficult
for some organizations.
● Overhead: Load optimization algorithms can introduce additional overhead
into the system, which can reduce its overall efficiency and performance.
● Flexibility: Load optimization algorithms may be inflexible and may not be
able to effectively handle changes in the system, such as new devices or
changes in demand.
● Accuracy: Load optimization algorithms may not always provide accurate
results, especially in complex and dynamic systems, which can limit their
effectiveness.
● Resource constraints: Load optimization algorithms may be limited by the
available resources in the system, such as computing power, bandwidth, and
storage, which can reduce their effectiveness.
In conclusion, while load optimization can bring many advantages to cognitive
IoT systems, it is important to consider the potential disadvantages and carefully
evaluate the trade-offs before implementing load optimization in a real-world
system.

6.8 Comparative analysis

Nature-inspired algorithms, evolutionary algorithms, and bio-inspired algorithms
are all types of optimization algorithms that draw inspiration from natural systems
and processes to solve complex problems. While there is some overlap between
these techniques, they each have their own strengths and weaknesses. In the context
of network optimization, these algorithms can be used to optimize various aspects
 Application and challenges of optimization in IoT 167

Nature-inspired These algorithms are very flexible

and can be applied to wide range of
Algorithms optimization problems.

Suitable for the problems with a

Evolutionary
large number of variables or where
Algorithms the search space is complex

Bio-inspired Effective for complex or adaptive

system.
Algorithms

Involves the optimization of network

Network
performance, including resource
Optimization allocation, routing and energy
Algorithms management

Figure 6.5 Comparative analysis of optimization algorithms

of network performance, including resource allocation, routing, and energy man-

agement. Here is a comparative analysis of these algorithms in the context of
optimization (Figure 6.5).
● Network optimization: Network optimization involves the optimization of
various aspects of network performance, including resource allocation, rout-
ing, and energy management. Optimization algorithms can be used to optimize
these aspects of network performance [43–45].
● Nature-inspired algorithms: Nature-inspired algorithms, such as ant colony
optimization, particle swarm optimization, and artificial bee colony optimiza-
tion, are based on the behavior of social animals and other natural phenomena.
These algorithms are often well-suited to solving complex optimization pro-
blems and can be highly effective at finding near-optimal solutions. However,
they can be computationally intensive and may require a large number of
iterations to converge to a solution.
● Evolutionary algorithms: Evolutionary algorithms are a class of nature-
inspired algorithms that simulate the process of natural selection to find opti-
mal solutions to problems. Evolutionary algorithms, such as genetic algorithms
and differential evolution, are based on the principles of natural selection and
genetics [46,47]. These algorithms are often well-suited to solving complex
optimization problems with multiple objectives and can be highly effective at
finding near-optimal solutions. However, they can be computationally intensive
and may require a large number of iterations to converge to a solution [48].
● Bio-inspired algorithms: Bio-inspired algorithms, such as neural networks,
fuzzy logic, and swarm intelligence, are based on the behavior of biological
systems. These algorithms are often well-suited to solving complex problems
168 Nature-inspired optimization algorithms and soft computing

that require intelligence, learning, and adaptation [49,50]. They can be highly
effective at finding solutions to complex problems, but may require large
amounts of data to train the model and can be computationally intensive. In the
context of the IoT, bio-inspired optimization algorithms can be used to opti-
mize various aspects of the system, including resource allocation, routing, and
energy management.
In general, nature-inspired algorithms are well-suited to solving problems that
require the exploration of large solution spaces, while evolutionary algorithms are
well-suited to solving problems with multiple objectives. Bio-inspired algorithms
are often used in IoT applications that require learning and adaptation over time.
Ultimately, the choice of algorithm will depend on the specific needs and
requirements of the IoT system being designed.

References
[1] D. Yadav, “Blood coagulation algorithm: a novel bio-inspired meta-heuristic
algorithm for global optimization.” Mathematics, vol. 9, no. 23, pp. 3011,
2021, https://doi.org/10.3390/math9233011.
[2] E. Kanniga and P. S. Jadhav, “A study paper on forest fire detection using
wireless sensor network.” International Journal of Psychosocial Rehabilitation,
vol. 23, no. 4, pp. 397–407, 2019, https://doi.org/10.37200/ijpr/v23i4/pr190199.
[3] X. Cui and G. Chen, “Application of improved ant colony optimization in
vehicular ad-hoc network routing.” In: 2021 IEEE 3rd Eurasia Conference
on IOT, Communication and Engineering (ECICE), 2021, doi: 10.1109/
ecice52819.2021.9645678.
[4] S. J. Wagh, M. S. Bhende, and A. D. Thakare, “Blockchain and IoT opti-
mization.” In: Energy Optimization Protocol Design for Sensor Networks in
IoT Domains, pp. 205–224, 2022, doi: 10.1201/9781003310549-9.
[5] A. Bhavya, K. Harleen, and C. Ritu, Transforming the Internet of Things for
Next-Generation Smart Systems. IGI Global, 2021.
[6] A. Kumar, P. S. Rathore, V. G. Diaz, and R. Agrawal, Swarm Intelligence
Optimization. John Wiley & Sons, 2021.
[7] U. Kaur and Shalu, “Blockchain- and deep learning-empowered resource
optimization in future cellular networks, edge computing, and IoT: open
challenges and current solutions.” In: Blockchain for 5G-Enabled IoT,
pp. 441–474, 2021, doi: 10.1007/978-3-030-67490-8_17.
[8] Introduction to Internet of Things (Basic Concept, Challenges, Security
Issues, Applications & Architecture). Nitya Publications, 2020.
[9] M. A. R. Khan, S. N. Shavkatovich, B. Nagpal, et al., “Optimizing hybrid
metaheuristic algorithm with cluster head to improve performance metrics
on the IoT.” Theoretical Computer Science, vol. 927, pp. 87–97, 2022,
https://doi.org/10.1016/j.tcs.2022.05.031.
[10] K. Gulati, R. S. Kumar Boddu, D. Kapila, S. L. Bangare, N. Chandnani, and
G. Saravanan, “A review paper on wireless sensor network techniques in
 Application and challenges of optimization in IoT 169

Internet of Things (IoT).” In: Materials Today: Proceedings, 2021, doi:

https://doi.org/10.1016/j.matpr.2021.05.067.
[11] N. N. Srinidhi, S. M. Dilip Kumar, and K. R. Venugopal, “Network opti-
mizations in the Internet of Things: a review.” Engineering Science and
Technology, an International Journal, vol. 22, no. 1, pp. 1–21, 2019, doi:
https://doi.org/10.1016/j.jestch.2018.09.003.
[12] Y. Zhang, and S. Sun. “Real-time data driven monitoring and optimization
method for IoT-based sensible production process.” In: 2013 10th IEEE
International Conference on Networking, Sensing and Control (ICNSC).
IEEE, 2013.
[13] P. Ramezani, Y. Zeng, and A. Jamalipour, “Optimal resource allocation for
multiuser Internet of Things network with single wireless-powered relay.”
IEEE Internet of Things Journal, vol. 6, pp. 1–1, 2018, doi: https://doi.org/
10.1109/jiot.2018.2879373.
[14] N. Balakrishnan, A. Rajendran, D. Pelusi, and V. Ponnusamy, “Deep belief
network enhanced intrusion detection system to prevent security breach in
the Internet of Things.” Internet of Things, vol. 14, p. 100112, 2019, doi:
https://doi.org/10.1016/j.iot.2019.100112.
[15] S. Barshandeh, M. Masdari, G. Dhiman, V. Hosseini, and K. K. Singh, “A
range-free localization algorithm for IoT networks.” International Journal of
Intelligent Systems, vol. 37, no. 12, pp. 10336–10379, 2021, doi: https://doi.
org/10.1002/int.22524.
[16] A. Mohamed, Handbook of Nature-Inspired Optimization Algorithms.
Springer Nature, 2022.
[17] A. Reda and Z. C. Johanyák, “Survey on five nature-inspired optimization
algorithms.” Gradus, vol. 8, no. 1, pp. 173–183, 2021, doi: https://doi.org/
10.47833/2021.1.csc.001.
[18] A. Thongprasert and A. Jiamsanguanwong, “New product development
processes for IOT-enabled home use medical devices: a systematic review.”
Engineering Journal, vol. 25, no. 2, pp. 15–48, 2021, doi: https://doi.org/
10.4186/ej.2021.25.2.15.
[19] L. Sciullo, L. Gigli, A. Trotta, and M. D. Felice, “WoT store: managing
resources and applications on the web of things.” Internet of Things, vol. 9,
pp. 100164, 2020, doi: https://doi.org/10.1016/j.iot.2020.100164.
[20] Y. Chen, S. Hao, and H. Nazif, “IoT-enabled product development method
to support rapid manufacturing using a nature-inspired algorithm.” Journal
of Management & Organization, pp. 1–23, 2022, doi: https://doi.org/
10.1017/jmo.2022.62.
[21] Y. Chen, S. Hao, and H. Nazif, “A privacy-aware approach for managing the
energy of cloud-based IoT resources using an improved optimization algo-
rithm.” IEEE Internet of Things Journal, vol. 99, pp. 1–1, 2021, doi: https://
doi.org/10.1109/jiot.2021.3112474.
[22] P. Bujok, J. Tvrdik, and R. Polakova, “Nature-inspired algorithms in real-
world optimization problems.” MENDEL, vol. 23, no. 1, pp. 7–14, 2017, doi:
https://doi.org/10.13164/mendel.2017.1.007.
170 Nature-inspired optimization algorithms and soft computing

[23] X. Wang and Y. LI, “Solving Shubert function optimization problem by using
evolutionary algorithm.” Journal of Computer Applications, vol. 29, no. 4,
pp. 1040–1042, 2009, doi: https://doi.org/10.3724/sp.j.1087.2009.01040.
[24] A. Ghaedi, A. K. Bardsiri, and M. J. Shahbazzadeh, “Cat hunting optimi-
zation algorithm: a novel optimization algorithm.” Evolutionary
Intelligence, vol. 16, pp. 417–438, 2021, doi: https://doi.org/10.1007/
s12065-021-00668-w.
[25] M. Emami, A. Amini, and A. Emami, “Stock portfolio optimization with
using a new hybrid evolutionary algorithm based on ICA and GA: recursive-
ICA-GA (Case Study of Tehran Stock Exchange).” SSRN Electronic
Journal, 2012, doi: https://doi.org/10.2139/ssrn.2067126.
[26] Y. Gao and K. Zhu, “Hybrid PSO-Solver algorithm for solving optimization
problems.” Journal of Computer Applications, vol. 31, no. 6, pp. 1648–1651,
2012, doi: https://doi.org/10.3724/sp.j.1087.2011.01648.
[27] A. Majumder, “Termite alate optimization algorithm: a swarm-based nature
inspired algorithm for optimization problems.” Evolutionary Intelligence,
vol. 16, pp. 1–21, 2022, doi: https://doi.org/10.1007/s12065-022-00714-1.
[28] S. Yi and S. Yue, “Study of logistics distribution route based on improved
genetic algorithm and ant colony optimization algorithm.” Internet of Things
(IoT) and Engineering Applications, vol. 1, pp. 11–17, 2016, doi: https://doi.
org/10.23977/iotea.2016.11003.
[29] G. Yu and L. Kang, “New evolutionary algorithm based on girdding for
dynamic optimization problems.” Journal of Computer Applications, vol. 28,
no. 2, pp. 319–321, 2008, doi: https://doi.org/10.3724/sp.j.1087.2008.00319.
[30] G. G. Wang, S. Deb, and L. D. S. Coelho, “Earthworm optimization algo-
rithm: a bio-inspired metaheuristic algorithm for global optimization pro-
blems.” International Journal of Bio-Inspired Computation, vol. 1, no. 1,
pp. 1, 2015, doi: https://doi.org/10.1504/ijbic.2015.10004283.
[31] M. Kumar, S. Kumar, P. K. Kashyap, et al., “Green communication in the
Internet of Things: a hybrid bio-inspired intelligent approach.” International
Journal Sensors, vol. 10, no. 22, pp. 3910, 2022, https://doi.org/10.3390/
s22103910
[32] S. Arslan and S. Aslan, “A modified artificial bee colony algorithm for clas-
sification optimization.” International Journal of Bio-Inspired Computation,
vol. 1, no. 1, pp. 1, 2022, doi: https://doi.org/10.1504/ijbic.2022.10049021.
[33] X. S. Yang, “Firefly algorithm, stochastic test functions and design optimi-
sation.” International Journal of Bio-Inspired Computation, vol. 2, no. 2,
pp. 78, 2010, doi: https://doi.org/10.1504/ijbic.2010.032124.
[34] D. Devassy, J. Immanuel Johnraja, and G. J. L. Paulraj, “NBA: novel bio-
inspired algorithm for energy optimization in WSN for IoT applications.”
The Journal of Supercomputing, vol. 78, no. 14, pp. 16118–16135, 2022,
doi: https://doi.org/10.1007/s11227-022-04505-4.
[35] S. Hamrioui and P. Lorenz, “Bio inspired routing algorithm and efficient
communications within IoT.” IEEE Network, vol. 31, no. 5, pp. 74–79, 2017,
doi: https://doi.org/10.1109/mnet.2017.1600282.
 Application and challenges of optimization in IoT 171

[36] C. X. Mavromoustakis, G. Mastorakis, J. M. Batalla, and Springerlink

(Online Service, Internet of Things (IoT) in 5G Mobile Technologies. Cham:
Springer International Publishing, 2016.
[37] M. P. Raju and A. J. Laxmi, “IOT based online load forecasting using
machine learning algorithms.” Procedia Computer Science, vol. 171,
pp. 551–560, 2020, doi: https://doi.org/10.1016/j.procs.2020.04.059.
[38] S. Dash, Subhendu Kumar Pani, A. Abraham, and Y. Liang, Advanced Soft
Computing Techniques in Data Science, IoT and Cloud Computing. Springer
Nature, 2021.
[39] H. Sounni, N. El kamoun, and F. Lakrami, “Load balancing algorithm based-
SDN for the IoT applications.” Indonesian Journal of Electrical Engineering
and Computer Science, vol. 21, no. 2, pp. 1209, 2021, doi: https://doi.org/
10.11591/ijeecs.v21.i2.
[40] O. Y. Abdulhammed, “Load balancing of IoT tasks in the cloud computing by
using sparrow search algorithm.” The Journal of Supercomputing, vol. 78,
pp. 3266–3287, 2021, doi: https://doi.org/10.1007/s11227-021-03989-w.
[41] B. T. Geetha, P. Santhosh Kumar, B. Sathya Bama, S. Neelakandan, C.
Dutta, and D. Vijendra Babu, “Green energy aware and cluster based com-
munication for future load prediction in IoT.” Sustainable Energy
Technologies and Assessments, vol. 52, pp. 102244, 2022, doi: https://doi.
org/10.1016/j.seta.2022.102244.
[42] D. Sharma and S. Jain, “Optimized LOADng routing protocol parameters
using black widow optimization algorithm for IoT.” Webology, vol. 19, no.
1, pp. 535–550, 2022, doi: https://doi.org/10.14704/web/v19i1/web19038.
[43] L. Kakkar, D. Gupta, and S. Tanwar, “Comparative analysis of various
encryption algorithms used in IoT security.” In: 2021 9th International
Conference on Reliability, Infocom Technologies and Optimization (Trends
and Future Directions) (ICRITO), 2021.
[44] D. J. Hemanth and J. Anitha, “Modified Genetic Algorithm approaches for
classification of abnormal magnetic resonance brain tumour images.”
Applied Soft Computing, vol. 75, pp. 21–28, 2019, https://doi.org/10.1016/j.
asoc.2018.10.054
[45] H. Kurdi, F. Ezzat, L. Altoaimy, S. H. Ahmed, and K. Youcef-Toumi,
“MultiCuckoo: multi-cloud service composition using a cuckoo-inspired
algorithm for the Internet of Things applications.” IEEE Access, vol. 6,
pp. 56737–56749, 2018, https://doi.org/10.1109/access.2018.2872744.
[46] R. C. H, V. D and V. R. K, “Comparative analysis of applications of identity-
based crpto system in IoT.” Information Security Education Journal (ISEJ),
vol. 5, no. 2, pp. 64, 2018, doi: 10.6025/isej/2018/5/2/64-70.
[47] L. Kakkar, D. Gupta, and S. Tanwar, “Comparative analysis of various
encryption algorithms used in IoT security.” In: 2021 9th International
Conference on Reliability, Infocom Technologies and Optimization (Trends
and Future Directions) (ICRITO), 2021, doi: 10.1109/icrito51393.
2021.9596234.
172 Nature-inspired optimization algorithms and soft computing

[48] O. Speck, D. Speck, R. Horn, J. Gantner, and K. Sedlbauer, “Biomimetic

bio-inspired biomorph sustainable? An attempt to classify and clarify
biology-derived technical developments.” Bioinspiration & Biomimetics,
vol. 12, no. 1, pp. 011004, 2017, https://doi.org/10.1088/1748-3190/12/1/
011004.
[49] O. Hegazy, O. S. Soliman, and M. Salam, “A comparative study between
FPA, BA, MCS, ABC, and PSO algorithms in training and optimizing of
LSSVM for stock market prediction.” International Journal of Advanced
Computer Research, vol. 5, no. 18,pp. 35, 2015.
[50] S. Jabbar, R. Iram, A. A. Minhas, I. Shafi, S. Khalid, and M. Ahmad,
“Intelligent optimization of wireless sensor networks through bio-inspired
computing: survey and future directions.” International Journal of
Distributed Sensor Networks, vol. 9, no. 2, p. 421084, 2013, https://doi.org/
10.1155/2013/421084.
 Chapter 7
Optimization applications and implications in
biomedicines and healthcare
Niharika Kulshrestha1, Abhinav Kumar2, Vinay Kumar2
and Subodh Srivastava2

Optimization is a set of methods which provides the best possible solution for a
given problem. In this chapter, the role of optimization in healthcare systems,
medical diagnosis, biomedical informatics, biomedical image processing, ECG
classification, feature extraction and classification, and intelligent detection of
disordered systems have been discussed in detail. Approaches for predictive ana-
lytics in healthcare and innovations and technologies for smart healthcare have
been discussed using various methods. Some of the issues and challenges while
using optimization algorithms for smart healthcare and wearables have been
demonstrated in this chapter.

7.1 Introduction

Computer vision (CV) is a branch of computer technology which includes image

processing, image analysis, and image understanding [1]. Computer vision is
related to the construction of the 3D world from the observed 2D images.
Figure 7.1 shows the different branches of image processing [2] which include
digital image processing, CV, computer graphics, and artificial intelligence (AI). In
a CV, input is an image, and output is a description. When input is a description,
output is an image in computer graphics, whereas in AI, input is a description, and
output is also a description.
Figure 7.2 shows the steps to be followed in digital image processing. First, an
input image is acquired using either ultrasound or magnetic resonance imaging
(MRI) or mammography or any other imaging modalities [3] and then preproces-
sing of the input image is done, which includes the denoising of the image followed
by the image restoration [4]. Image enhancement is the next step to highlight the

1
Department of Physics, GLA University, India
2
Department of Electronics and Communication Engineering, National Institute of Technology –
Patna, India
174 Nature-inspired optimization algorithms and soft computing

Output
Input Image Description

Image Digital image processing Computer vision

Description Computer graphics Artificial intelligence

Figure 7.1 Branches that are correlated with digital image processing

Denoised Image Image

Input image
image restoration enhancement

Feature Feature
Image
selection extraction
classification

Figure 7.2 The steps to be followed in digital image processing

feature of the image and then these features are classified called features classifi-
cation [5]. These features are then selected, this step is known as feature selection
and then selected features are extracted called feature extraction to give the resul-
tant output to find the abnormality in the image [6]. This image processing is used
in the healthcare system widely for the early detection of chronic and harmful
diseases such as heart disorders [7], lung cancer, breast cancer [8], and brain cancer
[9].
These branches of digital image processing can be processed using machine
learning (ML) [10] or deep learning (DL) [11] method. ML is an important
advanced technology in health informatics. ML is a growing technology used to
mine knowledge from data, i.e., automatic learning from volumetric data. ML
implies proposing algorithms which can learn and progress over time and can be
used for predictions [12]. Various applications of ML [13] are machine vision,
biometric recognition, handwriting recognition, medical diagnosis, alignment of
biological sequences, drug design, speech recognition, text mining, natural lan-
guage processing (NLP), fault diagnostics, load forecasting, control and automa-
tion, and business intelligence. In this chapter, ML is discussed for the early
detection of disorders or diseases in the human body. It can also be used in smart
healthcare systems and technologies. The main objective of the ML system is to
determine the accuracy. It provides patients’ safety and healthcare quality [14]. DL
[15] techniques are the advanced version of ML techniques. This technique reduces
Optimization applications and implications in biomedicines and healthcare 175

the need of solving the problem manually. Deep learning allows the computer to
build complex concepts out of simpler concepts. One of the examples of a DL
model is the feed-forward deep network or multilayer perceptron (MLP) [16]. An
MLP is just a mathematical function mapping some set of input values to output
values. The function is formed by composing many simpler functions [17].
Optimization algorithm (OA) [18] is the utilization of the minimum resources
to obtain the best result. Although optimization may not seem like a task for
machine learning, optimization techniques are frequently used in machine learning
algorithms. Evolutionary computation is a branch of evolutionary biology that
creates search and optimization techniques to aid in the completion of challenging
tasks and issues. Genetic algorithm (GA) [19], evolution methods, evolutionary
programming, and genetic programming are the main branches of evolutionary
computation. These all are fundamental characteristics of the evolution process.
The objective of the optimization is to minimize the different types of constraints to
upsurge the performance of the proposed algorithm for getting the high perfor-
mance, efficacy, and accuracy of the system [20]. OA can be applied anywhere and
there are various kinds of optimization algorithms such as cuckoo search [21], ant
colony [22], honey bee [23], fruit fly [24], a particle swarm [25], random search
[26], and halving interval [27].

7.2 Role of optimization algorithms in

healthcare systems
In healthcare systems, the role of OAs is rigorously significant that includes
operational level, scheduling decisions as well as the design of healthcare plans
[28]. In developed countries, expenditures in healthcare and its system account for
a substantial portion of the gross domestic product (GDP). For instance, GDP in
France is 12.2% and Germany is 12.8%, of the UK is about 9%, and the highest in
USA is about 18% [29]. Healthcare delivery involves high cost and so efficient
distribution of healthcare is crucial. In spite of spending huge amount in healthcare,
USA is lagged behind other industrialized countries. The advantages of applying
optimization algorithms are that.
(I) It can improve the distribution and supply of healthcare providers.
(II) It can reduce in traveling of patients.
(III) It can maximize the number of facilities and healthcare access equality.
In the modern smart healthcare system; the OA imparts a key role in the data
delivery to end users. The emerging fields of analytics and machine learning, as
well as traditional health economic analysis approaches, can all connect with
optimization methodologies. Several healthcare systems need advanced technolo-
gies like the Internet of Things (IoT) [30], Big Data, Cloud Computing, and AI to
make the healthcare system intelligent and realizable [31]. As IoT assists in data
collection, Big Data gives out data analysis, Cloud Computing provides computing,
and AI enables predictive analytics. But, these all technologies need efficient
176 Nature-inspired optimization algorithms and soft computing

optimization algorithms to improve reliability in the modern healthcare system

[32]. Moreover, the optimization has a major role in modeling and solving the
healthcare operations management problems, too. For example, decision-making
problems in healthcare delivery, which could be appointment scheduling, operation
room scheduling, workforce scheduling, capacity planning, etc. [33]. Such kind of
operational management optimization could maximize time used in operation
theatre and matches nursing skills toward appropriate cases of patients for the day.
In addition to these, population health issues like matching organ donors and
recipients or developing radiation therapy protocols that limit patient harm could be
addressed clinically via optimization [34].
Decision-making is necessary when using optimization algorithms in the
delivery of healthcare, and this involves scheduling appointments, operating rooms
(OR), capacity, and manpower. A key consideration factor in healthcare delivery
systems, appointment scheduling, is made up of appointment times for each service
at which the necessary resources will be made available, such as when planning
procedures in an OR. The main challenges in designing optimal appointment
schedules are uncertainty in service time such as patient hang around time,
healthcare expertise inactive time, and over limit of the time are not specific [35].
Patients do not have complete information about their preferences, so a more rea-
listic framework has been proposed that gets automatically updated when an
appointment is done [36]. OR scheduling, often known as surgery scheduling, takes
into account resources such ORs, surgeons, and surgical equipment. OR three tiers
of scheduling decisions are made sequentially [37]. In order to plan surgeries in
ORs on a daily basis, available OR time must first be assigned. Once this is done, a
block schedule is created. This stage involves determining the number of surgeries,
their order, and scheduled start times (appointment times).
Due to the requirement for reaching high service levels, high costs, and spe-
cialized resources, capacity planning is a significant concern in the healthcare
industry. Bed capacity planning is also a type of capacity planning [38]. This is
important because the factors like unexpected arrival of patients, indefinite time of
patient’s stay, and preplanned patient’s stay can affect the availability of bed
capacity. With a specific number of “virtual beds” and the ability for “virtual
patients” to arrive, stay, and depart at rates that correspond to the actual historical
trends, a simulation model is proposed in PASCAL utilizing a three-phase simu-
lation shell (or any other patterns as required) [39]. Workforce scheduling refers to
the planning of staff schedules based on the management of the staff required for
healthcare facilities at both the operational and aggregate levels. It mainly deals
with the management of nurses involved in workforce scheduling. Direct care and
indirect care are the two types of nursing care. Indirect care refers to the time nurses
spend on other tasks while providing direct care, which is a face-to-face interaction
between nurses and patients. Nurse has a larger impact on patient health negatively
due to their lack of time toward patient’s intensive care. GA can be used for
workforce scheduling [40]. Tabu search [41] is used as the primary technique,
while knapsack and flow models are imposed in the pre-processing and post-
processing models to create a software tool [42]. Ant colony optimization has also
Optimization applications and implications in biomedicines and healthcare 177

been proposed for nurse scheduling programs [43]. A mixed integer programming
(MIP) model can also be used for workforce scheduling [44].
The primary function of an optimization algorithm is to support the location of
medical facilities, organ transplantation, vaccine development, and disease
screening. GA has been used for Human resource (HR) planning. Fuzzy Delphi is
used to identify the factor influencing HR planning. This strategy encourages 36%
of the talents and includes recognizing the competencies that could be in demand in
emergencies [45].

7.3 Optimization algorithms in medical diagnosis

AI is the invention of fast as well as advanced computer technology. In the present
time, AI has been frequently used in clinical purposes. Significant means in AI is
the concept of machine learning, proposed by Arthur Samuel in 1959, which refers
to employing algorithms to make machines learn from a vast amount of data in
order to develop new methods for data analysis and study [46]. All kinds of medical
examinations of human body are done using some kind of machines, which use
some computing tools. These computer-aided tools diagnose diseases using auto-
mated processes for higher performance and better cure of the disease. To refine the
results of these automated processes and for the early-stage detection of disease;
some optimization algorithms are used for feature extraction and selection of the
disease. Some highly used OAs are cat swarm [47], flower pollination [48], glow
worm swarm [49], genetic [19], gray wolf [50], particle swarm [51], whale [52],
lion [53], grasshopper [54], moth flame [55], crow search [56], elephant search
[57], social spider [58], ant colony [22], chimp [59], bees optimization algorithms,
and many more [60].
Swarm intelligence [61] is a characteristic of systems of non-intelligent agents
that exhibit collectively intelligent behavior despite having insufficient individual
intelligence. Ant colony optimization (ACO) [22] and particle swarm optimization
(PCO) [61] are the two main types of optimizations used in swarm intelligence.
ACO is based on social insects and PCO follows the ability of human societies to
process knowledge. Both ACO and PCO depend on the population of individuals.
ACO is derived from ants, which are obviously significant creatures that are cap-
able of finding effective solutions to issues that are well beyond their individual
colony members’ capabilities, such as the most efficient route to a food source. The
emulation of human social behavior, or the ability of human society to process
knowledge, is what drives PSO. A population of individuals having the ability to
interact with their surroundings and one another is taken into account in particle
swarm theory. Thus, interactions between individuals will affect the population
level. Although some animal societies’ collective behavior and particle systems
served as the approach’s first sources of inspiration, the algorithm’s main focus is
on its social adaptation of information.
In medical diagnosis, various features of a disease are extracted using a spe-
cified algorithm or automated process. Then to reduce the extracted features for the
178 Nature-inspired optimization algorithms and soft computing

selection of the disease; any of the OA is used. One of the most common cancers
that affect roughly one million people worldwide is skin cancer. Its early detection
can initiate early treatment and cure for the patients. Thermal exchange OA is used
to detect skin cancer at its early stages [62]. A hybrid system is proposed to identify
various diseases with an optimizing classifier parameter for the support vector
machine (SVM) [63] and MLP. A hybrid intelligent system has been proposed
considering the objectives like prediction accuracy [64], sensitivity, and specificity
[65]. Global optimization-based techniques have been used to improve the diag-
nosis of breast cancer in patients [66]. Computer-aided diagnosis (CAD) [67] is
designed with the aid of SVM where parameters are altered by optimization algo-
rithms [14]. Over a decade, computer-aided diagnostics have proven to be effective
and accurate for diagnosis. It has been suggested to use deep learning techniques
along with supervised machine learning and meta-heuristic algorithms to conduct
an accurate diagnosis [68]. For feature extraction mainly convolutional neural
network (CNN) [69] is used and for feature selection, ant colony optimization is
used [68]. Accuracy of more than 99% is obtained for the diagnosis of brain tumors
and COVID-19. Cancer [70], diabetes, and heart disease can all be detected using
an artificial neural network (ANN) that has been tuned using the directed bee
colony (DBC) algorithm [71]. For the diagnosis of diseases including Alzheimer’s
disease, brain disorders, diabetes, liver diseases, COVID-19, etc., metaheuristic
optimization algorithms have been utilized [60]. Cancer classification and auto-
matic detection from 1,000 microscopic have been proposed for biopsy images [6].
It has four essential tissues: connective, epithelial, muscular, and nervous.

7.4 Optimization algorithms in biomedical informatics

Bioinformatics is an interdisciplinary field that combines physics, chemistry,
biology, computer science, information engineering, mathematics, and statics and
develops software for understanding large and complex biological data [14]. In
essence, biomedical informatics is an application of information science to the field
of biomedicine. According to Musen and Van Bemmel [72], techniques and sys-
tems for the collection, processing, and interpretation of patient data in medical
informatics are designed and evaluated with the use of scientific knowledge.
Biomedical data are a resource for comprehending fundamental research and
helping with healthcare decisions based on evidence. The creation of contemporary
analytical approaches, such as optimization algorithms and machine learning
techniques, was motivated by the fact that it was historically very challenging to
convert complicated and vast data sets into useful knowledge. Methods like trend
reporting, data statistical modeling, and visualization techniques like association
and correlation analysis are used to evaluate historical data. Also, pay attention to
prediction and decision-making applications that construct predictive models using
a known training dataset that comprises input data attributes and response values
[73]. The GA [19], ACO [22,74], PSO [75], and other metaheuristic algorithms are
effective solutions to optimization issues. The swarm intelligence algorithm is a
Optimization applications and implications in biomedicines and healthcare 179

solution for the complex problems of bioinformatics [51]. This algorithm provides
fast and reasonably accurate solutions to such problems [25]. ACO and cuckoo
algorithms [76] are computationally intensive and too complex where as meta-
heuristic algorithm is quite simple [21]. Gradient descent optimization [77] is a
traditional minimization optimization technique that determines how the weights
are changed. Least Mean Square algorithm (LMS) [78] is the computation of
weight to minimize the mean square error (MSE) between desired and actual out-
puts. Probabilistically, the LMS algorithm’s solution and the MSE optimization’s
solution are equivalent.

7.5 Optimization algorithms in biomedical

image processing

The enormously broad topic of biomedical image processing encompasses signal

acquisition, image synthesis, image processing, image representation, and medical
diagnosis based on features retrieved from images. Some fundamental image pro-
cessing techniques include outlining, deblurring, noise cleaning, filtering, search-
ing, classical analysis, and texture analysis [79]. There are mainly two kinds of
imaging techniques, namely, radiological imaging and microscopic imaging.
Radiological imaging includes ultrasound, X-ray, MRI, nuclear, and computer
tomography (CT). Capsule endoscopy (CE) is used to detect gastrointestinal tract
lesions called ulcers. Though the time of CE is about 6–8 h and can produce 60,000
images that become a difficult task for the medical practitioner to analyze such a
large number of images for the optimization algorithm can be used such CAD tools
can be used [80]. Biomedical image segmentation is a significant constituent of
image analysis and diagnosis. The largest problem in image processing is resi-
lience, hence optimization techniques have been employed to boost the effective-
ness of earlier research on medical image segmentation. In the imaging modalities
like X-rays, CT, and mammograms, researchers are interested in a certain region of
the image such regions are extracted using image segmentation [81].

7.6 Optimization algorithms for ECG classification

Basically, the ECG assists to detect the motion of the heart rate. For the accurate
classification of various kinds of arrhythmias in the ECG, some optimization is
needed [82]. Feature extraction is the main feature in ECG classification.
Preprocessing, feature extraction, GA optimization, and classification constitute the
classification technique. The original ECG signals are preprocessed to eliminate
noise. The WPD-statistical approach is used for feature extraction to acquire ECG
characteristics. The feature dimensions are reduced, and the weights and biases of
the BPNN are optimized using GA. The term “classification” refers to the division
of ECG signals into the six categories of N, L, R, P, V, and A. Furthermore, it can
be also achieved by using a swarm optimization algorithm that combined with
classifiers to search the best-fitted classification parameters [82].
180 Nature-inspired optimization algorithms and soft computing

7.7 Feature extraction and classification

The first major decision in object recognition is to decide the set of features that is
helpful for classification [83]. The feature can be defined as the characteristics and
attributes of the object [5]. Image feature is the characteristics of an object that
helps to distinguish an object from another object. Features can be artificial or
natural. Natural features are the visual appearance of an image that is natural to the
object such as brightness and texture. Artificial features are derived from image
manipulations the examples are amplitude histograms and frequency spectra. A
feature vector is a vector that contains n measured values as:

x ¼ ½x 1 ; x 2 ; x 3 . . . x n  (7.1)

The classifier [83] takes the feature vector as input and performs the classifi-
cation. Any pattern recognition task such as classification or clustering can be done
to detect the object. The essential characteristics of good features are as follows:
(i) Robustness – the property of a feature’s invariance to operations such as
translation, rotation, noise, and illumination is called robustness.
(ii) Shift invariance – the capability to maintain its state during when shift
operations are performed.
(iii) Rotation invariance – this is the capability to maintain its state originally
stated when it is rotated.
(iv) Size invariance – this is the ability to hold steady as its size changes.
(v) Mirror shear and affine invariance – features hold true when mirroring,
shearing, and affine transformation are applied.
(vi) Occlusion invariance – the attribute of the features that do not change when
all or a portion of the item is obscured is referred to as occlusion invariance.
(vii) Discrimination – there should not be any overlapping features and the
properties should be able to tell one object apart from another.
(viii) Reliability – similar objects should have comparable values; hence, the
values should be reliable.
(ix) Independence – if two traits are statistically uncorrelated from one another,
they are said to be independent. In other words, changing the value of one
attribute does not change the value of the others.
(x) Resistance to noise – an effective feature should be impervious to noise,
artifacts, etc.
(xi) Compactness – the features must be few in number in order to be displayed
in a short space.
All features do not exhibit these properties. Hence, suitable parameters that
distinguish the object uniquely should be identified and extracted. The feature can
be classified as shown in Figure 7.3.
After image segmentation, the next step is featuring extraction. The technique
of extracting and creating features to help in object classification is known as
feature extraction [80]. This stage is crucial since the features’ quality affects how
Optimization applications and implications in biomedicines and healthcare 181

Features

Nature of
Nature features Domain Information
object

Application-
General Edge-based Region-based Space domain- Preserving
specific
features features features based features features
features

Frequency
Pixel values Structural Structural Non-preserving
domain-based
features features features features
features

Local value Global features

Global features
features

Global features

Figure 7.3 Classification of features

well they perform the categorization task. The feature vector is a vector that stores
the expanded set of features. The categories of feature extraction include
histogram-anchored characteristics like mean, median, variance, skewness, kurto-
sis, and first- and second-order moments, as well as texture features, also known as
Harlicks features, which can be smooth or rough, etc. [5]. Area, perimeter, orien-
tation, equivalent diameter, circularity, eccentricity, image curvature, and wavelet
features [84] are some examples of geometric features, also referred to as shape
features, that are used for the detection of transient changes in abnormalities like
microcalcification and spiculated masses, among others.

7.7.1 Case study 1: feature extraction in mammography

For the purpose of identifying brief alterations in anomalies such microcalcifica-
tion, spiculated masses, etc. in mammogram classification can be seen using a
feature hybrid set that includes all feature categories [85]. Mean square energy and
mean amplitude, or orientation, are recovered for the given image at three levels
of Gabor wavelet decomposition, resulting in six orientation features, and are
employed for improved categorization. In statistics, histogram, z-score, gradient
information, etc. come under the statistical feature category. Since the hybrid fea-
ture set contains both redundant and non-redundant features, feature selection must
be done after feature extraction of the spatial frequency, mean of the z-score, nor-
malized gray level variance, the mean energy of the gradient, and threshold gra-
dient. The difference between the image’s maximum and minimum gray level
values is known as the histogram range. The median of the absolute deviations
from the median of the data is known as the median absolute deviation (MAD).
The z-score, also known as the standard score, shows how many standard
deviations above or below the mean an observation or datum is. The population’s
182 Nature-inspired optimization algorithms and soft computing

mean, or z-score, is equal to the raw score’s x value plus the population’s standard
deviation. The ratio of variation to an image’s mean is known as the normalized
gray level variance (Nvar). The gradient of an image gives information about fine
edges and structures in an image. The mean energy of the gradient measures the
energy of the gradient of image. Texture characteristics are crucial in the classifi-
cation of mammograms. Haralick’s texture features have been obtained from the
gray level co-occurrence matrix (GLCM) probabilities as follows: homogeneity or
angular second moment, contrast, correlation, variance1, variance2, standard
deviation1, standard deviation2, dissimilarity measure, local homogeneity or inverse
difference moment, energy, entropy haralick’s correlation, cluster shade, cluster
prominence, sum average, sum entropy, sum variance, difference variance, differ-
ence entropy, information measure1, and information measure2. Geometric features
or geometric shape feature describe the geometric properties of the masses.
Geometric elements are crucial to medical diagnostics for breast cancer detection.

7.7.2 Features are extracted by discrete wavelet transform

In the analysis of mammograms, wavelet-based [86] features include their effective
de-correlation behavior, ease of spotting useful local and transient features, the
capability of multi-resolution analysis, and accessibility of quick and reliable
algorithms to compute the discrete wavelet transform (DWT) and its inverse.
After performing a 2D discrete on the images, the wavelet characteristics are
extracted. The main energy components of an image are measured by the group
of coefficients known as approximation coefficients. The three remaining
wavelet coefficients, abbreviated as cA, cH, cV, and cD, are referred to as
approximation and detail coefficients. It is measured as variation in intensity or
gray level along the horizontal, vertical, and diagonal axes, respectively. The
average value of gray level fluctuations in an image in the three directions of
horizontal, vertical, and diagonal is determined by adding the wavelet coeffi-
cients, which are specified as follows. The mean value of the detail coefficients
obtained at the first level of decomposition to the mean value of the sum of the
approximation coefficients obtained at the first, second, and third level wavelet
decomposition is known as the wavelet ratio.

7.7.3 Features are extracted by the Gabor filter

It is a linear filter which is used to extract local image features like edge detection.
Given that Gabor filters’ [87] representations of frequency and direction are com-
parable to those of the human visual system, they are suitable for texture repre-
sentation and discrimination. For the purpose of drawing out useful features from an
image, a collection of Gabor filters with various frequencies and orientations may be
useful. The two main important Gabor features extracted from the given mammo-
graphic images are mean square energy (Gabor Energy) and mean amplitude i.e.,
orientation. These two features have been extracted from the given images. The
Gabor energy is defined as the responses of the linear symmetric and anti-symmetric
image. The mean square energy in a window of fixed size.
Optimization applications and implications in biomedicines and healthcare 183

7.7.4 Case study II: feature extraction in speech and

pattern recognition
Speech feature extraction is in charge of converting speech signals into a
stream of feature vector coefficients that only contains the data needed to
identify a particular utterance. Typically, spectral analysis methods like Mel-
frequency cepstral coefficients [88], linear predictive coding [89], and wavelet
transforms are used to extract the feature vector of speech data. The required
phonetic features can be acquired using the feature extraction technique, which
provides good distinction, little correlation among coefficients, and is not
based on linear attributes. In a situation where speech is continuous, a frame
cannot hold information from two phonemes that follow one another. It only
offers a limited depiction because it only considers the power spectrum and
ignores the phase spectrum. Linear predictive coding (LPC) [89], a low-
dimension feature vector is used for speech recognition. The advantages of
LPC are listed as:

(I) Speech production and perceptual representation require more than

linear scales, so with the help of PLC, a strongly related feature set is
created.
(II) It is able to separate the source from the vocal tract.
(III) It is not being able to incorporate a priori knowledge for the speech signal
under test.

The features that are frequently utilized in automatic speech and pattern
recognition are MeI frequency cepstral coefficients (MFCC) [90]. MFCC feature
extraction processing steps are shown in Figure 7.4.
Mel frequency Cepstral coefficients is given as:

X
k1 p 
C ½r  ¼ 2y½k cos r ð2k þ 1Þ (7.2)
k¼0
2k

where C ½r is the MFCC coefficient and y½k  is the input signal.

1D signal Framing and FFT (Fourier

blocking Windowing
transform)

MFCC coefficients Inverse FFT MEI scale

Figure 7.4 MFCC feature extraction process in speech processing

184 Nature-inspired optimization algorithms and soft computing

7.7.5 Feature selection

Feature selection is done after feature extraction and it is an optimizing set [91]. A
statistical approach called feature selection can be used to trim the extracted fea-
tures into a manageable number of features. Feature selection is an optimization
step that removes unwanted features to create optimal feature set. It is based on the
performance study of the extracted features. The main goal of feature selection is to
choose a small set of features from which to create a pattern classifier with less
complexity and thus enhance classification performance. Figure 7.5 shows the
feature selection process after the extraction of features.
One of the most important operations in feature selection methods is evaluating
the discrimination power of the individual feature. A significant stage in classification
is feature selection because it directly influences a CAD [92] system’s overall
effectiveness. The feature selection algorithms may be categorized as filter based,
wrapper based, and embedded [93]. The wrapper-based feature selection methods are
proposed and examined to select minimum redundant and maximal relevant feature
subset from original hybrid feature set. These are cross-validated lowest miss-
classification errors of several supervised classifiers and forward sequential feature
selection based on mutual information for unsupervised GA methods.

Input

Feature extraction

Statistical test

Revised feature set

N
Apply performance
test and check if OK

Generate final feature set

End

Figure 7.5 Feature selection process

Optimization applications and implications in biomedicines and healthcare 185

7.7.6 Classification
A naive Bayes (NB) [83] classifier is based on classical Bayesian statistics and uses
all the attributes and permits them to contribute to the decisions, taking into account
the features as equally important and independent of each other, considering the
class. Naı̈ve Bayes algorithm is as follows:
 Y  
yNB ¼ arg maxq P yq P xj jyq (7.3)
j
 
where yNB signifies the class output by the naı̈ve Bayes classifier. P xj jyq is the
class-conditional probability, and P yq is the unknown probability.
K-nearest neighbor, also known as KNN, is a data-driven classifier that is
straightforward and intuitive like NB classifier [94]. The K-NN algorithm is a non-
parametric technique that makes no assumptions regarding the relationship between
characteristics ½x1 ; x2; x3 . . . xn  and class membership. The class-conditional
densities, or P xj jyq , of the feature-vector distributions are necessary for
classification.
Quantities that can have a wide range of values are ideally suited for estimation
using logistic regression models. Because classification issues are so prevalent,
statisticians have developed a method to use regression models for this problem.
The method or model that is produced is known as the logistic regression method or
model. For classification, Fisher’s Linear Discriminant [95] can also be utilized.
This n-dimensional classification challenge has been reduced to a potentially sim-
pler one-dimensional problem. Rosenblatt [96] proposed the machine perceptron
whose architecture encodes the structure of a linear discriminant function. Support
vector regression is a logical progression from classification-based techniques. All
the characteristics of SVM classifiers are retained through SVM regression analy-
sis. SVM uses an optimization process that is effective for both the linearly
separable and inseparable samples in order to get the weight vector that maximizes
the margin. SVM hinges on these two mathematical operations:
(i) Input nonlinear pattern is mapped into a high-dimensional space.
(ii) Building of the finest hyperplane for linear separating the feature vectors
revealed in step (I).
Using previously undiscovered data, the SVM criterion function of the biggest
margin offers a novel approach and ensures accurate classification. The criterion
function is the most obvious for the purpose of classification and is the number of
samples misclassified by the weight vector. Since this function is stepwise constant,
it is naturally a weak candidate for gradient search. The Perceptron algorithm
seems to be an alternative to the criterion function. Neural networks (NN) can be
used to overcome the perceptron’s limitations. The gradient approaches for mini-
mization are not applicable since the perceptron criterion function takes into
account incorrectly categorized data. The NN generally use gradient algorithms for
minimization while solving regression problems, taking into account all samples
and the minimal squared-error criterion.
186 Nature-inspired optimization algorithms and soft computing

7.8 Optimization algorithm-based intelligent detection of

heart disorders
CNN can be used to monitor cardiac rhythms and anticipate heart illnesses by
uncovering hidden links and patterns in the healthcare industry. CNNs have many
hyperparameters and many particular designs, which are expensive and make it
difficult to choose the optimal value out of all potential hyperparameter values. The
CNN-JSO approach, which uses the JSO optimization process to adjust CNN
hyperparameters, is used to predict cardiac disorders [97]. System for forecasting
the level of heart disease risk using a fuzzy genetic algorithm, where first the
dataset is preprocessed, then: (i) effective attributes are chosen using a variety of
techniques, (ii) weighted fuzzy rules are developed using GA on the basis of the
selected attributes, (iii) construct the fuzzy-based decision support using the gen-
erated fuzzy knowledge base, and (iv) predict heart disease.
The studies performed using real-world data sets demonstrate the efficacy of
the suggested novel strategy [98]. Heart disease is diagnosed using a GA-based
trained recurrent fuzzy neural network (RFNN), and the findings showed 97.78%
accuracy [99]. With feature selection, one can choose the best subset of features
and lessen the dimensionality of the data. It is suggested to use an enhanced version
of the salp swarm algorithm (SSA) to overcome feature selection issues and choose
the ideal feature subset in wrapper mode. The findings acquired over many types of
datasets [100] confirm that the wrapper feature selection mode can be applied in a
variety of expert and intelligent system applications. The present SVM classifier’s
parameters are proposed to be optimized using a hybrid optimization technique that
combines the PSO and migration-modified biogeography-based optimization
algorithms. This algorithm offers a better way to optimize the ECG signal’s para-
meters. The suggested framework improves the automatic detection of different
arrhythmias or rhythm irregularities, improving accuracy [101]. The choice of
features is made using the Chimp optimization algorithm (CHOA) [59], which
improves the classification accuracy of heart disease detection [102]. In this
method, the cardiac image’s noise is first cleaned out using a median filter. Finally,
from the cardiac picture, GLCM characteristics are retrieved. The CHOA algorithm
is used to select the best features from the extracted features. The classifier uses
these chosen attributes as input. Support vector neural network (SVNN) is
employed as the classifier in this method. The classifier assigns the image a normal
or abnormal classification. According to simulation findings, the CHOA-based
SVNN outperforms the traditional SVNN, ANN, KNN, and SVM in terms of
accuracy [59]. Log-transformation is used to pre-process medical data, converting
it to a uniform value range. The procedure of identifying important features to
classify medical data is then carried out using sparse fuzzy-c-means (FCM).
Although this sparse technique offers critical characteristics for detection and may
be used to handle high dimensional data, incorporating sparse FCM for the feature
selection process offers extra advantages for interpreting the models. The deep
belief network (DBN), which is trained using the suggested Taylor-based bird
swarm method (Taylor-BSA) for detection, is then given the chosen features. Here,
Optimization applications and implications in biomedicines and healthcare 187

the Taylor series and the bird swarm algorithm (BSA) are combined to create the
Taylor-BSA [103].

7.9 Using predictive analytics in healthcare

Increased long-term investment in developing new technologies that use artificial
intelligence and machine learning to foresee future events (potentially in real-time)
to enhance people’s health is a reflection of the present interest in predictive ana-
lytics for enhancing healthcare. Clinical prediction models, as they are sometimes
known as predictive algorithms, aid in the diagnosis and prognosis of patients who
are more likely to develop a disease. Predictive algorithms are used to provide
patient counseling and to make clinical treatment decisions based on unique patient
features rather than population averages in the era of personalized medicine.
With the growing accessibility of Big Data, medical imaging, routinely gathered
electronic health records, and national registry data [104], the rate at which new
algorithms are published doesn’t seem to be slowing down.
Any healthcare application must manage enormous amounts of data in var-
ious ways. The foundation for knowledge generation today is big data. Big data is
a word used to describe data that possess specific qualities such as volume,
velocity, value, truthfulness, and variability. Such enormous data must be saved,
handled, and analyzed in order to produce the desired outcomes. Medical data is
more difficult to interpret and anticipate results from, making them more
important to patient care. Due to its importance, it is necessary to provide
effective and higher-performing algorithms, methodologies, and tools to analyse
big data in medicine. Traditional algorithms, however, are unable to analyze such
complicated data. These types of data and analytics are highly suited for machine
learning techniques. Big data analytics is explored in important facets of
healthcare, including EHR upkeep, disease diagnosis, and patient emergency
condition prediction [105]. The finest technique for handling massive amounts of
data to perform predictive analysis or pattern identification is ML.
The healthcare business is facing challenges in key areas such as electronic record
management, data integration, and computer-assisted diagnostics and disease predic-
tions due to the need to lower healthcare costs and the shift to individualized treatment.
To deal with these issues, machine learning provides a variety of tools, methods, and
frameworks [106]. Healthcare’s use of complex predictive analytics also raises moral
and legal questions [58]. By allowing for more individualized healthcare delivery,
predictive analysis has the potential to reduce health risks at both the population and
individual levels [107]. The deployment of e-HPA must protect patient privacy, create
a team to monitor the health system, include predictive analytics in medical education,
and guarantee that electronic technologies do not supplant or obstruct physician and
patient decision-making [108]. The decision-tree method is used to forecast chronic
renal illnesses in the healthcare system [109]. Furthermore, crucial are predictive
models, which must be enhanced with new strategies [110]. A potent method for
analyzing predicted data is SVM regression analysis.
188 Nature-inspired optimization algorithms and soft computing

7.10 Optimization algorithm for smart healthcare:

innovations and technologies
A framework known as “smart healthcare” makes use of technologies like wear-
ables, the Internet of Medical Things (IoMT), sophisticated machine learning
algorithms, and wireless communication technology to link people, resources, and
organizations, provide seamless access to health records, and handle and respond to
health environment demands intelligently [111]. In poor nations, the availability of
ambulances, outdated infrastructure, and contemporary equipment make it difficult
to provide healthcare services. The government institutions are unable to take on
the thousands of competent professionals who have graduated from academic and
training institutions, which creates problems with the availability of jobs for the
trained health professionals who support health delivery. Thus, unemployment is a
disease that threatens the prosperity of any country.
In some cases, industrialized countries offer excellent incentives to get com-
petent healthcare professionals to work in their health sector, which causes an
exodus of qualified healthcare workers in quest of better opportunities. In Africa,
people frequently leave for a medical facility at the crack of dawn to escape the
stress of waiting in line at public health facilities. Although there are private
healthcare providers, one of the barriers keeping people from obtaining medical
care in the private healthcare sector is the high cost of service. Because of the poor
condition of the roads and the lack of adequate ambulance services, it is occa-
sionally necessary to carry pregnant women, elderly people, and sick people on
motorcycles and transport them to a medical facility. The system for delivering
healthcare needs to be improved. Therefore, it is crucial that African healthcare
systems are modernized. Using mobile technologies and applications to transform
the delivery of health services is one method to accomplish this. This transition is
anticipated to enhance service delivery and reduce the stress associated with
waiting in lines, particularly for senior individuals.
How can we use mobile technology and its applications to provide healthcare
services to the elderly is the fundamental topic driving this study. The elderly may
not be computer literate, but a close relative or other trusted person can request
basic healthcare services on their behalf. The “smart city” movement, a new
technological paradigm, has some potential. In wealthy nations, the idea of a “smart
city” [30] has significantly transformed society; nevertheless, emerging nations
have not yet adequately investigated the idea. In order to give healthcare treatments
in the quickest possible time, an algorithm is created to map the locations of the
necessary locations. Smart mobile healthcare system is the term used to describe
this system [112]. Transparency, easy and quick accessibility, security, efficiency,
and other benefits are provided by a smart healthcare system that combines a
blockchain data network with healthcare 4.0 operations [113,114]. Processes uti-
lized in healthcare 4.0 that make data accessible are intended to be validated using
statistical simulation-optimization techniques and algorithms. The Ethereum net-
work implements the blockchain together with related programming languages,
tools, and methodologies like solidity, web3.js, Athena, etc.
Optimization applications and implications in biomedicines and healthcare 189

The performance of blockchain-based, decentralized apps for the smart health-

care system is enhanced by this study of optimization algorithms applicable to trends
in healthcare 4.0. For the proposed system, smart contracts and their designs are also
ready to speed up the payment and trust-building processes. The outcomes of this
work simulation demonstrate that the required gas value which represents block size
and expenditure is within the current Ethereum network gas restrictions. Block uti-
lization is above 80%, which indicates that the proposed system is operational [115].
When a patient has suffered a brain injury, a neurological disorder, or a mental
illness, it is crucial to identify the patient’s behavior. In traditional healthcare, the
patient behavioral analysis presented considerably more of a challenge. The devel-
opment of smart healthcare will allow for proper analysis of patient behavior. The IoT
in healthcare is crucial for providing people with better medical facilities and sup-
porting hospitals and clinicians. The suggested system consists of several medical
devices that may gather and monitor a patient’s medical information and health data
as well as communicate with a doctor via network-connected devices, such as
mobile-based apps and sensors. Through the use of machine learning algorithms,
patient data are seen and exchanged over the Internet. In order to ascertain the
patient’s precise health status, the DBN analyses the patient’s specifics from health
data. The technology that was created demonstrated an average error rate of 0.04 and
guaranteed an accuracy of 99% while evaluating patient behavior [116].
A Cloud of Things (CoT) [117] is produced by the combination of cloud com-
puting and IoT. A new paradigm for pervasive and ubiquitous computing is created
by this combination. Yet, energy-efficient CoT architectures are necessary for reliable
CoT-based services, especially those that are very delay-sensitive, like healthcare.
The efficiency of CoT structures in smart healthcare has received a lot of attention
[118]. Smart healthcare can effectively lower the cost and risk of medical procedures,
increase the efficiency of the use of medical resources, encourage collaboration and
exchange between regions, advance telemedicine and self-service medical care, and
ultimately make personalized medical services widely available [119].

7.11 Issues and challenges in using optimization

algorithms for smart healthcare and wearables
Latency, limited computation power, interoperability, and security are some of the
challenges that exist in the IoMT-based smart healthcare systems [111]. IoT chal-
lenges include related to security, privacy, and the sensitivity of health-related data.
Electronic medical records, genetic data, and other sensitive and private informa-
tion are all found in healthcare data and should be kept private. According to pre-
dictions, 72% of malicious traffic targets patient data [120]. So, it is crucial to
defend against hackers by maintaining privacy and security, both physically and
digitally. Low security, misconfigured hardware, and network configurations are
further difficulties. Furthermore, because the nature of the data from these many
types of devices is typically heterogeneous and is frequently managed by outside
parties, it is difficult to ensure its governance, security, and privacy [121].
190 Nature-inspired optimization algorithms and soft computing

List of abbreviations
ACO Ant colony optimization
AI Artificial Intelligence
ANN Artificial neural network
BSA Bird swarm algorithm
CAD Computer-aided diagnosis
CE Capsule endoscopy
CHOA Chimp optimization algorithm
CoT Cloud of Things
CT Computer tomography
CV Computer vision
DBC Directed bee colony
DBN Deep belief network
DL Deep learning
DWT Discrete wavelet transforms
FCM Fuzzy-c-means
GDP Gross domestic product
GLCM Gray level co-occurrence matrix
HR Human resource
IoMT Internet of Medical Things
IoT Internet of Things
LPC Linear predictive coding
MAD Median absolute deviation
MFCC MeI frequency cepstral coefficients
MIP Mixed integer programming
ML Machine learning
MLP Multilayer perceptron
MRI Magnetic resonance imaging
NB Nı̈ve Bayes
NLP Natural language processing
NN Neural networks
Nvar Normalized gray level variance
OA Optimization algorithm
OR Operating room
PCO Particle swarm optimization
RFNN Recurrent fuzzy neural networks
SSA Salp swarm algorithm
SVM Support vector machine
SVNN Support vector neural network
Optimization applications and implications in biomedicines and healthcare 191

References
[1] G. Stockman and L. G. Shapiro, Computer Vision. Prentice Hall PTR, 2001.
[2] R. Szeliski, Computer Vision: Algorithms and Applications. Springer Nature,
2022.
[3] P. Kumar, S. Srivastava, and R. Srivastava, “Basic understanding of medical
imaging modalities,” in High-Performance Medical Image Processing,
Apple Academic Press, pp. 1–17.
[4] S. Srivastava, N. Sharma, R. Srivastava, and S. K. Singh, “Restoration of
digital mammographic images corrupted with quantum noise using an
adaptive total variation (TV) based nonlinear filter,” in 2012 International
Conference on Communications, Devices and Intelligent Systems (CODIS),
2012, pp. 125–128.
[5] V. P. Singh, S. Srivastava, and R. Srivastava, “Effective mammogram
classification based on center symmetric-LBP features in wavelet domain
using random forests,” Technol. Heal. Care, vol. 25, no. 4, pp. 709–727,
2017, doi: 10.3233/THC-170851.
[6] R. Kumar, R. Srivastava, and S. Srivastava, “Detection and classification of
cancer from microscopic biopsy images using clinically significant and
biologically interpretable features,” J. Med. Eng., vol. 2015, 457906, 2015.
[7] M. Nath, S. Srivastava, N. Kulshrestha, and D. Singh, “Detection and loca-
lization of S1 and S2 heart sounds by 3rd order normalized average Shannon
energy envelope algorithm,” Proc. Inst. Mech. Eng. Part H J. Eng. Med.,
vol. 235, no. 6, pp. 615–624, 2021.
[8] A. Kumar, P. Kumar, and S. Srivastava, “A skewness reformed complex
diffusion based unsharp masking for the restoration and enhancement of
Poisson noise corrupted mammograms,” Biomed. Signal Process. Control,
vol. 73, no. August 2021, p. 103421, 2022, doi: 10.1016/j.bspc.2021.103421.
[9] R. R. Kumar, A. Kumar, and S. Srivastava, “Anisotropic diffusion based
unsharp masking and crispening for denoising and enhancement of MRI
images,” in: 2020 International Conference on Emerging Frontiers in
Electrical and Electronic Technologies (ICEFEET 2020), pp. 0–5, 2020,
doi: 10.1109/ICEFEET49149.2020.9186966.
[10] T. M. Mitchell, Machine Learning, vol. 1. McGraw-Hill New York, 2007.
[11] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no.
7553, pp. 436–444, 2015.
[12] B. Nithya and V. Ilango, “Predictive analytics in health care using machine
learning tools and techniques,” in 2017 International Conference on
Intelligent Computing and Control Systems (ICICCS), 2017, pp. 492–499.
[13] J. G. Carbonell, R. S. Michalski, and T. M. Mitchell, “An overview of
machine learning,” Mach. Learn., pp. 3–23, 1983.
[14] E. V Bernstam, J. W. Smith, and T. R. Johnson, “What is biomedical
informatics?,” J. Biomed. Inform., vol. 43, no. 1, pp. 104–110, 2010.
192 Nature-inspired optimization algorithms and soft computing

[15] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. MIT press,

2016.
[16] P. Kumar, S. Srivastava, R. K. Mishra, and Y. P. Sai, “End-to-end improved
convolutional neural network model for breast cancer detection using
mammographic data,” J. Def. Model. Simul., p. 1548512920973268, 2020.
[17] L. G. Kabari and E. O. Nwachukwu, “Neural networks and decision trees for
eye diseases diagnosis,” Adv. Expert Syst., pp. 63–84, 2012.
[18] X.-S. Yang, Nature-Inspired Optimization Algorithms. Academic Press,
2020.
[19] S. Forrest, “Genetic algorithms,” ACM Comput. Surv., vol. 28, no. 1, pp. 77–
80, 1996.
[20] H. Chen and W. Ser, “Design of robust broadband beamformers with pass-
band shaping characteristics using Tikhonov regularization,” IEEE Trans.
Audio. Speech. Lang. Processing, vol. 17, no. 4, pp. 665–681, 2009.
[21] M. Mareli and B. Twala, “An adaptive Cuckoo search algorithm for opti-
misation,” Appl. Comput. informatics, vol. 14, no. 2, pp. 107–115, 2018.
[22] M. Dorigo and G. Di Caro, “Ant colony optimization: a new meta-heuristic,”
in Proceedings of the 1999 Congress on Evolutionary Computation – CEC99
(Cat. No. 99TH8406), 1999, vol. 2, pp. 1470–1477.
[23] M. Fathian, B. Amiri, and A. Maroosi, “Application of honey-bee mating
optimization algorithm on clustering,” Appl. Math. Comput., vol. 190, no. 2,
pp. 1502–1513, 2007.
[24] W.-T. Pan, “A new fruit fly optimization algorithm: taking the financial
distress model as an example,” Knowledge-Based Syst., vol. 26, pp. 69–74,
2012.
[25] Y. Zhang, S. Wang, G. Ji, et al., “A comprehensive survey on particle swarm
optimization algorithm and its applications,” Math. Probl. Eng., vol. 2015, 2015.
[26] J. Bergstra and Y. Bengio, “Random search for hyper-parameter optimiza-
tion,” J. Mach. Learn. Res., vol. 13, pp. 281–305, 2012.
[27] T. A. Jundale and R. S. Hegadi, “Skew detection and correction of devana-
gari script using interval halving method,” Commun. Comput. Inf. Sci.,
vol. 709, pp. 28–38, 2017, doi: 10.1007/978-981-10-4859-3_3.
[28] Z. A. Abdalkareem, A. Amir, M. A. Al-Betar, P. Ekhan, and A. I.
Hammouri, “Healthcare scheduling in optimization context: a review,”
Health Technol. (Berl)., vol. 11, pp. 445–469, 2021.
[29] I. Papanicolas, L. R. Woskie, and A. K. Jha, “Health care spending in the
United States and other high-income countries,” Jama, vol. 319, no. 10,
pp. 1024–1039, 2018.
[30] A. Kumar and V. Nath, “Study and design of smart embedded system for
smart city using Internet of Things,” in: V. Nath and J. Mandal (eds),
Nanoelectronics, Circuits and Communication Systems. Lecture Notes in
Electrical Engineering, vol. 511, 2019, pp 397–408, Springer, Singapore.
[31] R. K. Kodali, G. Swamy, and B. Lakshmi, “An implementation of IoT for
healthcare,” in 2015 IEEE Recent Advances in Intelligent Computational
Systems (RAICS), 2015, pp. 411–416.
Optimization applications and implications in biomedicines and healthcare 193

[32] L. M. Dang, M. J. Piran, D. Han, K. Min, and H. Moon, “A survey on

internet of things and cloud computing for healthcare,” Electronics, vol. 8,
no. 7, p. 768, 2019.
[33] S. Batun and M. A. Begen, “Optimization in healthcare delivery modeling:
Methods and applications,” in: Handbook of Healthcare Operations
Management Methods and Application, Springer, pp. 75–119, 2013.
[34] A. Reddy and D. Scheinker, “The case for mathematical optimization in
health care: building a strong foundation for artificial intelligence,” Heal.
Aff. blog, 2020.
[35] P. P. Wang, “Static and dynamic scheduling of customer arrivals to a single-
server system,” Nav. Res. Logist., vol. 40, no. 3, pp. 345–360, 1993.
[36] W.-Y. Wang and D. Gupta, “Adaptive appointment systems with patient
preferences,” Manuf. & Serv. Oper. Manag., vol. 13, no. 3, pp. 373–389,
2011.
[37] A. Jebali, A. B. H. Alouane, and P. Ladet, “Operating rooms scheduling,”
Int. J. Prod. Econ., vol. 99, no. 1–2, pp. 52–62, 2006.
[38] Y.-J. Gong, J. Zhang, H. Chung, et al., “An efficient resource allocation
scheme using particle swarm optimization,” IEEE Trans. Evol. Comput.,
vol. 16, no. 6, pp. 801–816, 2012.
[39] J. C. Ridge, S. K. Jones, M. S. Nielsen, and A. K. Shahani, “Capacity
planning for intensive care units,” Eur. J. Oper. Res., vol. 105, no. 2,
pp. 346–355, 1998.
[40] U. Aickelin and K. A. Dowsland, “An indirect genetic algorithm for a nurse-
scheduling problem,” Comput. & Oper. Res., vol. 31, no. 5, pp. 761–778,
2004.
[41] S. Topaloglu, “A shift scheduling model for employees with different
seniority levels and an application in healthcare,” Eur. J. Oper. Res.,
vol. 198, no. 3, pp. 943–957, 2009.
[42] K. A. Dowsland and J. M. Thompson, “Solving a nurse scheduling problem
with knapsacks, networks and tabu search,” J. Oper. Res. Soc., vol. 51, no. 7,
pp. 825–833, 2000.
[43] W. J. Gutjahr and M. S. Rauner, “An ACO algorithm for a dynamic regional
nurse-scheduling problem in Austria,” Comput. & Oper. Res., vol. 34, no. 3,
pp. 642–666, 2007.
[44] J. A. Castillo-Salazar, D. Landa-Silva, and R. Qu, “Workforce scheduling
and routing problems: literature survey and computational study,” Ann.
Oper. Res., vol. 239, pp. 39–67, 2016.
[45] A. Apornak, S. Raissi, A. Keramati, and K. Khalili-Damghani, “Human
resources optimization in hospital emergency using the genetic algorithm
approach,” Int. J. Healthc. Manag., vol. 14, no. 4, pp. 1441–1448, 2021.
[46] A. L. Samuel, “Some studies in machine learning using the game of
checkers,” IBM J. Res. Dev., vol. 3, no. 3, pp. 210–229, 1959.
[47] S.-C. Chu, P.-W. Tsai, and J.-S. Pan, “Cat swarm optimization,” in: PRICAI
2006: Trends in Artificial Intelligence: 9th Pacific Rim International
194 Nature-inspired optimization algorithms and soft computing

Conference on Artificial Intelligence Guilin, China, August 7–11, 2006

Proceedings 9, 2006, pp. 854–858.
[48] X.-S. Yang, “Flower pollination algorithm for global optimization,” in:
Unconventional Computation and Natural Computation: 11th International
Conference, UCNC 2012, Orléan, France, September 3–7, 2012.
Proceedings 11, 2012, pp. 240–249.
[49] C. Puttamadappa and B. D. Parameshachari, “Demand side management of
small scale loads in a smart grid using glow-worm swarm optimization
technique,” Microprocess. Microsyst., vol. 71, p. 102886, 2019.
[50] S. Mirjalili, S. M. Mirjalili, and A. Lewis, “Grey wolf optimizer,” Adv. Eng.
Softw., vol. 69, pp. 46–61, 2014.
[51] D. Wang, D. Tan, and L. Liu, “Particle swarm optimization algorithm: an
overview,” Soft Comput., vol. 22, pp. 387–408, 2018.
[52] S. Mirjalili and A. Lewis, “The whale optimization algorithm,” Adv. Eng.
Softw., vol. 95, pp. 51–67, 2016.
[53] M. Yazdani and F. Jolai, “Lion optimization algorithm (LOA): a nature-
inspired metaheuristic algorithm,” J. Comput. Des. Eng., vol. 3, no. 1,
pp. 24–36, 2016.
[54] B. Suits, The Grasshopper: Games, Life and Utopia. Broadview Press, 2014.
[55] S. Mirjalili, “Moth-flame optimization algorithm: A novel nature-inspired
heuristic paradigm,” Knowledge-based Syst., vol. 89, pp. 228–249, 2015.
[56] A. Askarzadeh, “A novel metaheuristic method for solving constrained
engineering optimization problems: crow search algorithm,” Comput.
Struct., vol. 169, pp. 1–12, 2016.
[57] G.-G. Wang, S. Deb, and L. dos S. Coelho, “Elephant herding optimization,”
in 2015 3rd International Symposium on Computational and Business
Intelligence (ISCBI), 2015, pp. 1–5.
[58] E. Cuevas, M. Cienfuegos, D. Zaldı́var, and M. Pérez-Cisneros, “A swarm
optimization algorithm inspired in the behavior of the social-spider,” Expert
Syst. Appl., vol. 40, no. 16, pp. 6374–6384, 2013.
[59] M. Pradhan, A. Bhuiyan, and B. P. Baliarsingh, “Chimp optimization
algorithm-based feature selection for cardiac image-based heart disease
diagnosis,” in: Ambient Intelligence in Health Care: Proceedings of ICAIHC
2022, Springer, 2022, pp. 61–70.
[60] S. Kaur, Y. Kumar, A. Koul, and S. Kumar Kamboj, “A systematic review
on metaheuristic optimization techniques for feature selections in disease
diagnosis: open issues and challenges,” Arch. Comput. Methods Eng.,
vol. 33, pp. 1–33, 2022.
[61] J. Kennedy and R. C. Eberhart, “A discrete binary version of the particle
swarm algorithm,” in: 1997 IEEE International Conference on Systems,
Man, and Cybernetics. Computational Cybernetics and Simulation, 1997,
vol. 5, pp. 4104–4108.
[62] L. Wei, S. X. Pan, Y. A. Nanehkaran, and V. Rajinikanth, “An optimized
method for skin cancer diagnosis using modified thermal exchange optimi-
zation algorithm,” Comput. Math. Methods Med., vol. 2021, 2021.
Optimization applications and implications in biomedicines and healthcare 195

[63] S. K. Tripathy, R. Sudhamsh, S. Srivastava, and R. Srivastava, “MuST-POS:

multiscale spatial-temporal 3D atrous-net and PCA guided OC-SVM for
crowd panic detection,” J. Intell. & Fuzzy Syst., no. Preprint, pp. 1–16, 2022.
[64] P. Kumar, A. Kumar, S. Srivastava, and Y. Padma Sai, “A novel bi-modal
extended Huber loss function based refined mask RCNN approach for
automatic multi instance detection and localization of breast cancer,” Proc.
Inst. Mech. Eng. Part H J. Eng. Med., p. 09544119221095416.
[65] M. R. Nalluri, D. S. Roy, K. Kannan, et al., “Hybrid disease diagnosis using
multiobjective optimization with evolutionary parameter optimization,” J.
Healthc. Eng., vol. 2017, Article ID 5907264, 2017.
[66] A. Bagirov, A. Rubinov, and J. Yearwood, “Using global optimization to
improve classification for medical diagnosis and prognosis.,” Top. Health
Inf. Manage., vol. 22, no. 1, pp. 65–74, 2001.
[67] S. Srivastava, N. Sharma, S. K. Singh, and R. Srivastava, “Quantitative
analysis of a general framework of a CAD tool for breast cancer detection
from mammograms,” J. Med. Imaging Heal. Informatics, vol. 4, no. 5,
pp. 654–674, 2014, doi: 10.1166/jmihi.2014.1304.
[68] Y. A. Kadhim, M. U. Khan, and A. Mishra, “Deep learning-based computer-
aided diagnosis (CAD): applications for medical image datasets,” Sensors,
vol. 22, no. 22, p. 8999, 2022.
[69] K. Jani, R. Srivastava, S. Srivastava, and A. Anand, “Computer aided med-
ical image analysis for capsule endoscopy using conventional machine
learning and deep learning,” in: 2019 7th International Conference on Smart
Computing & Communications (ICSCC), 2019, pp. 1–5.
[70] A. Saber, M. Sakr, O. M. Abo-Seida, and A. Keshk, “Automated breast
cancer detection and classification techniques – a survey,” in: 2021
International Mobile, Intelligent, and Ubiquitous Computing Conference
(MIUCC), 2021, pp. 200–207.
[71] S. Agrawal, B. Singh, R. Kumar, and N. Dey, “Machine learning for medical
diagnosis: a neural network classifier optimized via the directed bee colony
optimization algorithm,” in: U-Healthcare Monitoring Systems, Elsevier,
2019, pp. 197–215.
[72] J. H. Van Bemmel, “The structure of medical informatics: bibliography on
educational courses at the Free University, Amsterdam,” Med. Informatics,
vol. 9, no. 3–4, pp. 175–180, 1984.
[73] S. Huang, J. Zhou, Z. Wang, Q. Ling, and Y. Shen, “Biomedical informatics
with optimization and machine learning,” EURASIP Journal on
Bioinformatics and Systems Biology, vol. 2017, no. 1, pp. 1–3, 2016.
[74] M. Dorigo and T. Stützle, Ant Colony Optimization: Overview and Recent
Advances. Springer, 2019.
[75] S. Das, A. Abraham, and A. Konar, “Swarm intelligence algorithms in
bioinformatics,” Comput. Intell. Bioinforma., pp. 113–147, 2008.
[76] R. Rajabioun, “Cuckoo optimization algorithm,” Appl. Soft Comput., vol. 11,
no. 8, pp. 5508–5518, 2011.
196 Nature-inspired optimization algorithms and soft computing

[77] S. Ruder, “An overview of gradient descent optimization algorithms,” arXiv

Prepr. arXiv1609.04747, 2016.
[78] W. Liu, P. P. Pokharel, and J. C. Principe, “The kernel least-mean-square
algorithm,” IEEE Trans. Signal Process., vol. 56, no. 2, pp. 543–554, 2008.
[79] H. K. Huang, “Biomedical image processing,” Crit. Rev. Bioeng., vol. 5, no.
3, pp. 185–271, 1981.
[80] K. K. Jani, S. Srivastava, and R. Srivastava, “Computer aided diagnosis
system for ulcer detection in capsule endoscopy using optimized feature
set,” J. Intell. Fuzzy Syst., vol. 37, no. 1, pp. 1491–1498, 2019.
[81] T. Khin, K. Srujan Raju, G. R. Sinha, K. K. Khaing, and T. M. Kyi, “Review
of optimization methods of medical image segmentation,” in: Proceedings of
the Third International Conference on Computational Intelligence and
Informatics: ICCII 2018, 2020, pp. 213–218.
[82] S. Dalal and V. P. Vishwakarma, “GA based KELM optimization for ECG
classification,” Procedia Comput. Sci., vol. 167, pp. 580–588, 2020.
[83] S. Srivastava, N. Sharma, S. K. Singh, and R. Srivastava, “Design, analysis
and classifier evaluation for a CAD tool for breast cancer detection from
digital mammograms,” Int. J. Biomed. Eng. Technol., vol. 13, no. 3, pp. 270–
300, 2013.
[84] S. Srivastava, N. Sharma, S. K. Singh, and R. Srivastava, “A combined
approach for the enhancement and segmentation of mammograms using
modified fuzzy C-means method in wavelet domain,” J. Med. Phys., vol. 39,
no. 3, pp. 169–183, 2014, doi: 10.4103/0971-6203.139007.
[85] V. P. Singh, S. Srivastava, and R. Srivastava, “An efficient image retrieval
based on fusion of fast features and query image classification,” Int. J.
Rough Sets Data Anal., vol. 4, no. 1, pp. 19–37, 2017.
[86] R. K. Senapati, S. Srivastava, and P. Mankar, “RST invariant blind image
watermarking schemes based on discrete tchebichef transform and singular
value decomposition,” Arab. J. Sci. Eng., vol. 45, pp. 3331–3353, 2020.
[87] V. Kumar and S. Srivastava, “Performance analysis of reshaped Gabor filter
for removing the Rician distributed noise in brain MR images,” Proc. Inst.
Mech. Eng. Part H J. Eng. Med., vol. 236, no. 8, pp. 1216–1231, 2022.
[88] V. Tiwari, “MFCC and its applications in speaker recognition,” Int. J.
Emerg. Technol., vol. 1, no. 1, pp. 19–22, 2010.
[89] Y. Sharma and B. K. Singh, “Prediction of specific language impairment in
children using speech linear predictive coding coefficients,” in: 2020 First
International Conference on Power, Control and Computing Technologies
(ICPC2T), 2020, pp. 305–310.
[90] S. Gupta, J. Jaafar, W. F. W. Ahmad, and A. Bansal, “Feature extraction
using MFCC,” Signal Image Process. An Int. J., vol. 4, no. 4, pp. 101–108,
2013.
[91] S. Srivastava, N. Sharma, and S. K. Singh, “Empirical analysis of supervised
and unsupervised filter based feature selection methods for breast cancer
classification from digital mammograms,” Int. J. Comput. Appl., vol. 88, no.
8, 2014.
Optimization applications and implications in biomedicines and healthcare 197

[92] G. Kumar, N. Kulshrestha, and S. Srivastava, “Analysis of different fuzzy

c-means membership functions for the design of cad tool of early breast
cancer detection,” in: Proceedings of 2nd International Conference on
Advanced Computing and Software Engineering (ICACSE), 2019.
[93] M. Tahir, A. Tubaishat, F. Al-Obeidat, B. Shah, Z. Halim, and M. Waqas,
“A novel binary chaotic genetic algorithm for feature selection and its
utility in affective computing and healthcare,” Neural Comput. Appl.,
pp. 1–22, 2020.
[94] R. Kumar, R. Srivastava, and S. Srivastava, “Microscopic biopsy image
segmentation using hybrid color K-means approach,” Int. J. Comput. Vis.
Image Process., vol. 7, no. 1, pp. 79–90, 2017.
[95] D. M. Witten and R. Tibshirani, “Penalized classification using Fisher’s
linear discriminant,” J. R. Stat. Soc. Ser. B (Statistical Methodol., vol. 73,
no. 5, pp. 753–772, 2011.
[96] F. Rosenblatt, “The perceptron: a probabilistic model for information sto-
rage and organization in the brain.,” Psychol. Rev., vol. 65, no. 6, p. 386,
1958.
[97] A. A. Samir, A. R. Rashwan, K. M. Sallam, R. K. Chakrabortty, M. J. Ryan,
and A. A. Abohany, “Evolutionary algorithm-based convolutional neural
network for predicting heart diseases,” Comput. & Ind. Eng., vol. 161,
p. 107651, 2021.
[98] A. K. Paul, P. C. Shill, M. R. I. Rabin, and M. A. H. Akhand, “Genetic
algorithm based fuzzy decision support system for the diagnosis of heart
disease,” in 2016 5th International Conference on Informatics, Electronics
and Vision (ICIEV), 2016, pp. 145–150.
[99] K. Uyar and A. Ilhan, “Diagnosis of heart disease using genetic algorithm
based trained recurrent fuzzy neural networks,” Procedia Comput. Sci.,
vol. 120, pp. 588–593, 2017.
[100] M. Tubishat, N. Idris, L. Shuib, M. A. M. Abushariah, and S. Mirjalili,
“Improved Salp Swarm Algorithm based on opposition based learning and
novel local search algorithm for feature selection,” Expert Syst. Appl.,
vol. 145, p. 113122, 2020.
[101] M. Kaliappan, S. Manimegalai Govindan, and M. Kuppusamy, “Automatic
ECG analysis system with hybrid optimization algorithm based feature
selection and classifier,” J. Intell. Fuzzy Syst., no. Preprint, pp. 1–16, 2022.
[102] L. Ali, A. Rahman, A. Khan, M. Zhou, A. Javeed, and J. A. Khan, “An
automated diagnostic system for heart disease prediction based on c2 sta-
tistical model and optimally configured deep neural network,” IEEE
Access, vol. 7, pp. 34938–34945, 2019.
[103] A. M. Alhassan and W. M. N. Wan Zainon, “Taylor bird swarm algorithm
based on deep belief network for heart disease diagnosis,” Appl. Sci.,
vol. 10, no. 18, p. 6626, 2020.
[104] B. Van Calster, L. Wynants, D. Timmerman, E. W. Steyerberg, and G. S.
Collins, “Predictive analytics in health care: how can we know it works?,”
J. Am. Med. Informatics Assoc., vol. 26, no. 12, pp. 1651–1654, 2019.
198 Nature-inspired optimization algorithms and soft computing

[105] I. Goyal, A. Singh, and J. K. Saini, “Big Data in healthcare: a review,” in:
2022 1st International Conference on Informatics (ICI), 2022, pp. 232–234.
[106] S. K. Reddy, T. Krishnaveni, G. Nikitha, and E. Vijaykanth, “Diabetes
prediction using different machine learning algorithms,” in: 2021 Third
International Conference on Inventive Research in Computing Applications
(ICIRCA), 2021, pp. 1261–1265.
[107] V. Dhar, “Big data and predictive analytics in health care,” Big Data,
vol. 2, no. 3. New Rochelle, NY, pp. 113–116, 2014.
[108] R. Amarasingham, R. E. Patzer, M. Huesch, N. Q. Nguyen, and B. Xie,
“Implementing electronic health care predictive analytics: considerations
and challenges,” Health Aff., vol. 33, no. 7, pp. 1148–1154, 2014.
[109] B. Boukenze, H. Mousannif, A. Haqiq, and others, “Predictive analytics in
healthcare system using data mining techniques,” Comput Sci Inf Technol,
vol. 1, pp. 1–9, 2016.
[110] A. Ray and A. K. Chaudhuri, “Smart healthcare disease diagnosis and
patient management: innovation, improvement and skill development,”
Mach. Learn. with Appl., vol. 3, p. 100011, 2021.
[111] G. Muhammad, F. Alshehri, F. Karray, A. El Saddik, M. Alsulaiman, and
T. H. Falk, “A comprehensive survey on multimodal medical signals fusion
for smart healthcare systems,” Inf. Fusion, vol. 76, pp. 355–375, 2021.
[112] E. Freeman, I. E. Agbehadji, and R. C. Millham, “Nature-inspired search
method for location optimization of smart health care system,” in 2019
International Conference on Mechatronics, Remote Sensing, Information
Systems and Industrial Information Technologies (ICMRSISIIT), 2020,
vol. 1, pp. 1–9.
[113] G. L. Tortorella, F. S. Fogliatto, A. Mac Cawley Vergara, R. Vassolo, and
R. Sawhney, “Healthcare 4.0: trends, challenges and research directions,”
Prod. Plan. Control, vol. 31, no. 15, pp. 1245–1260, 2020.
[114] K. Ashok and S. Gopikrishnan, “Statistical analysis of remote health
monitoring based iot security models and deployments from a pragmatic
perspective,” IEEE Access, vol. 11, pp. 2621–2651, 2023.
[115] A. Kumar, R. Krishnamurthi, A. Nayyar, K. Sharma, V. Grover, and E.
Hossain, “A novel smart healthcare design, simulation, and implementation
using healthcare 4.0 processes,” IEEE Access, vol. 8, pp. 118433–118471,
2020.
[116] R. M. K. Mohamed, O. R. Shahin, N. O. Hamed, H. Y. Zahran, and M. H.
Abdellattif, “Analyzing the patient behavior for improving the medical
treatment using smart healthcare and IoT-based deep belief network,” J.
Healthc. Eng., vol. 2022, 2051642, 2022.
[117] M. M. E. Mahmoud, J. J. P. C. Rodrigues, K. Saleem, J. Al-Muhtadi, N.
Kumar, and V. Korotaev, “Towards energy-aware fog-enabled cloud of
things for healthcare,” Comput. & Electr. Eng., vol. 67, pp. 58–69, 2018.
[118] M. M. E. Mahmoud, J. J. P. C. Rodrigues, S. H. Ahmed, et al., “Enabling
technologies on cloud of things for smart healthcare,” IEEE Access, vol. 6,
pp. 31950–31967, 2018.
Optimization applications and implications in biomedicines and healthcare 199

[119] S. Tian, W. Yang, J. M. Le Grange, P. Wang, W. Huang, and Z. Ye, “Smart

healthcare: making medical care more intelligent,” Glob. Heal. J., vol. 3,
no. 3, pp. 62–65, 2019.
[120] W. Li, Y. Chai, F. Khan, et al., “A comprehensive survey on machine
learning-based big data analytics for IoT-enabled smart healthcare system,”
Mob. Networks Appl., vol. 26, pp. 234–252, 2021.
[121] P. A. H. Williams and V. McCauley, “Always connected: the security
challenges of the healthcare Internet of Things,” in: 2016 IEEE 3rd World
Forum on Internet of Things (WF-IoT), 2016, pp. 30–35.
This page intentionally left blank
 Chapter 8
Applications and challenges of optimization in
industrial automation
Devarani Devi Ningombam1

8.1 Factory digitalization

The word “automation” can be used for a lot of different things and refers to a wide
range of modern technologies that can work without constant human interaction or
supervision [1]. The first benefit that comes from this change is that people will
have less work to do. If we use an automated system or software at several points in
our workflow, we will have more time for other projects and access to more
resources. This will allow you to save time. Automation technology not only lowers
operating costs by letting fewer people make the same or more products with the
same or fewer people, but it also improves consistency and efficiency by making it
impossible to make mistakes while the machine is running.
Industrial digitization is driven in large part by the need to respond quickly to
changing consumer needs and market conditions [2]. If businesses time their pro-
duction runs to match the seasons, they will waste less and have fewer unhappy
customers. By replacing manual tasks with automated ones, you can get rid of
rework, downtime, mistakes, and bottlenecks. You can also improve procedures,
performance monitoring, and making decisions. Hence, you could save money,
time, and manpower. Using intelligent technologies like artificial intelligence (AI)
and the Industrial Internet of Things (IIoT) to automate manufacturing means
turning analog processes into digital ones, automating human activities and
knowledge into actions, making decisions for the system, and performing constant
optimization [3].
“Factory digitalization” is the process of using cutting-edge technology to
improve old ways of making things. Smart technologies, data analytics, and linked
devices are helping industrial manufacturing practices greatly improve their effi-
ciency, productivity, and accuracy as part of the industry’s ongoing digital revo-
lution. The operations, methods, and energy footprint of industries and supply
chains are all being altered as a result of digitalization. In the context of manu-
facturing, “digital transformation” is the process of replacing analog activities with

1
National Institute of Technology Patna (NITP), India
202 Nature-inspired optimization algorithms and soft computing

Figure 8.1 Evolution of factory digitalization

digital infrastructure and processes to minimize complexity, boost efficiency,

lessen the number of errors handled, and cut costs across the board [4]. The
practically definite increase in an industry’s profitability is the most persuasive
argument in favor of the widespread use of digital technology by industrial
enterprises. Product engineers can acquire a better understanding of future product
enhancements by combining numerous technologies, such as an enterprise
resource planning (ERP) system and product sensors.

8.1.1 Birth of factory automation

During the Industrial Revolution of the 1800s, automated techniques and technol-
ogies were used to increase manufacturing output (ISO 2003). This gave rise to
today’s version of automation, which continues to be in use. Around the same time,
industries began utilizing relay logic and were electrified, which is the process of
obtaining power from electricity [5].
In the 1920s, this caused the development of industrial automation to speed up
very quickly. The use of centrally located power plants and the installation and use
of new high-pressure boilers, electrical substations, and steam turbines have
increased the need for instruments and controls. Putting electricity into factories in
the 1920s made them more productive and changed how things worked on the
factory floor. Production factories started using electric motors, and as time went
on, fewer and fewer places still used steam engines to power shaft and belt drives at
the front of the line. During this time of transition, industrial production increased
by around 30 percent. This was because electric motors were substantially more
efficient than steam engines, required less maintenance, and lacked the huge fric-
tion losses generated by line shafts and belts.
The Industrial Automation Initiative started in 2006 as part of a high-tech
German strategic plan that focused the country’s research and innovation policy on
a few projects that would lead to scientific and technical breakthroughs in the next
10–15 years [6]. It uses computer, software, and Internet technologies to make its
idea of an integrated industry a reality. In order to build stronger links between
knowledge and skills, the German approach puts a lot of emphasis on collaboration
between business and science [5]. They make it possible to set up networks that
 Applications and challenges of optimization in industrial automation 203

cover the whole process of making something, turning factories into smart places.
Intelligent equipment, warehousing systems, and production facilities that provide
end-to-end integration constitute the links. Its integration includes inbound logis-
tics, production, marketing, outbound logistics, and service. Factory automation is
projected to enhance tighter collaboration between business partners such as sup-
pliers and customers and among employees, hence creating new opportunities for
all parties involved to profit from one another. Figure 8.1 depicts the evolution of
factory digitalization.
As a result of industrial digitization, we can expect to see new business models
and big opportunities. Already, we can see the beginnings of this trend. Direct
metal laser sintering (DMLS), an additive method like 3-D printing that deposits
melted metal powder layers by laser, is used by some businesses to fabricate low-
volume metal products from digital 3-D models. These components are made of a
totally dense metal that has superior mechanical qualities. The DMLS process is
intriguing because it can create intricate geometries that cannot be made using
conventional machining techniques.
Implementing a digitization plan that considers how the organization’s opera-
tions and structure will change is very important. On a cultural level, it can be hard to
accept change, and leaders and other important people may be resistant. Because of
this, it is important to set digitalization goals, create a digitalization strategy, choose
the right technological enablers, build technology leadership, train your staff, and
create a digital culture. All of these tasks should be done at the same time.

8.2 Product flow monitoring

The term “production monitoring” refers to the process of inspecting your products
on-site every single working day of your production. An inspector will look at your
manufacturing facility to make sure it meets your requirements. They will pick
random units for inspection and look for and fix any flaws they find [7].
Monitoring production gives you information about the whole manufacturing
process, makes you aware of any mistakes that might have happened at any point, and
helps you avoid delivery delays. With production monitoring, we can keep an eye on
several parameters at the spindle, machine, and factory levels at the same time.
Automated and real-time data collection makes it possible to look at the data and find
bottlenecks and other useful information. In the context of production monitoring,
tracking entails keeping tabs on information in order to analyze the health of machinery.
By analyzing production data, we can monitor and improve our processes more effec-
tively. Enterprise ERP systems have the potential to incorporate monitoring technolo-
gies that expand their scope and improve their precision across the organization.
Nowadays, it is highly motivated to automate all monitoring operations. When
we refer to automated monitoring, we mean a system of continuous monitoring. This
sort of system is intended to notify a certified operator through an alarm, dialer, or
pager in the event that a water treatment facility or water distribution system
fails during regular operation. Automation reduces expenses and time spent on
204 Nature-inspired optimization algorithms and soft computing

problem-solving, resulting in more productive teams overall. Due to the lessened

need for manual intervention in the detection of failures, automation of monitoring
results in not only improved alarms but also accelerated problem resolution.
The necessity of automated monitoring increases when a microservices
architecture-based scenario is involved [8]. Monitoring microservices can be sev-
eral orders of magnitude more complicated than monitoring traditional apps if
monitoring tasks are not automated. Due to the fact that multiple systems operating
on a range of technical platforms are connecting with one another and increasing or
decreasing the number of instances they run, it is vital that adequate monitoring
mechanisms for applications be created and not left as an option [9]. In this section,
some of the best practices for automating monitoring are discussed, as follows.

8.2.1 Creating applications with monitoring in mind from

the start
Monitoring frameworks should be built right into the codebases of projects so that
they can deliver better metrics. For instance, a great number of monitoring fra-
meworks include methods for capturing metrics, which can then be shown in a
dashboard. For example, Micrometer is a Java framework that was developed
specifically for microservices. It has the ability to capture parameters by making
use of a wide variety of various formats. Customers will not be able to see these
metrics unless the app was built on a framework that gives you access to them. For
the same reason, it may be hard to add certain monitoring features to older systems:
the code may not be easy to build to gather useful information about the software
[10]. In addition, new monitoring frameworks may necessitate updated library
versions, while obsolete systems may require prior releases of the same library to
function well. In such a situation, refactoring is the only way to enable the use of
sophisticated monitoring metrics on such antiquated systems.

8.2.2 Organize products into several categories

Product organization is vital because it allows people to specialize their skill sets to
better serve a clearly defined product. It has the ability to alleviate difficulties in
areas such as communication, product creation, and sales that a business may
encounter [11]. On the basis of the features of the products and the variables
affecting the purchase, the product can typically be classified into one of the fol-
lowing four primary groups:
(i) Convenience products: Convenience products are easy to obtain and readily
available at a variety of retailers, making them more important than product
uniqueness or special features.
(ii) Shopping products: Shopping products are not bought as often as con-
venience products, but you can still find them in certain stores or online.
They require research and comparing things like price, durability, reputation
of the brand, features, and compatibility. There is a lot of competition, and
customer or technical support after a purchase can make the time between the
customer and the company longer.
 Applications and challenges of optimization in industrial automation 205

(iii) Specialty products: Specialty products are often status or prestige purchases, so it
is important to use messaging to build brand loyalty and keep customers through
new product launches, feature updates, future directions, and brand innovations.
(iv) Unsought products: Promoting brand recognition can help customers get to
know a company’s product and image, which can build trust and keep cus-
tomers for longer. For example, a pest control company may use a catchy
musical jingle on local radio and television stations to increase the likelihood
that customers think of the company before another.
Product classification can help you understand how different products fit into
the greater context of making purchases in a certain market, which is why product
category research is so crucial. There are numerous types of products and services
currently available on the market. Due to the fact that the conditions surrounding
the purchase of each type of product are distinct, it is feasible to build marketing
strategies that are adapted to the product, the consumer, and the purchasing style in
order to increase sales.
When used on products that are already in high demand, aggressive marketing
methods that are meant for products that are not selling well may be unnecessary or
even harmful. Similarly, you may not want to invest money from a marketing
budget on a widespread advertising campaign for a highly specialized product that
is only likely to appeal to a portion of the market. Spending some time considering
product categories can help you make more efficient use of advertising resources
by providing a more focused, precise message and rooting it in the customer
behavior insights that are related to product classification. This might help you
make better use of your advertising budget.

8.2.3 Include real-time tracking technologies

The monitoring system makes sure that a quick response can be made, that data can
be kept safe and collected, and that the system as a whole is in great shape. Even if
monitoring does not solve problems, it makes computer systems more reliable and
stable in the long run. Monitoring factories enables businesses to manage and
optimize their equipment assets more efficiently. With a production monitoring
platform such as Machine Metrics, manufacturers are able to discover and optimize
equipment capacity [12]. Moreover, producers can better identify the processes that
surround machinery. It also gives information about the whole manufacturing
process, makes aware of any mistakes that might have happened at any point, and
helps to avoid delivery delays. And knowing and controlling every step of the
manufacturing process can only mean two things for your business: better products
and happy customers.

8.3 Inventory management

The Institute for Supply Management (ISM) recently conducted a survey that
revealed record-long lead times, the unavailability of vital basic supplies, rising
206 Nature-inspired optimization algorithms and soft computing

commodity prices, and difficulty transferring across industries due to the unpre-
cedented spike in client demand [13]. Depending on how production processes
work, digital transformation and the Internet of Things (IoT) could be both helpful
and harmful. Because of this, some technological advances, like AI, robotics and
drones, electric cars, and on-demand delivery, could change the way we look at the
traditional supply chain. There are a number of different ways that this could be
accomplished. Even if their long-term goal is to increase the efficiency and cost-
effectiveness of e-commerce processes, one of the most difficult aspects of
accomplishing this goal is integrating these systems and services across all of a
company’s existing supply chain activities. This is one of the components that must
be accomplished [14].
Inventory management is important for almost all business owners, but it is
even more important if we have more than one way to sell our products (or even a
physical store). Attempting to manually balance all of these variables raises the
probability of making mistakes. Not to mention the length of time required. What
if, however, it is informed that a team could use that time to focus on more
important tasks while still managing the inventory? At this stage, automated
inventory management comes into play. By using the features of a retail operating
system, we may reduce the amount of time spent on inventory management each
year while also improving the accuracy of those procedures.
Both retailers and wholesalers can keep an eye on their stock in real-time
thanks to automated inventory management systems. As a result of the systems,
workflows are simpler and more effective. By creating your retail automation with
pre-built circumstances, we will be able to focus on other crucial responsibilities
with the assurance that the inventory will be managed automatically. This will give
us an extra time to complete tasks. The majority of modern businesses, whether or
not they specialize in e-commerce, employ automated inventory management to
track and organize their stock, suppliers, and sales. Using an automated system
enables merchants to monitor stocks in real-time and make timely business-critical
choices. The main features of the automated inventory management are listed in
Figure 8.2.
For instance, if one of our products has reached the lowest possible stock level
and it is, therefore, necessary to reorder, the software that manages our inventory
will tell us automatically (or even reorder for us). Our inventory automation should
also work with other retail management systems, like your point-of-sale (POS)
software and order management system. In order to achieve real-time accuracy in
inventory management, we must keep track of sales across all channels. Only then
can we hope to achieve that objective. The main benefit of automated inventory
management is that we will know exactly which items sell better than others, which
products sell poorly, how well each product is doing, and how much profit you are
making. When we add this information to your demand projections, risk manage-
ment, cash-flow projections, and expected profit margins, we can make plans that
are more accurate and solid [15].
In the next section, the primary advantages of automated inventory manage-
ment are highlighted in Figure 8.3.
 Applications and challenges of optimization in industrial automation 207

Figure 8.2 Main features of automated inventory management

Figure 8.3 Main benefits of automated inventory management

8.3.1 Time efficiency

Traditional inventory management and control systems rely on manually analyzing
data from multiple locations, which is inefficient and time-consuming. It is prob-
able that forecasting future sales, calculating inventory levels, and generating
reports will all be inaccurate and inefficient. Integration of point-of-sale systems
and software for demand forecasting are two examples of sophisticated automated
solutions. These technologies continuously update data and produce significant
insights, resulting in the optimization of time. Reducing the number of hours spent
manually inputting data can have a favorable influence on customer retention by
minimizing inventory issues.
For example, consider all the hours spent manually updating spreadsheets with
inventory data. With an automated system, the stock levels will be precisely and
208 Nature-inspired optimization algorithms and soft computing

automatically updated across all channels. The automation engine will auto-
matically register and update the system whenever an item is sold, returned, or
received, freeing up the time to focus on what truly matters.

8.3.2 Scalability
Using cutting-edge software makes it much easier for a company to expand its
operations and improve its products. With the help of automated inventory systems,
businesses can be sure that building new warehouses will be profitable without
hiring more people or spending more money on tracking and sending purchase
orders by hand. This is made possible by automated inventory management sys-
tems. Real-time inventory data connected with other management systems ensures
the accuracy of decisions regarding the expansion of the firm.
In the end, automation has the potential to bring about a lot of scalability and
growth opportunities. Software that automates inventory management saves time,
makes it more accurate, and makes it less likely that mistakes will be made while
processing thousands of transactions per day.

8.3.3 Accuracy
Automated inventory management software provides retailers and producers with
accurate, real-time stock data. This enables the system to automatically replace
supplies in the event that they run out. In addition, by analyzing stock data patterns,
organizations may make more precise and thorough predictions and suggestions
regarding recruiting, optimal reordering points, arranging shifts to meet anticipated
demand, changing targets, and eventually increasing sales. These forecasts and
suggestions can be derived from an analysis of data trends.
When data is entered manually, there is a much higher chance of human error
occurring. What is the answer? Do away with any and all requirements for manually
entering inventory data. When we choose to automate the operations involved in
inventory management, the software will handle the administration of data entry on
its own. This includes adding, deleting, forecasting, and restocking stock in real-time.

8.3.4 Synchronization
Seventy-two percent of stock-outs were caused by mistakes in the way stores
ordered and restocked items, according to a study that looked at 600 stores in 29
counties. All of these things can hurt a company’s reputation: when demand fore-
casts are made by hand, they are often wrong, wrong orders can lead to lost sales,
and not having enough goods in the warehouse can make customers unhappy. With
automated inventory control software, stores, warehouses, and manufacturers are
always aware of what needs to be reordered. This helps them fulfill their duties
without delay, which reduces the number of errors made.
When talking to factories that use several distribution networks, the most
common problem they wish to fix is managing inventory availability. Having to
manually type in inventory availability as purchases are made, all while trying to
avoid out-of-stocks and cancellations, is not only stressful but also expensive and
 Applications and challenges of optimization in industrial automation 209

resource intensive. Errors are likely to occur when orders are entered manually
from one system into another. It is vital to regularly monitor and oversee sales
across all channels in order to have a complete understanding of the situation and
synchronize all data. All of this data can be synchronized in one central location so
that reports can be generated to prevent underselling and overselling. This will
ultimately result in cost savings and improved customer satisfaction.

8.3.5 Quality of delivery (QoD)

A study of 600 retail stores in 29 counties found that faulty in-store ordering and
restocking procedures were to blame for 72% of stock-outs. All of these issues harm
the company’s reputation. The automated inventory control software regularly noti-
fies retailers, warehouses, and manufacturers of the need for reordering. This allows
them to do their jobs without delay, thereby decreasing the number of errors.

8.4 Safety and security

The opportunities for intelligent manufacturing and industrial automation continue
to expand, evolve, and amaze. It is interesting to watch how new technical solutions
may be deployed to improve the management and productivity of factory floors.
This category includes inventions that enhance security, optimize process man-
agement, streamline edge computing, and coordinate motion.
The first step in doing the individual assignment is to stop people or things
from going into areas that could be dangerous. Safety laser scanners, light curtains,
and safety mats are all important parts of the process of turning off devices auto-
matically and finding intruders without human intervention. Safety systems assist
in ensuring people, plants, and equipment keep functioning safely. Improved pro-
cess control could introduce higher levels of throughput and eliminate disruptions.
Terminal management systems include any form of application in a terminal sto-
rage market to combine your operations [16].
Safety through EtherNet/IP protocol-based automation safety solutions boosts
diagnostic data accessibility, boosts productivity, and reduces wiring complexity,
and the connected enterprise (CE) is responsible for providing the network con-
nectivity. This new perspective on safety not only makes it simpler to keep people
and processes risk-free but also significantly increases overall productivity [17].
Intelligent safety devices provide enhanced visibility into operations and aid users
in developing a deeper comprehension of process states, environmental variables,
and other factors that affect productivity and safety.
Moreover, in an effort to save expenses, a number of organizations have turned
to automation. While the system’s increased efficiency is undeniably persuasive, it
is the system’s enhanced security that establishes the argument for its imple-
mentation. Here are several examples:
● Processes and operations: With automation, the number of operational injuries
can be kept to a minimum since workers no longer have to do tasks that could
be dangerous.
210 Nature-inspired optimization algorithms and soft computing

● Actions to take in an emergency: In an emergency, automated systems can

provide real-time monitoring, which lets people act more quickly. Problems
that were formerly thought to be inevitable can now be solved before they
even arise.
● Risk management: With the help of safety automation software, businesses can
quickly and easily change safety procedures and come up with a plan for how
to handle an accident. Furthermore, it can be accomplished in a timely and
efficient fashion that causes minimal interference with ongoing activities.
Some of the main benefits of the safety and security of factory automation are:
● When automation is used, it often increases efficiency because it makes things
go faster, makes better products, and reduces waste. Maintaining your system
may require an initial investment, but it will be considerably cheaper in the
long run than keeping up with your existing costs.
● The anticipated annual expenditures connected with occupational injuries and
illnesses are decreased by automation.
● As a result of having more time to focus on intellectually interesting work,
employees are happier now that the majority of manual labor has been auto-
mated. In addition to expressing care for their safety, removing personnel from
a potentially hazardous situation indicates the company’s dedication to
that asset.
● Advantage automation makes operations more secure and efficient, which
increases productivity and gives businesses an edge in the market.
Steps to be taken to ensure safety and security in optimizing the automation
industry are shown in Figure 8.4.
● Training employees: Automated systems can be helpful, but only when they
are effectively used. Training staff is therefore essential. One of the many
benefits of training is the ability to identify and avoid potentially hazardous
situations.

Figure 8.4 Steps to be taken to ensure safety and security in optimizing the
automation industry
 Applications and challenges of optimization in industrial automation 211

● Keeping to a regular schedule: Because long-term use can cause wear and tear,
it is important to do regular maintenance checks on automated systems. The
return on investment for your system may also improve.
● Risk analysis: Before doing maintenance on a system, you need to know
everything there is to know about its risk profile. Knowing about safeguards
and proper procedures can help you save a significant amount of money. To
make progress, all involved parties must have a shared understanding of the
situation.
● Lockout/tagout: When doing maintenance, it is best to follow the Lockout/
Tagout procedures for controlling dangerous materials. Locking and tagging
involve isolating energy sources that could be dangerous and could lead to the
sudden release of stored mechanical energy. It is advised that this method be
utilized as frequently as feasible; but, due to the need for a power source for
programming and software upgrades, this is not always practicable.
● Realizing obligatory legal standards: Keep in mind that there are regulations
governing the standards that must be met by an organization in terms of worker
safety and maintenance practices. You should stay current with any mod-
ifications that may occur.

8.5 Quality control

In the modern business world, there has always been a need to make and deliver
products quickly and well. Long delays can cause money to be lost and waste
resources that could have been used to get new business [18]. Liberalization and
digitalization have changed how companies operate by letting them sell their
goods in many different countries. In response, there is a lot more competition
between brands, and companies are being held to the highest standards of
customer satisfaction. In order to be viable in today’s application world,
industries must rely on experienced people, sophisticated tools, and dependable
procedures that allow the company to deliver more commercial value in less
time [19]. To execute slightly elevated development services in a timely and
trouble-free manner, a modern product testing approach that makes efficient use
of test automation capabilities is expected. A “digital technology known as
automated quality control” helps to increase the quality of inspections by
helping to reduce the number of mistakes made during the workflow. Businesses
are able to increase their performance and get their items to market more
quickly without sacrificing product quality if they eliminate the need for manual
proofreading.
Automatic data quality control technology makes it possible for businesses to
get products to market faster without lowering the quality of those products. These
technologies accelerate the entire process of quality control and minimize the
amount of time spent on proofreading by automating manual tasks that, without
automation, would take hours to complete. In turn, inefficiencies like fatigue and
human error can be minimized, allowing for the prospect of faster and more
212 Nature-inspired optimization algorithms and soft computing

accurate error detection [20]. Controlling quality guarantees that consistency and
high standards of business practices may be maintained across the many production
processes that are carried out on a daily basis at your organization. Moreover,
automatic quality control at multiple revision points can help you find deviations as
soon as they happen, so you can fix them before the number of them gets out of
hand. Your business will be able to make the most of its resources and the pro-
duction of its employees if it makes use of the advantages offered by quality con-
trol. These advantages include the following:
● saves energy spent searching for faults in content;
● in comparison to hours spent manually reviewing files, it inspects files in
seconds;
● it easily detects all types of faults at any point in the process;
● achieve faster turnaround times by expediting approvals and decreasing quality
delays;
● can compare two files side by side to check whether there are any differences
between them;
● can find out which of your company’s touchpoints are prone to errors and
eliminate them;
● it can make use of the time that you would have spent otherwise editing in
order to bring jobs to market more quickly.
For optimizing manufacturing automation while retaining the quality of the
data, we should focus on the following factors as shown in Figure 8.5. Moreover,

Figure 8.5 Data quality factors

 Applications and challenges of optimization in industrial automation 213

here are five tips that will help you improve your quality assurance process and use
the best methods and procedures in your company.
(i) Establish a work environment in which employees are encouraged to look for
methods to enhance the service quality of the organization. A way to ensure
that quality management remains at the forefront of employees’ minds is to
cultivate an environment that promotes continuous improvement in this area.
Giving individuals extra money when they exhibit a commitment to quality
is one approach to reinforcing this attitude in the workplace.
(ii) Keep the machinery and automated software in good working order by
scheduling frequent servicing. Preventative maintenance of machinery is
essential for reliable production and worker security. When preventive
maintenance is neglected, faulty plastic parts are produced, which leads to
the breakdown of the device.
(iii) Develop a comprehensive training program. The foundation of great quality
management is a well-organized training program. During staff orientation,
quality management and its significance should be introduced. Continuous
training and the introduction of new processes and procedures are additional
chances to emphasize adherence to quality standards.
(iv) Establish a comprehensive overview for verifying product quality. Accurate
product design relies heavily on the results of the quality inspection phase of
production. A comprehensive inspection procedure is essential since it is
frequently the last chance for a manufacturer to identify any design problems
before a product is shipped.
(v) Schedule regular audits of the business’s processes. Internal audits of the
supply chain can be used to assess the efficiency of the operation and whe-
ther or not all regulations are being followed. Component manufacturers can
improve customer satisfaction and get ready for external safety and com-
pliance evaluations by performing internal audits.

8.6 Packaging optimization

“Package optimization” refers to the use of intelligent packaging to secure, iden-

tify, and protect goods throughout their delivery from a factory, warehouse, or
distribution center to a client. Automation is a significant benefit that companies
have in today’s competitive market, and it is often one of the most effective
methods for lowering packaging [21]. To stay ahead of the competition in the face
of rising demand and escalating customer expectations, such as the expectation of
one- or two-day delivery, firms will need new resources and strategies. The quality
of the product’s packaging affects how long it lasts, how happy the end user is with
it, and how it affects the environment. To optimize the packing of a product from
top to bottom, one must consider not just one, but all of the various containers that
will be utilized during shipment. Packing can be classified into primary, secondary,
and tertiary. Primary packaging compartmentalizes the goods well and offers a
positive experience for the user. Display packaging is intended to use as little retail
214 Nature-inspired optimization algorithms and soft computing

space as feasible, whereas, secondary and tertiary packaging are the extra layers of
protection that make it possible to ship a product anywhere. Several aspects must
be considered while maximizing the packaging, which includes:
(i) Choosing the right materials for packaging: It is necessary to test the product
to see how fragile it is and if the materials will protect it enough during
shipping. Its primary package should facilitate an appealing opening
experience for consumers. By selecting the appropriate secondary and ter-
tiary packing, protection from potential shipping-related handling and
environmental problems can be ensured.
(ii) Appropriate package design: The correct package design, employing the correct
quantity and type of packing materials, is a crucial component that can have a
variety of effects on your supply chain. Non-standard package forms that attract
consumers’ attention on store shelves may not be the most space-efficient way
to package your product in order to maximize space in trucks and warehouses.
(iii) Choosing the right amount of packaging has a big impact on your supply chain
and the environment. Underpackaging results in damaged products, which
increase your costs through product returns and waste. In addition to introducing
an excessive amount of garbage into the environment, excessive packaging also
results in greater packing costs and waste. In addition to causing stress on the
supply chain, overpackaging also necessitates extra room for storage and transit.
(iv) Product count optimization: If we are bundling many things for shipping
purposes, we may be able to reduce costs and increase productivity by ree-
valuating how we mix the products into packs and pallets.

When companies work hard to improve their packaging designs, customer

satisfaction goes up because customers are impressed by how quickly their pro-
ducts arrive and how clean they are. If the company sends fragile goods to custo-
mers all over the world, customers should use packaging that can absorb shock,
resist water, and keep people from opening it. Using an optimized packaging
technique can help save money in the long run. Any organization can save money
on shipping costs and labor by redesigning the packaging to make the most efficient
use of available space in trucks and storage facilities. The benefits of optimizing
packaging automation are discussed as follows:

(a) The number of packages from the packaging bridge can be significantly
boosted by integrating automated procedures to supplement human labor. To
meet the demands of single-line and multi-line orders, third-party logistics
providers might use techniques that enable them to manufacture up to 1,000
unique packages every hour. These configurations can alleviate the workload
of staff and help a corporation react to demand variations. This can be a major
source of competitive advantage as the e-commerce business evolves and
client expectations increase.
(b) Automation is a cost-effective way for any merchant or logistics provider to
improve operations and make them better fit the current market. Typically,
businesses benefit most from solutions that help them avoid bottlenecks and
 Applications and challenges of optimization in industrial automation 215

save on labor expenses. Depending on the business’s size, packaging needs,

and the disruption a new solution may cause, systems that manage the entire
packaging process or handle a specific task may be beneficial.
In any case, the success of a business will depend on how well a new
automated technology or solution is put into place. Best practices will help a
business integrate new procedures with the equipment and processes already
in place, which will save time and money. So, automated packaging can have
a big and good effect on business, and companies that want to get ahead of the
competition should strongly consider using it.

8.7 Logistics and supply chain optimization

The process of making sure that orders are filled on time and in full (OTIF) is called
logistics optimization. This is accomplished through streamlined logistics and
distribution. The main goal of supply chain optimization is to make supply net-
works work better and more efficiently. This enables companies to deliver custo-
mers precisely what they want, when they want it, while also being lucrative and
sustainable [22]. A company can do end-to-end quality control with an optimized
supply chain, from getting the raw materials to delivering the finished products to
customers. With this level of control, a company can cut down on waste and better
manage key performance indicators (KPIs), which leads to better business results.
People usually think of the logistics operation and the supply chain as functions
that help with other business tasks. But this view is slowly changing, and senior
management is starting to see that the supply chain is a crucial part of the organi-
zation’s ability to stay competitive. This is because the quality of the product, the
time it takes to get it to the customer, and the cost are all better when the supply chain
is streamlined and planned. In many markets, both now and in the future, customers

Inventory

Logistic Procurement
network and sourcing
Supply chain
optimization issues

Sales forecast Transportation

Figure 8.6 Supply chain optimization issues

216 Nature-inspired optimization algorithms and soft computing

want this [23–26]. So, every industry will have to make improving their supply chain
a top priority if they want to reach operational excellence. Corporations must make
the best use of their industrial and logistics capabilities, optimize and streamline
flows throughout the supply chain, and get the most out of their resources if they
want to keep their stock low while also being more responsive and reducing
delivery times. Figure 8.6 depicts the issues faced during the optimization of supply
chain.
While we are trying to optimize the supply chain, the primary challenges that
we focus on are as follows:
(a) Logistics network design
(b) Procurement and sourcing optimization
(c) Inventory optimization
(d) Transportation planning and optimization
(e) Sales forecasting

References
[1] Nitzan, D. and Rosen, C. A. Programmable industrial automation. IEEE
Transactions on Computers, C-25(12), 1259–1270, 1976.
[2] Atack, J., Margo, R., and Rhode, P. Automation of manufacturing in the late
nineteenth century: the hand and machine labor study. Journal of Economic
Perspectives, 33, 51–70, 2019. Doi:10.1257/jep.33.2.51.
[3] Wollschlaeger, M., Sauter, T., and Jasperneite, J. The future of industrial
communication: automation networks in the era of the internet of things and
industry 4.0. IEEE Industrial Electronics Magazine, 11.1, 17–27, 2017.
[4] Lu, Y., Xu, X., and Wang, L. Smart manufacturing process and system
automation – a critical review of the standards and envisioned scenarios.
Journal of Manufacturing Systems, 56, 312, 2020.
[5] Kagermann, H., Wahlster, W., and Helbig, J. Recommendations for imple-
menting the strategic initiative INDUSTRIE 4.0. Final Report of the
Industrie 4.0, 2013, p. 82.
[6] Mabkhot, M. M., Al-Ahmari, A. M., Salah, B., and Alkhalefah, H.
Requirements of the smart factory system: a survey and perspective.
Machines, 6.2, 23, 2018.
[7] Edrington, B., Zhao B., Hansel A., Mori M., and Fujishima M. Machine
monitoring system based on MTConnect technology. Procedia CIRP, 22,
92–97, 2014.
[8] Wang, L., Xu, S., Qiu, J., et al. Automatic monitoring system in underground
engineering construction: review and prospect. Advances in Civil Engineering,
5, 1–16, 2020.
[9] Doukas, C., Chantzis, D., Stavropoulos, P., Papacharalampopoulos, A., and
Chryssolouris, G.. Monitoring and control of manufacturing processes: a
review. Procedia CIRP, 8, 421–425, 2013. Doi:10.1016/j.procir.2013.06.127.
 Applications and challenges of optimization in industrial automation 217

[10] Nour, M. and Hussain, M. A review of the real-time monitoring of fluid-

properties in tubular architectures for industrial applications. Sensors, 20
(14), 3907, 2020. https://doi.org/10.3390/s20143907.
[11] Cheng, C.-Y., Pourhejazy, P., Hung, C.-Y., and Yuangyai, C. Smart mon-
itoring of manufacturing systems for automated decision-making: a multi-
method framework. Sensors, 21, 6860, 2021. https://doi.org/10.3390/
s21206860
[12] Sridharan, M., Devi, R., Dharshini, C.S., and Bhavadarani, M. IoT based
performance monitoring and control in counter flow double pipe heat
exchanger. Internet of Things, 5, 34–40, 2019.
[13] Kidonge, K.K., Automation of Inventory Management Process, Project
Report, submitted to Makerere University (Uganda), 2006. Obtained from:
https://silo.tips/download/automation-of-inventory-management-process
[14] Mascarenhas, M., Lamani, A. Matkar, C. Dessai, A.R., and Kotharkar. A.
An automated inventory management system. International Journal of
Computer Applications, 176, 21–23, 2020.
[15] Eswari, P. and Praveena, S.M. Optimization of Efficacy in Industrial
Automation Control and Monitoring System using IIOT. 9001. ISO, ISO.
“14649-1, 2008. Industrial automation systems and integration – Physical
device control – Data model for computerized numerical controllers – Part 1:
Overview and fundamental principles.” International Organization for
Standardization, 2003.
[16] Macher, G., Armengaud, E., Brenner, E., and Kreiner, C. A review of threat
analysis and risk assessment methods in the automotive context.
International Conference on Computer Safety, Reliability, and Security,
pp. 130–141, 2016. Doi:10.1007/978-3-319-45477-1_11.
[17] Kriaa, S., Bouissou, M., Piètre-Cambacedes, L., and Halgand, Y. A survey
of approaches combining safety and security for industrial control systems.
Reliability Engineering & System Safety, 139, 156–178, 2015. Doi:10.1016/
j.ress.2015.02.008.
[18] Winkler, D. and Biffl, S. Improving quality assurance in automation systems
development projects. In M. Savsar (ed.), Quality Assurance and
Management. InTech, pp. 379–398, 2012. Doi:10.5772/33487.
[19] Buzatu, C. and Orzan, I.A. Product quality control optimization for selection
of measurement and control devices. In Chiru, A. and Ispas, N. (eds.),
CONAT 2016 International Congress of Automotive and Transport
Engineering. CONAT 2016. Springer, Cham, 2017. https://doi.org/10.1007/
978-3-319-45447-4_69.
[20] Lönnroth, V. Improving Quality Control in Automation Projects Using
Simulation Systems (Dissertation), 2021. Retrieved from http://urn.kb.se/
resolve?urn=urn:nbn:se:kth:diva-293928
[21] Perez-Vidal, C., Gracia, L., Paco, J., Wirkus, M., Azorin, J., and Fernández,
J. Automation of product packaging for industrial applications. International
Journal of Computer Integrated Manufacturing, 31, 1–9, 2017. Doi:10.1080/
0951192X.2017.1369165.
218 Nature-inspired optimization algorithms and soft computing

[22] Sathyamyla, K. and Bengt L. Modeling and optimization of synchronous

behavior for packaging machines, IFAC Proceedings, 47(3), 1684–1691,
2014.
[23] Braccesi, L. Monsignori, M., and Nesi, Paolo. Monitoring and optimizing
industrial production processes. In Proceedings of the IEEE International
Conference on Engineering of Complex Computer Systems, ICECCS,
pp. 213–222, 2004. Doi:10.1109/ICECCS.2004.1310920.
[24] Kusiak, A. Smart manufacturing. International Journal of Production
Research, 56(1–2), 508–517, 2018.
[25] Plósz, S., Schmittner, C., and Varga, P. Combining safety and security
analysis for industrial collaborative automation systems. In Proceedings of
the International Conference on Computer Safety, Reliability, and Security,
pp. 187–198, 2017. Doi:10.1007/978-3-319-66284-8_16.
[26] Xu, X. Machine Tool 4.0 for the new era of manufacturing. The
International Journal of Advanced Manufacturing Technology, 92, 1893–
1900, 2017.
 Chapter 9
Expectations from modern evolutionary
approaches for image processing
R. Srikanth1 and V. Rajagopal2

Nature-inspired optimization algorithms (NIOAs) are a class of algorithms that are

designed to solve complex optimization problems by emulating the natural pro-
cesses of various biological and physical systems. These algorithms have been
extensively studied and applied in various fields, including engineering, computer
science, physics, economics, and biology, among others.
In recent years, there has been an increasing interest in the application aspects
of NIOAs, as researchers and practitioners seek to harness their power and potential
to solve real-world problems. This is due to the fact that NIOAs have several
advantages over traditional optimization algorithms, including their ability to
handle non-linear and non-convex problems, their ability to explore large solution
spaces efficiently, and their ability to adapt to dynamic environments.
There are several types of NIOAs, including genetic algorithms, particle
swarm optimization, ant colony optimization, and many others. Each of these
algorithms is inspired by a different natural process, and they differ in terms of their
implementation, strengths, and weaknesses. Applications of NIOAs are diverse and
range from engineering design problems, such as structural optimization and con-
trol system design, to data mining and machine learning applications, such as
clustering and classification problems. NIOAs have also been successfully applied
in financial engineering, robotics, transportation, and energy management, among
other fields.
Despite their potential, NIOAs are not a panacea for all optimization problems,
and their performance can be highly dependent on the problem characteristics and
the algorithm parameters. Therefore, researchers and practitioners need to carefully
evaluate and tune NIOAs to ensure their effectiveness and efficiency. In summary,
NIOAs are a promising class of optimization algorithms that have shown great
potential in solving complex real-world problems. With ongoing research and

1
Department of Electronics & Communication Engineering, Kakatiya Institute of Technology and
Science, India
2
Department of Electrical & Electronics Engineering, Kakatiya Institute of Technology and Science,
India
220 Nature-inspired optimization algorithms and soft computing

development, NIOAs are likely to become even more powerful and versatile,
paving the way for new applications and solutions in various fields.
Types of nature-inspired optimization algorithms
The list of nature-inspired optimization algorithms along with their descrip-
tions and acronyms:

1. Particle swarm optimization (PSO): PSO is a computational method that

uses the collective intelligence of a group of particles to move through the
search space in search of the best solution. The particles exchange informa-
tion with one another and adjust their positions and velocities to find the
global optimum. PSO has been used successfully to solve a variety of opti-
mization problems.
2. Genetic algorithm (GA): GA is a search and optimization algorithm inspired
by the natural selection process. It starts with a population of candidate
solutions, each encoded as a string of parameters. These candidate solutions
evolved over successive generations by applying genetic operators such as
mutation, crossover, and selection. The fittest individuals from each genera-
tion are selected to produce the next generation, eventually leading to an
optimal solution. GA is widely used in optimization problems, machine
learning, and artificial intelligence.
3. Ant colony optimization (ACO): ACO is a population-based optimization
algorithm that is inspired by the behavior of ant colonies. In ACO, a popu-
lation of artificial ants moves through a search space to find the optimal
solution. The ants deposit pheromone trails to communicate with each other
and adjust their movement based on the pheromone levels. Ants deposit
pheromones on their path to a food source, which other ants can follow to
reach the same source. ACO mimics this behavior by depositing virtual
pheromones on the search space to guide the search toward the optimal
solution.
4. Artificial bee colony algorithm (ABC): ABC is a population-based opti-
mization algorithm inspired by honeybee colony behavior. A population of
artificial bees moves through a search space to find the best solution in ABC.
Bees communicate with one another and adjust their movements based on the
quality of food sources.
5. Firefly algorithm (FA): FA is a population-based optimization algorithm
inspired by the flashing behavior of fireflies. In FA, a population of artificial
fireflies moves through a search space in search of the best solution. Fireflies
use their light to attract potential mates, and their flashing patterns are
influenced by the flashing patterns of nearby fireflies. FA mimics this beha-
vior by adjusting the brightness and position of the fireflies to optimize the
objective function.
6. Cuckoo search algorithm (CS): CS is a population-based optimization
algorithm that is inspired by the brood parasitism behavior of some bird
species. In CS, a population of artificial cuckoos lays eggs in other cuckoos’
 Expectations from modern evolutionary approaches for image processing 221

nests to find the optimal solution. The cuckoos adjust the quality of their eggs
and the host cuckoo discards eggs with lower quality.
7. Differential evolution (DE): DE is a population-based optimization algorithm
that is inspired by the process of natural selection. In DE, a population of
solutions is evolved through differential mutation and crossover operations.
The fittest individuals are selected to create the next generation of solutions.
8. Genetic algorithm (GA): GA is an optimization algorithm inspired by the
process of natural selection. It uses a population of candidate solutions to
evolve better solutions through selection, crossover, and mutation.
9. Ant colony optimization (ACO): ACO is a metaheuristic algorithm
based on ant behavior that uses pheromone trails to determine the shortest
path between a food source and the nest. It is frequently employed in the
way to solve of optimization problems such as the traveling salesman
problem.
10. Harmony search algorithm (HSA): HSA [1] is an optimization algorithm
inspired by the process of music improvisation. It generates new solutions by
improvising new melodies, with the aim of finding the optimal solution. It is a
metaheuristic optimization algorithm inspired by the process of musical
improvisation. The algorithm was developed by Geem et al. in 2001 and has
since been applied to various optimization problems, including engineering
design, water resource management, and image processing. In HSA, a set of
decision variables is represented as musical improvisation, and the optimi-
zation process is modeled as a process of generating better improvisations.
The improvisations are evaluated using an objective function, and the better
ones are used to create new improvisations through a process of improvisation
memory, pitch adjustment, and harmony memory. Improvisation memory
involves storing the best improvisations encountered so far, while pitch
adjustment involves changing the decision variables of an improvisation.
Harmony memory involves selecting the best improvisations from the
improvisation memory and using them to generate new improvisations. HSA
has been shown to be effective in solving a wide range of optimization pro-
blems, including function optimization, parameter tuning, and feature selec-
tion. It has several advantages over other optimization algorithms, such as its
ability to handle discrete, continuous, and mixed-variable problems, and its
fast convergence rate.
11. Differential evolution (DE): DE is an optimization algorithm that uses the
difference between two solutions in the population to generate a new candi-
date solution. It is based on the principles of evolution and natural selection.
12. Cuckoo search (CS): CS is an optimization algorithm inspired by the beha-
vior of cuckoo birds. It uses a population of cuckoos to lay eggs in host nests,
with the aim of finding the optimal solution.
13. Glowworm swarm optimization (GSO): GSO is an optimization algorithm
inspired by the behavior of glowworms. It uses the bioluminescent behavior
of glowworms to search for the optimal solution.
222 Nature-inspired optimization algorithms and soft computing

14. Cat swarm optimization (CSO): CSO is an optimization algorithm inspired

by the behavior of cats. It uses the social behavior of cats to search for the
optimal solution.
15. Bat algorithm (BA): BA is an optimization algorithm inspired by the echo-
location behavior of bats. It uses the frequency and loudness of bat calls to
find the optimal solution.
16. Whale optimization algorithm (WOA): WOA is an optimization algorithm
inspired by the hunting behavior of humpback whales. It uses the bubble-net
hunting behavior of whales to search for the optimal solution.
17. Gray wolf optimizer (GWO): GWO is a metaheuristic algorithm inspired by
humpback whale hunting behavior. It enhances a given objective function
using search agents and is commonly used for optimization problems like
function optimization, data clustering, and feature selection.
18. Grasshopper optimization algorithm (GOA): GOA is an optimization
algorithm inspired by the jumping behavior of grasshoppers. It uses the
jumping and searching behavior of grasshoppers to find the optimal solution.
19. Flower pollination algorithm (FPA): FPA is an optimization algorithm
inspired by the pollination behavior of flowers. It uses the pollination beha-
vior of flowers to find the optimal solution.
20. Moth flame optimization (MFO): MFO is an optimization algorithm
inspired by the navigation behavior of moths. It uses the natural behavior of
moths to move toward the optimal solution.
21. Dragonfly algorithm (DA): DA is an optimization algorithm inspired by the
hunting behavior of dragonflies. It uses the swarming behavior of dragonflies
to find the optimal solution.
22. Artificial ecosystem-based optimization (AEBO): AEBO is an optimiza-
tion algorithm inspired by the principles of ecological systems. It uses the
principles of co-evolution, symbiosis, and ecological niches to find the opti-
mal solution.
23. Monkey algorithm (MA): MA is a nature-inspired optimization algorithm that
is based on the foraging behavior of monkeys. The algorithm was proposed in
2012 by Mirjalili and Lewis to solve complex optimization problems. The MA
algorithm is designed to mimic the behavior of monkeys as they search for food
in the forest. Monkeys use a combination of random exploration and memory
to find food sources. They remember locations where they have found food in
the past and use this information to guide their search for new food sources.
These are just a few examples of natural-inspired optimization algorithms, but
there are several others in the literature.

9.1 Application domains of nature-inspired optimization

algorithms
Nature-inspired optimization algorithms [2] are a type of optimization technique that
resembles the characteristics of natural systems. Examples include GA, PSO, ACO,
 Expectations from modern evolutionary approaches for image processing 223

and simulated annealing. Because of their ability to solve complex optimization

problems that traditional optimization techniques struggle with, these algorithms are
gaining popularity. Because of their ability to solve complex problems efficiently,
bio-inspired optimal algorithms have found uses across a broad spectrum of domains.
Here are some application domains where nature-inspired optimization algo-
rithms are commonly used.

Engineering design: Nature-inspired optimization algorithms can be used to

optimize engineering designs for products, processes, and systems, such as aero-
space design, automotive design, and structural engineering. For example, particle
swarm optimization has been used to optimize the design of airplane wings.
Finance: These algorithms can be used in finance for portfolio optimization, risk
management, and stock price prediction. For instance, ant colony optimization has
been used to solve the portfolio optimization problem, which involves selecting a
mix of assets to maximize returns while minimizing risks.
Machine learning: Optimization algorithms that take inspiration from nature are
widely employed to enhance the performance of various machine learning models,
including but not limited to neural networks, decision trees, and support vector
machines. As an illustration, genetic algorithms have been utilized to optimize the
architecture of neural networks.
Image and signal processing: These algorithms can be used in image and signal
processing to optimize parameters in image segmentation, filtering, and feature
extraction. For example, particle swarm optimization has been used to optimize the
parameters in image segmentation algorithms.
Robotics: NIOAs can be used to optimize the control of robotic systems. For
example, ant colony optimization has been used to optimize the path planning of
mobile robots.
Energy systems: These algorithms can be used to optimize energy systems such as
power generation, transmission, and distribution. For instance, particle swarm
optimization has been used to optimize the placement of wind turbines to maximize
energy output.
Engineering design optimization: NIOAs are used to optimize various engineer-
ing designs, such as structural designs, mechanical designs, and electronic designs,
for improved performance and efficiency.
Power system optimization: NIOAs are applied to optimize the operation and
control of power systems for improved energy efficiency, reliability, and stability.
Transportation system optimization: Algorithms are used to optimize transpor-
tation systems, including routing, scheduling, and resource allocation, for reduced
congestion and improved performance.
Robotics and automation: NIOAs are used to optimize robot behavior, motion
planning, and control for improved performance in various applications such as
manufacturing, logistics, and space exploration.
Image and signal processing: Algorithms are applied to optimize image and sig-
nal processing techniques for improved image quality, noise reduction, and signal
extraction.
224 Nature-inspired optimization algorithms and soft computing

Data mining and machine learning: Algorithms are used to optimize machine
learning algorithms and data mining techniques for improved accuracy, efficiency,
and scalability.
Financial optimization: Bio-inspired algorithms are applied to optimize financial
models, portfolio selection, and risk management strategies for improved financial
performance.
Biomedical engineering: Bio-inspired algorithms are used to optimize various
biomedical engineering applications, such as prosthetics, implants, and drug
delivery systems, for improved efficiency and effectiveness.
Agriculture and environmental optimization: NIOAs are applied to optimize
agricultural and environmental systems for improved crop yields, water manage-
ment, and environmental sustainability.
Game theory and optimization: NIOAs are used to optimize game strategies and
outcomes in various fields such as economics, psychology, and computer science.
Disaster management and emergency response: NIOAs are applied to optimize
disaster management and emergency response strategies for improved disaster
preparedness, response time, and resource allocation.
Social network analysis and optimization: Bio-inspired algorithms are used to
optimize social network analysis and management for improved communication,
collaboration, and information sharing.
Bioinformatics and genomics: NIOAs are applied to optimize bioinformatics and
genomic data analysis for improved accuracy, efficiency, and scalability.
Smart grid optimization: Algorithms are used to optimize the operation and
control of smart grids for improved energy efficiency, reliability, and stability.
Multi-objective optimization: Algorithms are applied to optimize multiple
objectives simultaneously, such as cost, performance, and environmental impact,
for improved decision-making.
Quality control and assurance: NIOAs are used to optimize quality control and
assurance techniques for improved product quality, defect detection, and
efficiency.
Supply chain optimization: Optimization algorithms are applied to optimize
supply chain management for improved efficiency, responsiveness, and cost-
effectiveness.
Chemical engineering: Bio-inspired algorithms are used to optimize chemical
processes and systems for improved efficiency, cost-effectiveness, and environ-
mental sustainability.
Construction and project management: NIOAs are applied to optimize con-
struction and project management processes for improved cost-effectiveness, effi-
ciency, and quality.
Energy management and optimization: NIOAs are used to optimize energy
consumption, production, and management for improved energy efficiency and
sustainability.

Overall, nature-inspired optimization algorithms have a wide range of appli-

cation domains and can be applied to a wide range of optimization problems.
 Expectations from modern evolutionary approaches for image processing 225

Nature-inspired optimization algorithms have resolved complex optimization pro-

blems in these and many other application domains. These algorithms could even
find near-optimal solutions in a reasonable amount of time and have proven to be
extremely effective in solving complex optimization problems that are difficult to
solve using traditional optimization methods.

9.1.1 Implementation
To discuss the process of implementation of a naturally inspired algorithm, an
optimization technique is considered and discussed its features and procedure of
implementation. The harmony search optimization algorithm is selected to study
how to implement an naturally inspired optimization algorithm to find an optimized
solution by maximizing or minimizing a given objective function.
Image segmentation method based on a multilevel thresholding technique is
considered for the explanation of the implementation of optimization techniques.

9.1.2 Finding optimized threshold level using harmony

search optimization algorithm
The procedure of image segmentation is partitioning an image into its constituent
parts, and a significant approach for extracting interesting features from images.
Over a couple of decades, many efficient image segmentation approaches were
formulated for various applications. Still, it is a challenging and complex issue,
especially for color image segmentation.
To compute the optimized threshold values, Otsu’s variance and Kapur’s
entropy [1] are deployed as fitness functions; both values should be maximized to
locate optimal threshold values. In both Kapur’s and Otsu’s methods, the pixels of
an image are classified into different classes based on the threshold level selected
on the histogram. Optimal threshold levels give higher efficiency of segmentation.
Digital image segmentation is a technique of partitioning the image into
regions to extract information about features of an image with homogeneous fea-
tures in terms of intensity level, texture structure, color information, etc. Image
segmentation schemes available from the literature, multi-level thresholding [1] of
grayscale on the histogram of an image is a highly established method and used in
various applications from satellite image segmentation to medical images. The
important multilevel thresholding-based segmentation techniques are Kapur’s and
Otsu’s methods [3]. The segmentation can be used often as a preprocessing step in
object recognition, computer vision, image analysis, and so on in different appli-
cations like medical, agricultural, industrial, fault detection, and weather forecast-
ing. In general, the majority of segmentation techniques are based on discontinuity
and similarity, among abundant methods available thresholding is the most
important technique for both grayscale and color images.
In general, to find out the optimized threshold values, two types of computa-
tional techniques are called parametric and nonparametric [4]. In the case of para-
metric techniques, statistical parameters are used, depending on initial conditions,
and hence are inflexible to be applied. In the case of nonparametric techniques,
226 Nature-inspired optimization algorithms and soft computing

thresholds are computed based on some criteria like Otsu’s inter-class variance and
Kapurs’s entropy functions [5–9]. The thresholding method holds the properties like
simplicity, accuracy, and robustness, which can be classified into two major cate-
gories: bi-level and multilevel [1,7], the pixels of an image are classified into dif-
ferent classes based on the threshold level selected on the histogram. In the case of
bi-level thresholding, all pixels are classified into two groups based on threshold
level. Pixels in the second category of multilevel thresholding are classified into
more than two classes. Nonetheless, the primary constraint in multilevel thresholding
is accuracy, stability, and execution time, among many other things.
In the case of color images [10], each pixel consists of three components (red,
green, and blue) [10], due to this heavy load, segmentation of color images might
be more exigent and intricate. Accordingly, it is essential to find the optimal
thresholds by using optimization algorithms by maximizing the inter-class variance
in Otsu’s and the histogram entropy in the case of Kapur’s on a histogram of an
image. As per the no-free-lunch (NFL) principle [11], no algorithm can solve all
types of optimization problems, one optimization algorithm may be useful well in
one type of application and not succeed to solve other kinds of applications; thus, it
is indispensable to devise and transform new algorithms.
Techniques with histogram plots are incapable of owning spatial contextual
information to compute optimized thresholds. To conquer these drawbacks, a novel
methodology is presented in this chapter, a curve which is having similar character-
istics that of the histogram, and considers spatial contextual information of image
pixels is named “Energy Curve” [3] can be used in the place of the histogram, HSA [1]
used to select optimized gray levels, energy curve characteristics are similar to a his-
togram. For each value in an image, energy is computed in the grayscale range of that
image. The threshold levels can be computed based on the values and peak points on
the energy curve. In the literature, numerous optimization techniques along with the
efficiencies and applications in particular fields are available; a few of them mentioned
are PSO [12], ACO [13], BFO [14], ABC [15], GWO [16], MFO [17], SSA [18], FA
[19], WOA [20], SCA [21], KHO [22], BA [23], FPA [24], and MVO [25]. Moreover,
several customized algorithms are used in multilevel thresholding. For example, Chen
et al. [26] proposed an improvised algorithm (IFA) to segment compared with PSO
[27], and other methods [10,27–29].

9.1.2.1 Image segmentation with multilevel thresholding

Otsu method
This technique [2,5,6] is used for multi-level thresholding ðMTÞ, in which gray
levels will be partitioned into different regions or classes; in this process, thresh-
olding ðthÞ levels need to be selected, and the set of rules to be followed for bi-level
thresholding are

C1 p if 0  p < th; C2 pif th  p < L  1 (9.1)

where C1 and C2 are two classes, p indicated pixel value for the gray levels
f1; 2; 3; :::::; L  1g in an image, and L  1 has indicated the maximum gray level.
 Expectations from modern evolutionary approaches for image processing 227

If the gray level is below the threshold th, then that pixel is grouped into class C1,
else grouped into class C2. The set of rules for MT are

C1 p if 0  p < th1
C2 p if th1  p < th2
(9.2)
Ci p if thi  p < thi
Cn p if thn  p < thn þ 1

From (9.2), C1; C2; :::::Cn indicates different classes, and threshold levels to
find objects are represented by fth1; th2; ::: :; thi; thi þ 1; thng; these thresholds can
be computed based on either a histogram or an energy curve. By use of these
threshold levels, the entire pixels will be classified into different regions (C1,
C2 ... Cn). The significant methods of segmentation of images based on threshold
levels are Otsu’s and Kapur’s methods, and in both cases, threshold levels can be
computed by maximizing the cost function (interclass variance). In this chapter,
optimized threshold levels thn are computed by Otsu’s method th values [17]. In
this method, interclass variance is considered the objective function, also called a
cost function. For experimentation, grayscale images are considered. The below
expression gives the probability distribution for each gray level:

hic XNP
Phic ¼ ; Phi ¼ 1 (9.3)
NP i¼1 c

From (9.3), pixel value is denoted by, range of grayscale is ð0  i  L  1Þ,

where c ¼ 1; 2; 3 for RGB and c ¼ 1 for grayscale image, the total image pixels are
represented by NP, and the histogram of considered images is represented by hic . In
bi-level thresholding, the total pixels in the image are grouped into two classes.

Phc1 Phcth Phcthþ1 PhcL

C1 ¼ ; ::: ; C2 ¼ ; ::: (9.4)
wc0 ðthÞ wc0 ðthÞ wc1 ðthÞ wc1 ðthÞ

Whereas w0 ðthÞ and w1 ðthÞ are the probability distributions for C1 and C2, as
it is shown below

X
th X
th
wc0 ðthÞ ¼ Phci ; wc1 ðthÞ ¼ Phci (9.5)
j¼1 j¼thþ1

The mean or average of two classes mc0 and mc1 to be computed, the variance
c
between classes s2 is given by (9.6) and (9.7)

Xth
iPhci X
L
iPhci
mc0 ¼ ; m c
¼ (9.6)
wc ðthÞ 1 i¼thþ1 wc1 ðthÞ
i¼1 0

c
s2 ¼ sc1 þ sc2 (9.7)
228 Nature-inspired optimization algorithms and soft computing

Notice that for both (9.6) and (9.7), c determined by the type of image, where
sc1 and sc2 in (9.8) are the variances of classes C1 and C2
 2  2
sc1 ¼ wc0 mc0 þ mcT ; sc2 ¼ wc1 mc1 þ mcT (9.8)

where mcT ¼ wc0 mc0 þ wc1 mc1 and wc0 þ wc1 ¼ 1: Based on the values of sc1 and sc2 , (9.8)
gives the objective function:
 c 
J ðthÞ ¼ max s2 ðthÞ ; 0  th  L  1 (9.9)
c
From (3.9), s2 ðthÞ is the total variance between two various regions after
segmentation by Otsu’s scheme for given th, optimization techniques required to
compute the threshold level ðthÞ by maximizing the fitness function is shown in
(9.9). Similarly, for MT, the objective (or fitness) function J(th) (as shown in
(9.10)) to segment an image into k classes, requires k variances.
 c 
J ðTH Þ ¼ max s2 ðthi Þ ; 0  thi  L  1; where i ¼ 1; 2:::k (9.10)

where TH is a vector, TH ¼ ½th1 ; th2 ; th3 ::: :::thk1  for multi-level thresholding; the
variances between classes can be computed from (9.11)

c
X
k X
k  2
s2 ¼ sci ¼ wci mci  mcT (9.11)
i¼1 i¼1

where ith represents i class, wci indicates probability of ith classes, and mcj is the mean
of ith class. For MT segmentation, these parameters are anticipated as

X
th1 X
th1 X
th1
wc0 ðthÞ ¼ Phci ; wc1 ðthÞ ¼ Phci    wck1 ðthÞ ¼ Phci (9.12)
i¼1 i¼th1 þ1 i¼thk þ1

And, the averages of each class can be computed as

Xth1
iPhci X
th2
iPhci XL
iPhci
mc0 ¼ ; m c
¼    m c
¼ (9.13)
i¼1 0
wc ðth1 Þ 1 i¼th þ1 wc0 ðth2 Þ k1
wc ðthk Þ
i¼th þ1 1
1 k

Multilevel thresholding with Kapur’s method

Kapur’s method [1,2], which uses entropy as an objective function, is an important
nonparametric technique for determining optimal threshold values. This method
focuses on finding the best thresholds by maximizing total entropy. Entropy is a
measure of the compactness and separability of classes. Kapur’s method’s objective
function for multilevel is defined as
!
X k
J ðTH Þ ¼ max Hi ; 0  thi  L  1; where i ¼ 1; 2 ::: k
C
(9.14)
i¼1
 Expectations from modern evolutionary approaches for image processing 229

where TH is a vector, ½th1 ; th2 ; th3 ::: :::thk1 . Each entropy is calculated separately
with its th value, given for k entropies
X
th1   Xth1  c
Phc Phci Phci Phi
H c1 ¼ c
i
ln H c
2 ¼ ln    H ck
i¼1
w 0 wc0 i¼1
w c
1 w c
1

X
th1  c
Phci Phi
¼ c ln (9.15)
w
i¼th þ1 k1
wck1
k

Phci is the probability distribution of the particular intensity levels, which is

obtained using (9.3). The probability distribution values (wc0 , wc1 , wc2 ... ... wck1 ) of
the k classes are obtained using (9.5). In the end, by using (3.3), classify the pixels
into various classless.
Energy curve
To find effective optimized threshold levels, the energy curve [3] can be used as an
alternative of the histogram of an image for various applications.
Equation of energy curve
Consider an image indicated as I ¼ xði; jÞ, where i and j are spatial coordinates,
i ¼ 1; 2; :: N and j ¼ 1; 2; ::: M, and size of image is X ¼ M  N . For an image,
spatial correlation among neighboring the pixels can be devised by devising the
neighborhood
 system with N of order d of an image with spatial coordinates ði; jÞ
as Nijd ¼ ði þ u; j þ vÞ; ðu; vÞ 2 N d , various configurations of the neighborhood
are described in Ref. [22]. Neighborhood systems with order 2 are calculated for
the creation of an energy curve, i.e., ðu; vÞ 2 fð1; 0Þ; ð0; 1Þ; ð1; 1Þ; ð1; 1Þg
(Figure 9.1).
The foremost step is to get the energy of each pixel of an image considered,
generate a binary matrix Bx ¼ bij ; 1  i  M; 1  j  N , the bij ¼ 1 if xij >x;
else bij ¼ 1. Let C ¼ cij ; 1  i  M; 1  j  N is another matrix, where
cij ¼ 1,8ði; jÞ. At each pixel value x and the energy value Ex of the image, I can be
computed with the below expression:
X
M X
N X X
M X
N X
Ex ¼  bij bpq þ cij cpq (9.16)
i¼1 j¼1 pq2Nij2 i¼1 j¼1 pq2Nij2

From (9.16), its second term should be a constant; consequently, the energy
associated with each pixel is Ex  0. From the above equation, we can clear that the
energy for a particular gray, level is zero if each element of Bx either 1 or  1, can
be put forward in another way as all the pixels of image I ði; jÞ with gray level either
greater than x or less than x, otherwise energy level at a particular gray value x is
positive as given in Figure 9.2.
Problem statement
The optimal threshold levels can be computed by (i) maximizing inter-class var-
iance for Otsu’s method and maximizing total entropy for Kapurs’s method. For the
230 Nature-inspired optimization algorithms and soft computing

Take image
x(i,j)

Fing Histogram
of image

lntialize the HSA parameters and

Hamony Memory Matrix

Find optimized threshold levels

Apply Otsu’s or Kapur’s method

for image segmentation

Is termination No
criteria satisfied

Yes

Obtain optimal thresholded image

End

Figure 9.1 Flowchart to find optimized threshold levels using HSA

multilevel, the objective function of Kapur’s method is defined as

!
Xk
J ðTH Þ ¼ max H Ci ; 0  thi  L  1; where i ¼ 1; 2 ::: k
i¼1

The inter-class variance for Otsu’s method is given as

 c 
J ðthÞ ¼ max s2 ðthÞ ; 0  th  L  1

9.1.2.2 Harmony search optimization algorithm

The HSA generates solutions, referred to as “harmonies,” that are n-dimensional
real vectors. Initially, random values are assigned to create the population and store
them in the harmony memory (HM). The HM is then used to produce a new can-
didate harmony by adjusting the pitch or selecting random elements for updating.
The worst HM vector is compared with the newly computed candidate harmony to
update the elements in the HM. This process is repeated several times until the
termination condition is met. Coupling HSA [1] and Otsu’s algorithm can yield
various segmentation algorithms.
 Image

Butterfly Brain Bridge Baboon Man

8000 0.09 0.012 3000 12000

7000 0.08
0.01 2500 10000
0.07
6000
Probability of Gray Value

Probability of Gray Value

0.06 0.008 2000 8000
Histogram

5000
0.05
4000 0.006 1500 6000
0.04
3000
0.03 0.004 1000 4000
2000 0.02
0.002 500 2000
1000 0.01

0 0 0 0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300
GVl
x 10 5 Energy Curve of Image
3.5 x 105 Energy Curve of Image 5 Energy Curve of Image
6
x 104 Energy Curve of Image 6 8
x 10 5 Energy Curve of Image 3 x 10
3
6 2.5
5 5
2.5 6

Energy Value

Energy Value
4 2
Energy Value
Energy Value

Energy Value

4
5
2
3 3 4 1.5
1.5
3
2 2 1
1 2
1 1 1 0.5
0.5
0 0
0 0 0 0 50 100 150 200 250 300 0 50 100 150 200 250 300
0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250
GrayValue

Figure 9.2 First row: input images; second row: histograms; third row: energy curves
232 Nature-inspired optimization algorithms and soft computing

Parameters of the algorithm

In the algorithm, the problem statement and variables can be summarized as the mini-
mization or maximization of the function ðxÞ; x ¼ ðxð1Þ; xð2Þ; ::: ::: :: xðnÞÞ 2 Rn , The
HSA aims to maximize the inter-class variance, which is expressed as the objective
function in (9.11). The x indicates a set of variables to be designed.
Where xðk Þ 2 ½Xl ðk Þ; Xu ðk Þ for the range k ¼ 1; 2; ::: ::n (9.17)
The objective function, f ðxÞ, is a cost function that is dependent on a set of
variables denoted by x, where n is the length of the variable set. The upper and
lower limits of the design variables, xðk Þ, are represented by Xu ðk Þ and Xl ðk Þ,
respectively. The HSA use several parameters, including the size of the harmony
memory (HM), pitch adjusting rate (PAR), harmony-memory consideration rate
(HMCR), and the number of iterations or improvisations (NI) distance
bandwidth (BW).
Harmony memory initialization
In this stage, it is crucial to set up the initial components of the harmony search
algorithm, known as the harmony memory (HM) vectors. Each HM vector denoted
as xi ¼ fxið1Þ; xið2Þ; ::: xiðnÞg is generated randomly using the following equation:
xiðk Þ ¼ Xl ðk Þ þ ðXu ðk Þ  Xl ðk ÞÞ rand ð0;1Þ for k ¼ 1; 2; ::: :; n and i ¼ 1; 2; ::: :;
HMS, that is length of HM, where HMS is the length of the HM. Here, Xl ðk Þ and
Xu ðk Þ represent the lower and upper limits of the search space, respectively. These
vectors are used to form the HM matrix, which contains HMS harmony vectors.
2 3
x1
6 x2 7
6 7
HM ¼ 6 .. 7 (9.18)
4 . 5
xHMS

Updating harmony memory and harmonic vector

The harmony search algorithm generates a new harmony vector, x_new, from the
harmony memory (HM) matrix using three operations: memory consideration,
random reinitialization, and pitch adjustment [1]. In the first phase, the initial
decision value, xnew ð1Þ, is selected randomly from the harmony memory set
fx1 ð1Þ; x2 ð1Þ::: : xHMS ð1Þg. A random number, r1, is generated within the range
[0,1], and if it is less than HMCR, then xnew ð1Þ is generated by memory con-
sideration. Otherwise, xnew ð1Þ is selected from the search range ½Xl ðk Þ; Xu ðk Þ. The
remaining decision values, xnew ð2Þ, xnew ð3Þ, ..., xnew ðnÞ, are selected similarly. The
two operations, memory consideration, and random reinitialization can be sum-
marized as follows:
8
< xi ðk Þ 2 fx1 ðk Þ::: ::: ::: ::: : xHMS ðk Þg
xnew ðk Þ ¼ Xl ðk Þ þ ðXu ðk Þ  Xl ðk ÞÞ rand ð0;1Þ (9.19)
:
with probability of 1  HMCR
 Expectations from modern evolutionary approaches for image processing 233

After generating each new decision value, xnew , the harmony search algorithm
checks whether pitch adjustment is required or not. To do this, a PAR is defined,
which is a combination of the frequency and bandwidth factor (BF). These two
factors are adjusted to obtain a new xnew value from a local search around the HM.
The pitch-adjusted solution, xnew ðkÞ, can be calculated as xnew(k)+/-rand(0,1)•BW
with a probability of PAR. This pitch adjustment process is similar to the mutation
process in other bio-inspired algorithms. The range of pitch adjustment is restricted
by the limits ½Xl ðk Þ; Xu ðk Þ.
In the finishing stage, a new harmony vector, xnew , is computed and evaluated
against the worst harmony vector, xw , in the harmony memory. The fitness of xnew is
compared with that of xw , and if it is better, xw is replaced by xnew and becomes part
of the harmony memory. This process is repeated to maximize the objective function.
The optimization algorithm used in this study considers k variables, repre-
sented by threshold values thk , to form a harmony or solution. For multilevel seg-
mentation, the algorithm’s population is represented by
HM ¼ ½x1 ; x2 ::: :; x1HMS T ; xi ¼ ½th1 ; th2 ::: ::: ::; thk  (9.20)
where each element xi in the harmony memory represents a threshold value for each
of the k variables. The HMS is the maximum size of HM. In this study, an 8-bit
image with intensity values ranging from 0 to 255 is used, and the algorithm’s
parameters are set to HM = 50, HMCR = 0.5, PAR = 0.2, BW = 0.1, and NI = 3,500.
The number of regions considered for segmentation ranges from 2 to 5.
Computing optimized threshold levels
The introduction section of this chapter outlines various multilevel thresholding
techniques for image segmentation and highlights the limitations of the histogram-
based methods. In the proposed method, an Energy Curve is utilized in place of the
histogram, and the optimization of threshold levels is achieved using HSA. This is
done by maximizing the inter-class variance for Otsu’s method and entropy for
Kapur’s method, as shown in (9.10) and (9.14). The new approach is visually
represented in Figure 9.1.
From the flow chart, take an image for experimentation xði; jÞ for multilevel
thresholding-based segmentation, and plot the energy curve of the considered color
image by using (3.1), then assign the design parameters of HSA, and the solution
matrix values are filled with arbitrary numbers, initially denoted as xi (set of
threshold levels) as per (3.18), then divide the entire pixels in the image as per
selected threshold levels into different classes or regions as per Otsu’s technique
and Kapur’s method, then find the inter-class variance and entropy of the seg-
mented image, as given in (9.14). Afterward, find the new set of threshold levels
with (9.2) again, find the fitness and compare it with the previous fitness function,
run this procedure until no improvement in the objective function or reach the
specified number of iterations, and lastly find the optimized threshold valued (xnew )
and classify the gray levels as (9.1) and (9.2) for the final segmentation of R, G, and
B components separately for color images. The results of this scheme are compared
with histogram-based techniques for evolution.
234 Nature-inspired optimization algorithms and soft computing

9.2 Results
This chapter introduces a novel technique for image segmentation that overcomes
the limitations of methods that rely solely on image histograms. Such methods lack
spatial contextual information and therefore may not yield optimal threshold levels.
To address this issue, the proposed method utilizes energy curve, which captures
the spatial relationship between pixels and their neighbors. The optimization of
threshold levels for multilevel segmentation is achieved using harmony search
algorithm (HSA) in combination with energy curve (Figures 9.3 and 9.4).
The primary objective of this article is to highlight the benefits of using the
energy curve in comparison to the histogram. The research adopts a primarily
quantitative approach for evaluating the proposed method, which involves ana-
lyzing nine images. To implement the proposed method, a MATLAB“ code is
developed, and the experiment is conducted on an eighth-generation Intel processor
CORE i5-8250u with a clock speed of 1.60 GHz and 8 GB of internal memory. The
MATLAB code is designed to generate energy curves and validation measures,
which are then presented in tables. Some of the results presented in this chapter are
taken from papers [30] authored by the authors of this chapter.
The resulting images are displayed in Figures 9.2–9.4, while comparative
metrics are provided in Tables 9.1 and 9.2. The optimized threshold levels for N =
2–5 are indicated for five standard images, and the mean fitness of the proposed
system is compared to various algorithms. The histograms and energy curves are
displayed with their corresponding threshold levels denoted by red spots. In the
tables, histogram is referred to as “H,” the energy curve as “E,” the number of
regions as “N,” and the threshold levels as “Th Levels.” The segmentation efficiency
improves with an increase in the mean value of the fitness function. Table 9.1 present
the mean fitness function values for the proposed technique and algorithms based on
histograms like Otsu’s method with HAS (HAS_O) [1], Kapur’s method based on
HSA (HAS_K) [1], PSO with Otsu’s method [31], Kapur’s method with EMO
(EMO_K) [32], and Otsu’s method with EMO (EMO_O) for optimized threshold
levels of N = 2–5 for five benchmark images. As shown in Table 9.1 for the bridge
and butterfly images, the proposed method has a higher mean fitness function value
than the other algorithms. Table 9.1 illustrates the proposed method’s threshold
levels and mean fitness function values for the baboon and man images, highlighting
that the proposed method provides higher fitness values. In comparison to the pro-
posed method, the HAS K and EMO K methods produce lower fitness function
values. When N = 2, the mean object function for the bridge image is 2.06E+03 with
the histogram and 1.58521E+11 with the energy curve.
Time taken for threshold level optimization and segmentation through histo-
gram and energy curve approaches are presented in Table 9.2. The findings
demonstrate that the proposed technique demands lesser time in comparison to
other optimized algorithms. On an average, the proposed method takes
13.7813 seconds, which is lower than the time required for the EMO technique
employing Otsu’s method and Kapur’s method. In summary, the threshold levels
optimized using the HSA based on energy curve outperforms the ones obtained via
 0.012 0.012 0.012 0.012

0.01 0.01 0.01 0.01

Probability of Gray Value

Probability of Gray Value
Probability of Gray Value

Probability of Gray Value

0.008 0.008 0.008 0.008

0.006 0.006 0.006 0.006

0.004 0.004 0.004 0.004

0.002 0.002 0.002

0.002

0 0 0
0 0 50 100 150 200 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300
0 50 100 150 200 250 300 Gray Value Gray Value Gray Value
Gray Value

Figure 9.3 First row: segmentation results of “bridge” image with histogram on N = 2, 3, 4, and 5. Threshold given in second row
is indicated by red dots on histogram.
 2 2 2
2
1.8 1.8 1.8
1.8
1.6 1.6

Probability of Gray Value

1.6

Probability of Gray Value

1.6
Probability of Gray Value

Probability of Gray Value

1.4 1.4 1.4 1.4

1.2 1.2 1.2 1.2
1 1 1 1
0.8 0.8
0.8 0.8
0.6 0.6
0.6 0.6
0.4 0.4
0.4 0.4 0.2
0.2
0.2 0.2 0
0 0 50 100 150 200 250 300
0 0 0 50 100 150 200 250 300
0 50 100 150 200 250 300 0 50 100 150 200 250 300 Gray Value
Gray Value Gray Value
Gray Value

Figure 9.4 First row: segmentation results of “bridge” image with histogram on N = 2, 3, 4, and 5. Threshold given in second row
is indicated by red dots on energy curve.
Table 9.1 Thresholding levels for different levels (N), and mean of fitness function with different segmentation methods

Image HSA_O HSA_K PSO_O EMO_O EMO_K Proposed method

N Th Mean of Th Mean of Th Mean of Th Mean of Th Mean of Th Mean of
fitness fitness fitness fitness fitness fitness
Bridge 2 127 2.063e+03 116 13.5021 120 1.944e+03 126 1.944e+03 116 9.573 122 1.585E+11
3 90 2.532e+03 90 12.678 95 2.56e+03 95 2.380e+03 92 12.690 93 1.45E+11
157 161 160 160 164 152
4 76 2.720e+03 65 15.418 75 2.722e+03 74 2.560e+03 65 15.450 77 1.45E+11
124 126 123 121 126 120
182 187 178 177 187 170
5 65 2.78E+03 53 17.803 61 2.822e+03 66 2.82e+03 57 17.782 67 1.45E+11
105 98 104 105 106 102
144 147 144 146 155 132
192 199 192 193 203 182
Butterfly 2 124 1.272e+03 213 12.040 122 1.272e+03 122 1.272e+03 213 12.040 122 3.195E+10
3 99 1.553e+03 27 16.795 98 1.553e+03 101 1.552e+03 120 16.662 99 3.195E+10
151 213 150 153 213 147
4 82 1.669e+03 27 21.417 81 1.669e+03 81 1.669e+03 27 21.417 84 3.698E+10
118 120 118 119 120 118
160 213 159 160 213 157
5 83 1.697e+03 27 25.292 70 1.711e+03 71 1.711e+03 27 25.295 78 3.659E+10
106 96 97 99 96 108
129 141 125 126 144 135
165 213 160 162 213 166
(Continues)
Table 9.1 (Continued)

Image HSA_O HSA_K PSO_O EMO_O EMO_K Proposed method

N Th Mean of Th Mean of Th Mean of Th Mean of Th Mean of Th Mean of

fitness fitness fitness fitness fitness fitness
Baboon 2 127 1.219e+03 201 13.100 127 1.2191e+03 128 1.2191e+03 193 13.100 123 4.395e+11
3 97 1.547e+03 94 19.568 96 1.548e+03 77 1.548e+03 74 18.330 100 4.154e+11
129 193 128 169 193 145
4 85 1.637e+03 69 22.8685 84 1.638e+03 85 1.638e+03 70 22.868 86 4.657e+11
123 124 124 125 125 121
155 193 160 161 193 162
5 71 1.692e+03 54 26.962 71 1.692e+03 72 1.692e+03 54 26.9627 77 4.257e+11
116 94 105 116 94 108
148 135 136 137 135 135
169 193 167 168 193 163
Man 2 85 2.700e+03 101 15.358 82 2.700e+03 82 2.700e+03 189 13.256 80 3.092e+10
3 51 3.064e+03 72 17.856 50 3.064e+03 51 3.064e+03 92 18.4609 57 3.092e+10
120 169 115 116 189 115
4 36 3.213e+03 51 22.525 33 3.213e+03 33 3.2134e+03 70 23.2409 45 3.0922e+10
87 117 85 85 151 89
137 163 134 134 199 135
5 30 3.267e+03 26 26.729 27 3.269e+03 31 3.269e+03 40 27.478 39 3.092e+10
72 95 65 73 96 79
111 139 104 112 125 103
146 183 142 147 179 146
 Expectations from modern evolutionary approaches for image processing 239

Table 9.2 Time required for segmentation based on the proposed method and
other methods

Time (s)

Image N HSA_O HSA_K PSO EMO_O EMO_K Proposed

Bridge 2 10.125 18.0511 8.2658 50.51403 94.0337 6.1258
3 16.2565 21.4523 12.566 104.4127 145.262 8.2365
4 19.2365 20.5066 19.8609 155.0894 196.625 12.569
5 24.369 79.7038 28.7158 218.9217 322.3668 16.547
Butterfly 2 7.8694 18.843 18.8806 64.44 94.9834 6.998
3 14.8945 23.0062 20.7511 130.793 142.674 13.569
4 19.658 23.0588 22.0223 182.19 291.81 16.458
5 24.255 26.5361 23.0086 244.834 254.536 18.1675
Baboon 2 13.569 18.235 15.5688 60.6078 92.8163 10.257
3 16.369 22.9658 19.9815 122.8779 145.5034 11.265
4 18.6254 29.458 19.2624 268.2846 206.9632 13.524
5 26.8027 41.8257 20.5374 261.3762 266.7319 16.654
Man 2 12.546 14.4032 11.8400 57.35 95.3515 9.657
3 13.369 22.6598 14.7413 118.759 148.6318 10.894
4 15.365 26.365 25.3560 181.8948 197.8918 12.324
5 18.254 28.369 29.5827 244.309 248.3340 14.258
Average 20.915 33.3438 22.9594 148.4578 179.811 13.7813

histogram methods. Consequently, the segmentation process with the energy curve
approach is superior to histogram techniques. This study predominantly utilized
quantitative methods to scrutinize the results.

9.3 Parameter tanning

Bio-inspired optimization algorithms typically have several parameters that can be
tuned to achieve better performance on different optimization problems. Some of
the common running parameters of bio-inspired optimization algorithms are:
1. Population size: This parameter controls the number of candidate solutions
generated and evaluated in each algorithm iteration. Larger population size
may lead to superior solutions, but it also increases the computational cost.
2. Mutation rate: This parameter controls the rate at which mutations are
introduced into the population. Mutations are random changes to the candidate
solutions that allow the algorithm to explore new areas of the search space. A
higher mutation rate can lead to more exploration, but may also reduce the
convergence speed.
3. Crossover rate: This parameter controls the rate at which crossover is applied
to the candidate solutions. Crossover is a genetic operator that combines two or
more candidate solutions to create new solutions. A higher crossover rate can
lead to more exploitation of good solutions, but may also reduce the diversity
of the population.
240 Nature-inspired optimization algorithms and soft computing

4. Selection strategy: This parameter determines how the candidate solutions are
selected for reproduction in each iteration of the algorithm. Common selection
strategies include roulette wheel selection, tournament selection, and rank-
based selection.
5. Termination criteria: This parameter specifies when the algorithm should be
stopped. Reaching a maximum number of iterations, a minimum
acceptable fitness value, or a specified amount of computing time all are
common termination criteria.
6. Initialization method: This parameter controls how the first set of candidate
solutions is generated. Random initialization, initialization based on domain-
specific knowledge, and initialization based on previous successful solutions
are all common initialization methods.
7. Fitness function: This parameter determines how the quality of candidate
solutions is evaluated. The fitness function should be chosen carefully to
accurately reflect the optimization problem being solved.
These parameters can be adjusted based on the distinctiveness of the optimi-
zation problem being solved to achieve better performance.
The GA and HSA are both population-based optimization algorithms that are
devised to locate the optimal solution for the problem under test. While both
algorithms have their own unique set of parameters, the most common parameters
that need to be tuned for both algorithms are:
Parameters for genetic algorithm
1. Population size: The number of individuals in each generation of the
algorithm.
2. Selection method: The method used to select individuals for the next gen-
eration (e.g., roulette wheel selection, tournament selection, etc.).
3. Crossover rate: The probability that crossover will occur between two indi-
viduals in the population.
4. Mutation rate: The probability that mutation will occur in an individual in the
population.
5. Fitness function: The function used to evaluate each individual’s fitness in the
population.
Parameters for harmony search algorithm
1. Harmony memory size: The size of the harmony memory used in the
algorithm.
2. Pitch adjusting rate: The probability that a pitch adjustment will occur in the
algorithm.
3. Harmony memory consideration rate: The rate at which the algorithm
considers the harmony memory when generating new solutions.
4. Maximum improvisation iterations: The maximum number of iterations the
algorithm will perform before converging to a solution.
5. Objective function: The function used to evaluate the fitness of each gener-
ated solution.
 Expectations from modern evolutionary approaches for image processing 241

It is important to note that the optimal values for these parameters may vary
depending on the problem being solved. Therefore, it is recommended to experi-
ment with different parameter values and analyze the results to determine the
optimal set of parameters for a specific problem.

9.4 Constrained and unconstrained optimization

The two different types of optimization techniques used to find the maximum or
minimum value of a function are constrained optimization and unconstrained
optimization. Unconstrained optimization is the process of finding the smallest or
maximum value of a function without any constraints. In other words, there are no
restrictions on the variables’ possible values. The goal of unconstrained optimiza-
tion is to find the variable values that result in the function’s minimum or
maximum value.
Constrained optimization involves finding the minimum or maximum value of
a function subject to one or more constraints. In other words, there are limitations
on the values that the variables can take on. The goal of constrained optimization is
to find the value of the variables that satisfy the constraints and result in the
minimum or maximum value of the function.
There are several differences between constrained and unconstrained optimi-
zation techniques:

1. Constraints: The most obvious difference is the presence of constraints in

constrained optimization problems. Unconstrained optimization has no
constraints.
2. Feasibility: In constrained optimization, the solution must satisfy the con-
straints, making the solution feasible. In unconstrained optimization, there are
no feasibility requirements.
3. Objective function: In both constrained and unconstrained optimization, the
objective function is the function that we are trying to optimize. In constrained
optimization, the objective function is subject to constraints.
4. Solution: Constrained optimization problems often have multiple solutions
that satisfy the constraints, whereas unconstrained optimization problems
usually have a unique solution.
5. Optimization techniques: Constrained optimization often requires more
advanced optimization techniques, such as linear programming or nonlinear
programming, whereas unconstrained optimization can often be solved using
simpler techniques, such as gradient descent or Newton’s method.
6. Computational complexity: Constrained optimization problems are generally
more computationally complex than unconstrained optimization problems due
to the presence of constraints.

Overall, the presence of constraints is the key difference between constrained

and unconstrained optimization techniques.
242 Nature-inspired optimization algorithms and soft computing

The examples of constrained optimization problems in engineering are given

below:
1. The optimal design of a truss structure is subject to constraints on the max-
imum stress, displacement, and weight.
2. Optimal control of a robotic arm is subject to constraints on the joint angles,
velocity, and acceleration.
3. The optimal sizing of a heat exchanger is subject to constraints on the flow
rate, pressure drop, and temperature difference.
4. The optimal allocation of resources in a manufacturing system is subject to
constraints on capacity, lead time, and inventory level.
5. The optimal placement of sensors in a power grid is subject to constraints on
the coverage, detection rate, and false alarm rate.
6. Optimal scheduling of production processes in a chemical plant is subject to
constraints on the reaction rate, yield, and purity.
7. Optimal trajectory planning of a UAV is subject to constraints on the flight
range, altitude, and speed.
8. The optimal placement of wind turbines in a wind farm is subject to con-
straints on the wind speed, turbulence, and wake effect.
9. Optimal design of a control system for a power plant subject to constraints on
stability, response time, and energy efficiency.
10. The optimal routing of vehicles in a transportation network is subject to
constraints on the traffic flow, capacity, and emissions.
These are just a few examples of the many constrained optimization problems
in engineering. In each case, the goal is to find the optimal solution while satisfying
a set of constraints that reflect the physical limitations and requirements of the
system.
The examples of unconstrained optimization problems in engineering are
given below:
1. Minimizing the weight of a cantilever beam subject to a fixed load and con-
straints on the stress and deflection.
2. Maximizing the efficiency of a heat engine is subject to constraints on tem-
perature, pressure, and heat transfer.
3. Minimizing the power consumption of a motor subject to constraints on the
torque and speed.
4. Maximizing the signal-to-noise ratio of a communication system is subject to
constraints on bandwidth and power.
5. Minimizing the vibration of a mechanical system subject to constraints on the
stiffness and damping.
6. Maximizing the output power of a photovoltaic system is subject to con-
straints on the irradiance and temperature.
7. Minimizing the total cost of a production process subject to constraints on the
throughput and quality.
8. Maximizing the accuracy of a machine learning model is subject to con-
straints on the training time and complexity.
 Expectations from modern evolutionary approaches for image processing 243

9. Minimizing the error of a control system subject to constraints on the stability

and response time.
10. Maximizing the fuel efficiency of a vehicle subject to constraints on speed
and acceleration.
In each case, the goal is to find the optimal solution to a single-objective or
multi-objective optimization problem without any constraints or limitations on the
decision variables.
The examples of constrained and unconstrained optimization problems in
image processing are given below:
1. Unconstrained: Minimizing the distortion of a compressed image subject to
no constraints.
2. Unconstrained: Maximizing the sharpness of an image subject to no
constraints.
3. Unconstrained: Minimizing the noise level of an image subject to no
constraints.
4. Unconstrained: Maximizing the contrast of an image subject to no constraints.
5. Unconstrained: Minimizing the computational cost of an image processing
algorithm subject to no constraints.
6. Constrained: Minimizing the distortion of a compressed image subject to a
maximum allowable bit rate constraint.
7. Constrained: Maximizing the sharpness of an image subject to a maximum
allowable noise constraint.
8. Constrained: Minimizing the noise level of an image subject to a minimum
allowable signal-to-noise ratio constraint.
9. Constrained: Maximizing the contrast of an image subject to a minimum
allowable brightness level constraint.
10. Constrained: Minimizing the computational cost of an image processing
algorithm subject to a maximum allowable processing time constraint.
In each case, the goal is to find the optimal solution to an optimization problem
that optimizes a certain image processing task while satisfying certain constraints
or limitations on the decision variables.

9.5 How to deal with constraints

Constraints in optimization algorithms refer to the limitations or restrictions on the
variables or parameters of the problem being optimized. These constraints may
arise due to physical, economic, or other real-world considerations. To deal with
constraints in optimization algorithms, you can follow the following steps:
1. Formulate the constraints: First, you need to define the constraints that apply to
the problem. The constraints can be of two types: equality constraints and
inequality constraints.
2. Use appropriate optimization technique: There are various optimization tech-
niques available to deal with constrained optimization problems. Some of the
244 Nature-inspired optimization algorithms and soft computing

popular optimization techniques include linear programming, quadratic pro-

gramming, non-linear programming, and dynamic programming. Choose the
appropriate technique based on the complexity and nature of the problem.
3. Convert constraints into penalties: One way to incorporate constraints in
optimization algorithms is by converting them into penalties. This involves
adding a penalty term to the objective function that is proportional to the
violation of the constraint. The penalty function is designed in such a way that
it becomes very large when the constraint is violated significantly.
4. Use Lagrange multipliers: Another way to incorporate constraints in optimi-
zation algorithms is by using Lagrange multipliers. This technique involves
adding a new variable to the objective function for each constraint. The
Lagrange multipliers are then used to ensure that the constraints are satisfied.
5. Use interior-point methods: Interior-point methods are a type of optimization
technique that can be used to solve constrained optimization problems. This
method works by transforming the original problem into an unconstrained
problem and then solving it using iterative techniques.
In summary, dealing with constraints in optimization algorithms involves
formulating the constraints, selecting an appropriate optimization technique, con-
verting constraints into penalties, using Lagrange multipliers, and using interior-
point methods.
Harmony search (HS) is a metaheuristic optimization algorithm that is com-
monly used to solve unconstrained optimization problems. However, there are
situations where constraints are present in the problem being optimized. In such
cases, the following methods can be used to deal with constraints in the harmony
search algorithm:
1. Penalty function method: This method involves incorporating the constraints
as penalty terms in the objective function of the harmony search algorithm. The
penalty term is a function of the violation of the constraints and is added to the
objective function. The penalty function is designed in such a way that it
becomes very large when the constraints are violated significantly. The HS
algorithm then minimizes the modified objective function that includes the
penalty terms.
2. Dynamic bandwidth adjustment method: This method involves dynamically
adjusting the bandwidth of the HS algorithm. The bandwidth is the parameter
that controls the pitch adjustment of the harmony vectors. The bandwidth is
adjusted such that the constraints are satisfied. The bandwidth adjustment is
done iteratively until the constraints are satisfied.
3. Augmented Lagrangian method: This method involves adding the Lagrange
multipliers to the objective function of the harmony search algorithm. The
Lagrange multipliers are used to enforce the constraints in the optimization
process. The augmented Lagrangian function is optimized using the HS algo-
rithm to obtain the optimal solution.
In summary, the penalty function method, dynamic bandwidth adjustment
method, and augmented Lagrangian method are three ways to deal with constraints
 Expectations from modern evolutionary approaches for image processing 245

in the harmony search optimization algorithm. These methods ensure that the
optimal solution is found while satisfying the constraints of the problem.

9.6 Feature selection

The process of selecting a subset of relevant features or variables from a larger set
of features available in a dataset is termed feature selection [33]. The goal of fea-
ture selection is to reduce data dimensionality, remove irrelevant or redundant
features, and improve machine learning model accuracy and efficiency. In many
real-world problems, datasets often contain a large number of features, but not all
of them contribute equally to the predictive power of the model. In some cases,
using all the available features can lead to overfitting, which means the method or
model performs well on the training data but poorly on the test data. Feature
selection can help to overcome this problem by selecting only the most informative
and relevant features for the model.
The process of selecting a subset of relevant features or descriptors that capture
important characteristics or patterns in an image is referred to as feature selection in
image processing. The features selected can be used for a variety of tasks such as
object recognition, image classification, and image segmentation.
Image feature selection is important because images can contain a large
number of features, and not all of them may be useful or relevant to the task at
hand. Moreover, using all the features can lead to overfitting or high computational
complexity, which can affect the performance of the algorithm.
There are different techniques for feature selection in image processing,
including:
1. SIFT (scale-invariant feature transform): This method extracts key features
from images that are invariant to scale, orientation, and affine distortion. The
SIFT features can be used for object recognition and image matching.
2. HOG (histogram of oriented gradients): This method extracts features that
capture the local orientation information of an image. The HOG features can be
used for pedestrian detection and human pose estimation.
3. SURF (speeded up robust features): This method extracts features that are
invariant to scale, rotation, and illumination changes. The SURF features can
be used for object recognition and image matching.
4. Deep learning: This involves using convolutional neural networks (CNNs) to
learn features directly from the raw pixels of an image. The CNNs can be
trained on large datasets to learn generic features that can be used for various
image-processing tasks.
Nature-inspired optimization algorithms, such as genetic algorithms, particle
swarm optimization, and ant colony optimization, can be used to perform feature
selection in machine learning tasks. This process can help to reduce the dimen-
sionality of the data and improve the accuracy and efficiency of the machine
learning model.
246 Nature-inspired optimization algorithms and soft computing

Nature-inspired optimization algorithms can be used to perform feature

selection by searching for the subset of features that maximizes the performance of
the machine learning model. These algorithms work by iteratively evaluating
candidate subsets of features and selecting the best subset based on a fitness
function that measures the performance of the model.
For example, in a genetic algorithm approach, the candidate subsets of features
can be represented as chromosomes, and the fitness function can depend on the
accuracy of the machine learning model trained on the selected subset of features.
The algorithm then evolves the population of candidate subsets over several gen-
erations, using selection, crossover, and mutation operators to generate new can-
didate solutions. Similarly, in a particle swarm optimization approach, the
candidate subsets of features can be represented as particles that move through a
search space of possible subsets, and the fitness function can be based on the
accuracy of the machine learning model trained on the selected subset of features.
The algorithm then updates the position of each particle based on its own best
solution and the best solution of the swarm.

9.7 Advantages of using optimization techniques in

engineering applications
Optimization techniques are widely used in engineering applications due to their
numerous advantages.
Here are some of the key advantages of using optimization techniques in
engineering applications:
1. Improved efficiency: Optimization techniques can help engineers to design
and develop systems that are more efficient, reducing energy consumption
and increasing productivity.
2. Enhanced performance: Optimization techniques can help engineers to optimize
the performance of systems, improving the accuracy and reliability of results.
3. Faster development: Optimization techniques can help engineers to reduce
the time it takes to develop new systems, allowing them to bring new products
to market more quickly. Optimization techniques are often able to find the
optimal solution faster than traditional methods. This is because they are
designed to search the solution space more efficiently and effectively, which
can save time and resources.
4. Increased safety: Optimization techniques can help engineers to design safer
systems by identifying and addressing potential safety risks during the
design phase.
5. Improved sustainability: Optimization techniques can help engineers to
design systems that are more environmentally sustainable, reducing their
carbon footprint and minimizing their impact on the environment.
6. Flexible design: Optimization techniques can help engineers to design sys-
tems that are more flexible and adaptable, allowing them to respond quickly
to changing conditions and requirements.
 Expectations from modern evolutionary approaches for image processing 247

7. Improved performance: Optimization techniques help to improve the per-

formance of engineering systems by finding the optimal solution. This can
lead to improved efficiency, reduced costs, increased productivity, and better
quality of output.
8. Increased robustness: Optimization techniques can help to improve the
robustness of engineering systems by identifying potential weaknesses and
finding solutions to mitigate them. This can lead to systems that are more
resilient to changes in the environment or unexpected events.
9. Ability to handle complexity: Engineering systems can be highly complex
and difficult to analyze using traditional methods. Optimization techniques,
however, are well-suited to handle complex systems and can help to identify
the best solutions even in highly complex scenarios.
10. Reduced costs: Optimization techniques can help to reduce costs in engi-
neering applications by identifying the optimal solution that meets the desired
performance requirements at the lowest cost. This can lead to significant
savings in material costs, manufacturing costs, and maintenance costs.
11. Increased innovation: Optimization techniques can help to promote inno-
vation in engineering applications by exploring new design concepts and
solutions that may not have been considered using traditional methods. This
can lead to new and improved products, processes, and systems that are more
efficient and effective.
12. Improved image quality: Optimization techniques can be used to improve
the quality of images by optimizing the parameters of image processing
algorithms. For example, optimization can be used to find the optimal set of
parameters for image denoising, image segmentation, or image registration
algorithms.
13. Efficient feature selection: In image processing, optimization techniques can
be used to reduce the dimensionality of the feature space and improve the
efficiency of subsequent processing steps. This can result in more precise and
faster image analysis.
14. Increased accuracy: Optimization techniques can be used to improve the
accuracy of image processing algorithms by finding the most efficient
solution. For example, optimization can be used to find the best set of
parameters for object detection, object recognition, or image classification
algorithms.
15. Automated optimization: Optimization techniques can be used to automate
the process of parameter tuning in image processing algorithms. This can save
time and reduce the need for manual intervention, leading to increased effi-
ciency and productivity.
16. Robustness: Optimization techniques can be used to improve the robustness
of image processing algorithms by finding solutions that are less sensitive to
changes in the input data or environmental conditions.
17. Customization: Optimization techniques can be used to customize image
processing algorithms to specific applications and data types. This can lead to
improved performance and accuracy in a variety of image-processing tasks.
248 Nature-inspired optimization algorithms and soft computing

Overall, optimization techniques are an essential tool for engineers in a wide

range of applications. They provide a powerful means of improving the efficiency,
performance, safety, and sustainability of engineering systems, while also reducing
costs and accelerating the development process.

9.8 Conclusion

Nature-inspired optimized algorithms such as genetic algorithms, particle swarm

optimization, and ant colony optimization have proven to be effective tools for
solving complex optimization problems in a variety of fields such as engineering,
finance, and healthcare. These algorithms are inspired by natural phenomena and
biological systems, and they use natural selection and evolution to find the best
solution. Constrained and unconstrained optimization algorithms have distinct
advantages and are appropriate for a variety of problems. Constrained optimization
algorithms take constraints into account when determining the best solution,
whereas unconstrained optimization algorithms do not. When the problem has
specific constraints, constrained optimization algorithms are more appropriate,
whereas unconstrained optimization algorithms are faster and more efficient.
Design is one of the engineering applications of optimization algorithms.

References

[1] Oliva D, Cuevas E, Pajares G, Zaldivar D, and Perez-Cisneros M. Multilevel

thresholding segmentation based on harmony search optimization. J Appl
Math 2013;2013. doi:10.1155/2013/575414.
[2] Abualigah L. Group search optimizer: a nature-inspired meta-heuristic opti-
mization algorithm with its results, variants, and applications. Neural Comput
Appl 2021;33:2949–2972. https://doi.org/10.1007/s00521-020-05107-y.
[3] Patra S, Gautam R, and Singla A. A novel context-sensitive multilevel
thresholding for image segmentation. Appl Soft Comput J 2014;23:122–127.
doi:10.1016/j.asoc.2014.06.016.
[4] Jia H, Ma J, and Song W. Multilevel thresholding segmentation for color
image using modified moth-flame optimization. IEEE Access 2019;7:
44097–44134. doi:10.1109/ACCESS.2019.2908718.
[5] Otsu N. A threshold selection method from gray-level histograms. IEEE
Trans Syst Man Cybernet 1979;9:62–66. doi:10.1109/TSMC.1979.4310076.
[6] Kwon SH. Threshold selection based on clustering analysis. Pattern Recogn
Lett 2004;25:1045–1050. doi:10.1016/j.patrec.2004.03.001.
[7] Resma KPB. and Nair MS. Multilevel thresholding for image segmentation
using Krill Herd Optimization algorithm. J King Saud Univ Comput Inf Sci
2021;33(5):528–541. doi:10.1016/j.jksuci.2018.04.007.
[8] Diego Olivaa EC, Gonzalo Pajares DZ, and Valentı́n Osuna. A Multilevel
thresholding algorithm using electromagnetism optimization. Neurocomputing
2014;139:357–381.
 Expectations from modern evolutionary approaches for image processing 249

[9] Hammouche K, Diaf M, and Siarry P. A comparative study of various meta-

heuristic techniques applied to the multilevel thresholding problem. Eng
Appl Artif Intell 2010;23:676–688.
[10] Bhandari AK, Kumar A, Chaudhary S, and Singh GK. A novel color image
multilevel thresholding based segmentation using nature inspired optimiza-
tion algorithms. Expert Syst Appl 2016;63:112–133.
[11] Wolper DH and Macready WG. No free lunch theorems for optimization.
IEEE Trans Evol Comput 1997;1(1):67–82.
[12] Kennedy J and Eberhart R. Particle swarm optimization. Proc IEEE Int Conf
Neural Netw 1995;4:1942–1948.
[13] Socha K and Dorigo M. Ant colony optimization for continuous domains.
Eur J Oper Res 2008;185(3):1155–1173.
[14] Passino KM. Biomimicry of bacterial foraging for distributed optimization
and control. IEEE Control Syst Mag 2002;22(3):52–67.
[15] Karaboga D and Basturk B. A powerful and efficient algorithm for numer-
ical function optimization: artificial bee colony (ABC) algorithm. J Global
Optim 2007;39(3):459–471.
[16] Mirjalili S, Mirjalili SM, and Lewis A. Grey wolf optimizer. Adv Eng Softw
2014;69:46–61.
[17] Mirjalili S. Moth-flame optimization algorithm: a novel nature-inspired
heuristic paradigm. Knowl-Based Syst 2015;89:228–249.
[18] Yu JJQ and Li VOK. A social spider algorithm for global optimization. Appl
Soft Comput 2015;30:614–627.
[19] Yang X-S. Multiobjective firefly algorithm for continuous optimization. Eng
Comput 2013;29(2):175–184.
[20] Mirjalili S and Lewis A. The whale optimization algorithm. Adv Eng Softw
2016;95:51–67.
[21] Mirjalili S. SCA: a sine cosine algorithm for solving optimization problems.
Knowl-Based Syst 2016;96:120–133.
[22] Gandomi AH and Alavi AH. Krill herd: a new bio-inspired optimization
algorithm. Commun Nonlinear Sci Numer Simul 2012;17:4831–4845.
[23] Yang X. A new metaheuristic bat-inspired algorithm. In Proceedings of the Nature
Inspired Cooperative Strategies for Optimization (NICSO), pp. 65–74, 2010.
[24] Yang XS. Flower pollination algorithm for global optimization. In
Proceedings of the Unconventional Computing and Natural Computation,
pp. 240–249, September 2012.
[25] Mirjalili S, Mirjalili SM, and Hatamlou A. Multi-verse optimizer: a nature-
inspired algorithm for global optimization. Neural Comput Appl 2016;27
(2):495–513.
[26] Chen K, Zhou Y, Zhang Z, Dai M, Chao Y, and Shi J. Multilevel image
segmentation based on an improved firefly algorithm. Math Problems Eng
2016;26:1–12.
[27] Akay B. A study on particle swarm optimization and artificial bee colony
algorithms for multilevel thresholding. Appl Soft Comput 2013;13(6):3066–
3091.
250 Nature-inspired optimization algorithms and soft computing

[28] Agarwal P, Singh R, Kumar S, and Bhattacharya M. Social spider algorithm

employed multi-level thresholding segmentation approach. In Proceedings
of the 1st International Conference on Information Communication
Technology for Intelligence Systems vol. 2, pp. 249–259, 2016.
[29] Khairuzzaman AKM and Chaudhury S. Multilevel thresholding using grey
wolf optimizer for image segmentation. Expert Syst Appl 2017;85:64–76.
[30] Srikanth R and Bikshalu K. Multilevel thresholding image segmentation based
on energy curve with harmony Search Algorithm. Ain Shams Eng J 2021;12
(1):1–20.ISSN 2090-4479,https://doi.org/10.1016/j.asej.2020.09.003.
[31] Chen Wei and Kangling F. Multilevel thresholding algorithm based on
particle swarm optimization for image segmentation. In 2008 27th Chinese
Control Conference, Kunming, 2008, pp. 348–351, doi:10.1109/CHICC.
2008.4605745.
[32] Diego Oliva EC, Gonzalo Pajares, and Valentı́n Osuna DZ. A multilevel
thresholding algorithm using electromagnetism optimization. Neurocomputing
2014;139:357–381.doi.org/10.1016/j.neucom.2014.02.020.
[33] Abu Khurma R, Aljarah I, Sharieh A, Abd Elaziz M, Damaševičius R, and
Krilavičius T. A review of the modification strategies of the nature inspired
algorithms for feature selection problem. Mathematics 2022;10:464. https://
doi.org/10.3390/math10030464
 Chapter 10
Conclusion
Aakansha Garg1, Varuna Gupta2, Vandana Mehndiratta2,
Yash Thakur2 and Dolly Sharma1

10.1 Concluding remarks

With the advancement in technology, day-to-day life has become a lot easier.
Resources requirement and task complexity are also increasing day by day. To
imbibe these requirements with less effort, using minimum resources is the most
adaptable solution. Optimization is performed to satisfy all these demands while
reducing overall cost, complexity, and enhancing efficiency. Some problems are
deterministic in nature like refrigerator cooling off mode, iron heating off mode,
etc. Contrary to this, most of the real-world problems are stochastic processes for
example traffic light controlling based on number of vehicles on the road, early
diagnosis of heart diseases based on various parameters, identify number of cellular
users who may demand network resources at same instance of time, image scenario
like background, posture, etc., while using image processing and assuming human
speed of working in a factory with automation processes and speed of other vehi-
cles for an automated vehicle on the road. Maximum real-world problems are
indeterministic in nature. Traditional methods like Newton’s method, Secant
method, Steffensen’s method, Principal component analysis [1], and Brent’s
method [2]. These are computation methods to find only the exact solutions for
real-world problems [3]. These methods always follow the traditional way only and
are incompatible to any spontaneous changes.
Since the last few decades, researchers are more focused toward optimization.
It may be done in different ways: To analyze a random data in much discrete way,
machine learning algorithms are used. Some examples are K-means [4], K-nearest
neighbor [5], reinforcement learning [6], Gaussian mixture model [7], Artificial
neural networks [8], Hidden Markov Model [9], logistic regression [10], and sup-
port vector machine [11]. To reduce the cost and latency of transmission, com-
pression algorithms like discrete cosine transform, discrete wavelet transform, and
Lempel-Ziv-Welch (LZW) [12] are used. To ensure security, encryption algorithms

1
ABES Institute of Technology, India
2
Christ (Deemed to be University) – Delhi, India
252 Nature-inspired optimization algorithms and soft computing

like advanced encryption standard, data encryption standard, and triple data
encryption standards are used [13]. To optimize parameters like energy consump-
tion and other real-time requirements, numerous other algorithms have been used
nowadays.
There are various problems to be solved in a specified steps or time. For
example, an autonomous car path is difficult to design as the condition of the road,
and other vehicles movements are non-deterministic. It is also difficult to predict
the time when movement of vehicles will be predictable. Such problems are called
NP-hard problems. To solve such problems, nature-inspired algorithms provide
satisfactory solutions [14]. Based on the inspiration source, nature-inspired algo-
rithms are categorized as follows. Stochastic algorithms are based on random
search, for example, stochastic hill climbing and Tabu search. Evolutionary algo-
rithms replicate the behavior of living organisms, for example, genetic algorithm
and strength Pareto evolutionary algorithm. Physical algorithms are based on
physical phenomena such as simulated annealing and harmony search. Bio-inspired
algorithms are inspired by Darwin’s theory of survival. Swarm-based algorithms
are based on behavior of social creatures, for example, ant colony optimization
and particle swarm optimization. Chemistry-based algorithms are based on opti-
mization of chemical reactions, for example, grenade explosion method and river
formation dynamics.
The nature-inspired algorithms has been adopted from different fields of sci-
ence, i.e., physics, chemistry, biology, mathematics, social science, etc. These
algorithms can be used to optimize single-objective problems and multi-objectives
problems [15]. Due to the nature-inspired algorithms’ precision, its application area
has been increased a lot. Genetic algorithm along with heuristic and machine
learning algorithm has been used to optimize various parameters like cost, energy
consumption, and latency using D2D communication for an IoT, 5G, and 6G sys-
tem[16]. A neoteric gravitational search algorithm for multi-constraint optimization
problem has been used to optimize throughput of served users in UAV commu-
nication [17]. To automate a 3D packaging system, particle swarm optimization
along with IoT has been designed [18]. The hill climbing algorithm has been used
for maximum power point tracking in a photovoltaic power harvesting system for
smart nodes of Internet of Things (IoT) [19]. An ensemble learning methods are
used as base classifier to identify the cross-semantic and structural features of cloud
to avoid massive data transmission in cloud computing systems [20]. Improved
simulated annealing is applied to physiological detection and positioning of a
visitor into the park [21]. The variety of bio-inspired algorithms is far beyond the
scope of our explanation.
Particle swarm optimization (PSO) is a random search algorithm based on
swarm, and it is simpler than other nature-inspired algorithms in terms of complexity
and fast convergence [22]. PSO has been implemented to solve many real-time pro-
blems like, PSO with gradient descent method and support vector machine has been
used to identify malicious data [23]. PSO has been used to allocate resources auto-
matically to a distributed data computation for large data computations. It ensures that
the subgroups will get resources directly proportional to computations [24]. The
 Conclusion 253

meta-heuristic PSO has been used to make decisions for node deployment fulfilling
multiple objectives such as QoS for energy consumption, delay, and throughput,
ensuring failure tolerance [25]. Since past two decades, PSO has been used in bio-
medical field and image processing. A single-particle optimizer-based local search
has been used for DNA sequence compression [26]. It is evident that in some appli-
cations, performance of PSO has been further improved by integrating with some
other algorithms. A hybridization of PSO with dragon fly optimization and gray wolf
optimization has been used to find a feature set for early detection of Alzheimer’s
disease [27]. A multi-swarm technique further improves the overall outcomes for a
problem. Initially, swarms are used for unconstrained local optimization. The results
are further optimized globally using constrained PSO [28]. For optimum utilization of
resources, PSO has been proposed. It optimizes routes based on arrival time and
waiting tolerance of the passengers [29]. Thus, PSO is applied to solve a large set of
engineering problems. To further improve the performance, these can be hybridized
with other algorithms. PSO and its allies can be used for linear, non-linear, uncon-
strained, and constrained problem optimizations. These algorithms can be applied for
solving single- and multi-objective problems.
Conventional nature-inspired algorithms and swarm optimization can be used
for optimization. IoT, 5G, and 6G systems are becoming more and more complex.
Conventional nature-inspired optimization algorithms like genetic algorithms,
PSO, and simulated annealing improves overall performance of the system. To
enhance it further, other optimization algorithms have been adapted. Cat swarm
optimization algorithm and its allies versions can be applied in various fields like
wireless sensor networks, electrical engineering, and communication engineering.
These algorithms are based on cat lazy behavior contrary to her alert mode, and
these are considered as seeking (local optimization) and tracing modes (global
optimization) [30]. Cuckoo optimization algorithm is generally used for continuous
problems, which are non-linear in nature. These are based on the cuckoo’s habit to
lay eggs in others’ nests and selecting the best nest where her eggs can nurture
without fail. This algorithm is used for complex problems, which are NP hard like
data fusion in wireless networks, train neural network, manufacturing scheduling,
and nurse scheduling [31]. Mine blast optimization algorithm (MBA) is based on
clearing the mines field while targeting the most explosive (optimum) mine in
which created pieces of shrapnel collide with other mines this is used as direction
for further process [32]. This algorithm is good for constrained optimization pro-
blems. MBA, its hybrid, and advanced versions has been used to find solutions to
various real-time problems. Water cycle algorithm (WCA) is based on the flow of
the river, i.e., how the river is formed at high mountains due to snowfall, flows
down to the valley, and ends up in the sea. This algorithm is widely used in various
mechanical engineering design problems [33]. WCA and its various hybrid and
advanced versions have been applied to solve a large number of engineering pro-
blems like spam e-mail detection, PI controller, and energy optimization [34].
The algorithms discussed till now can perform better but there is a limitation
that if an algorithm is stuck into the premature stage of local optima, then an
optimal value may not be obtained. This problem can be solved by using anarchic
254 Nature-inspired optimization algorithms and soft computing

society optimization algorithms (ASA). It is based on the human behavior of a

group of people who dislike stopping instead to keep visiting places which are less
visited by them. In engineering, it is less popular till now. It has been used to assign
various factories and machines to complete jobs in an area [35]. Teaching learning
method is a population-based algorithm which optimize in the two phases, i.e.,
teacher’s phase to teach students, student’s phase, i.e., to learn by other students. It
has been applied to many mechanical engineering design problems [36]. Another
meta-heuristic algorithm is inspired by the lion and ant intelligence. A lion sets a
trap for its prey and verifies the ant’s movement in its trap, whereas ant tries to
escape the trap by her intelligence. This algorithm can be applied to a large variety
of engineering applications like in wireless sensor networks, signal processing, PV
cells modeling, multilayer perceptron neural network, etc. [37]. Crow search
algorithm is based on the fact that a crow follows another crow to steal it. It has a
wide area of application in engineering like optimal power flow, feature selection,
image processing, and cloud computing [38]. It is evident that these advanced
meta-heuristic algorithms can be used to solve a wide range of engineering pro-
blems. Algorithms can be selected based on the problem type.
The nature-inspired algorithms are used in a large variety of application areas.
IoT is one of the applications in which optimization of resources, data, time, network,
battery, size of the device, etc. need to be optimized. Optimization helps to accom-
modate a greater number of users using the same resources which improves overall
efficiency and cost of the system. Various algorithms like genetic algorithm, simulated
annealing algorithm, differential evolution algorithm, particle swarm optimization, etc.
are used to optimize position of access point has been implemented [39]. Crowd-based
intelligence and genetic algorithms have been used to improve communication,
caching and communication capacity the device correlation and distance for
communication is optimized [40]. An discrete and evolutionary multi-objective
method has been used to improve the throughput and energy efficiency of the system
of a multi-UAV communication system [41]. The Hawk optimization algorithm,
genetic algorithm, artificial bee colony algorithm, moth flame optimization, and
whale optimization algorithms have been used to improve energy utilization in the
Internet of Things (IoT) applications [42]. The swarm intelligence-based algorithms,
i.e., flower pollination algorithm, slap swarm algorithm, and sine cosine algorithms
are implemented to calculate docking points for efficient data collection UAV [43].
Thus, various algorithms have been used to optimize different parameters for several
applications. The IoT technologies are further being used in various areas like
medical science, smart cities, etc. In smart cities, expectation is to sense various data,
software automation while using optimal usage of resources. A multi-swarm differ-
ential evolution particle swarm optimization is used to reduce energy utilization
without altering the target task [44]. To implement smart healthcare systems in smart
cities mobile cloud computing is used. This technology helps to reduce latency and
resource consumption by offloading the task computation to nearby devices. Other
optimization algorithms such as ant colony optimization and particle swarm optimi-
zation for optimal data offloading have been used [45]. Bat algorithm and fuzzy logic
are used to optimize energy consumption and comfort index in the large cities [46].
 Conclusion 255

Using other neighborhood devices for computation may fasten the process, but it may
interfere with the user’s privacy. Various security encryption algorithms are available
to ensure user’s privacy but ensuring privacy of user along with cloud-based power
system results best. An identity-based encryption with equality test ensures no leak-
age of data to ensure user’s security [47].
Apart from above various nature-inspired algorithms also finds applications in
the fields of biomedical, healthcare, and industrial automation. For early detection
of diseases various features have been incorporated in wearable devices like
smartwatches and smart phones. Algorithms like GA, swarm intelligence, neural
networks, and these elites are used for predictive analysis and to optimize the
resource [48]. These algorithms are also widely used in industrial automations like
warehouse replenishment where each time position allocations of a good is a new
permutation which very well mimics the behavior of chromosomes. Fault diagnosis
of rotation machinery is optimized by quantum genetic algorithm [49]. A robotic
assembly planning has been implemented using GA, ant colony optimization,
memetic algorithm, flower pollination algorithm, and teaching learning algorithms
[50]. The scope is beyond mentioned areas, i.e., nature-inspired algorithms can be
applied for optimization of a large variety of real-time problems.

10.2 Challenges and potentials of bio-inspired

optimization algorithms for IoT applications

Bio-inspired optimization algorithms are computational methods that simulate

natural biological processes to solve complex problems like mixed integer non-
linear programming problems, constraint-based problems, and NP hard problems.
These algorithms have shown great potential in various fields, including the IoT.
However, some challenges and limitations need to be addressed to fully exploit
their potential in IoT applications.

10.2.1 Challenges
Scalability: As IoT systems continue to grow in size and complexity, the scalability
of bio-inspired optimization algorithms becomes a significant challenge. These
algorithms may take a long time to converge or sometimes during local optimiza-
tion, a wrong convergent point may lead to no convergence during global optimi-
zation. For smart applications, time and resource optimization is desirable.
Real-time constraints: Real-time applications require a flexible system which may
adapt to real-time changes, which is a limitation as many bio-inspired optimization
algorithms are computationally complex.
Robustness: IoT applications often operate in dynamic and unpredictable envir-
onments. Therefore, the robustness of bio-inspired optimization algorithms to
changing conditions and perturbations is crucial.
Interpretability: Bio-inspired optimization algorithms are often considered as
“black-box” approaches, which may limit their interpretability and applicability in
certain IoT domains.
256 Nature-inspired optimization algorithms and soft computing

10.2.2 Potentials
Adaptability: Bio-inspired optimization algorithms are inherently adaptable and
can dynamically adjust to changes in the environment, making them suitable for
IoT applications that operate in dynamic and unpredictable environments.
Efficiency: Bio-inspired optimization algorithms are often highly efficient, making
them suitable for IoT applications with limited computational resources.
Multimodal optimization: Many IoT applications involve the optimization of
multiple objectives or criteria. Bio-inspired optimization algorithms can handle
such problems efficiently, making them suitable for IoT applications with complex
optimization objectives.
Diversity: Bio-inspired optimization algorithms can generate diverse solutions,
making them suitable for IoT applications that require diverse solutions to be
explored.
In summary, bio-inspired optimization algorithms have great potential for
IoT applications, but they also face challenges that need to be addressed to fully
exploit their potential. Researchers need to modify existing algorithms or propose
new algorithms and techniques that can overcome these challenges while lever-
aging the potentials of these algorithms to address the optimization problems in
IoT applications.

10.3 Challenges and opportunities of bio-inspired

optimization algorithms for biomedical applications

Bio-inspired optimization algorithms are computational techniques that mimic

natural processes to solve complex optimization problems. They have shown tre-
mendous potential in a wide range of biomedical applications, including drug
design, medical imaging, and bioinformatics. However, these algorithms also face
several challenges that need to be addressed to realize their full potential.
Lack of transparency is one of the significant challenges in the optimization
process. Bio-inspired optimization algorithms are often referred to as “black-box”
algorithms, which means that the decision-making process is not transparent. This
can be a challenge in biomedical applications where transparency is essential for
regulatory compliance and ethical considerations. Therefore, it is crucial to develop
explainable optimization algorithms that can provide a clear rationale for the
solutions they generate.
Lack of reliability is another drawback as sometimes bio inspired algorithms
are stuck into false local convergence which may lead to wrong global con-
vergence. For critical applications, it may cause a great detriment. So, in spite of all
virtues, it is difficult to adapt these algorithms for critical bio-medical applications.
Additionally, the optimization of biomedical problems often involves working
with noisy and uncertain data. Bio-inspired optimization algorithms may not per-
form well when working with noisy data, and it may be challenging to distinguish
between random noise and significant signal. Therefore, it is essential to develop
optimization algorithms that can handle noisy and uncertain data effectively.
 Conclusion 257

Accuracy of measured data is one of the important factors for response of

nature inspired algorithms. As these are adaptable to dynamic changes. Any mea-
sured changes of data at any instance may mislead to a convergent point. It is
difficult to determine the point where the device measured abrupt data.
Despite these challenges, bio-inspired optimization algorithms present numerous
opportunities for biomedical applications. For example, they can be used to develop
personalized medicine by optimizing treatment plans for individual patients based on
their genetic and clinical data. They can also be used to optimize the design of
medical devices and prosthetics, improving patient outcomes and quality of life.
Additionally, bio-inspired optimization algorithms can help in the discovery of new
drugs and therapies by searching vast chemical and biological spaces.
In conclusion, while bio-inspired optimization algorithms present numerous
challenges in biomedical applications, they also offer significant opportunities.
Addressing these challenges and developing more sophisticated algorithms will be
crucial in realizing the full potential of bio-inspired optimization in the
biomedical field.

10.4 Recent trends in smart cities planning based on

nature-inspired computing
Nature-inspired computing is a new change in smart city planning that aims to use
principles derived from natural systems to optimize and improve the efficiency of
urban systems. Some recent trends in this area include:
Swarm intelligence: One of the key principles of swarm intelligence is to mimic
the behavior of social insects, such as ants and bees. This idea is being used in
smart city planning to improve waste management, make traffic flow better, and
use less energy. For example, researchers are developing algorithms that mimic the
foraging behavior of ants to optimize garbage collection routes in cities.
Genetic algorithms: Genetic algorithms are inspired by the process of natural
selection and evolution. They are being used in smart city planning to optimize
complex systems such as transportation networks, energy grids, and water man-
agement systems. For example, researchers are developing genetic algorithms to
optimize the placement of electric vehicle charging stations in cities.
Artificial neural networks: Artificial neural networks are computational models
that are inspired by the structure and function of biological neurons. They are being
used in smart city planning to analyze data from sensors and other sources to
identify patterns and optimize urban systems. For example, researchers are using
artificial neural networks to optimize traffic light timings in cities.
Particle swarm optimization: Particle swarm optimization is a nature-inspired
algorithm that is based on the behavior of social insects, such as flocking birds. It is
being used in smart city planning to optimize a wide range of systems, including
transportation networks, energy grids, and water management systems. For exam-
ple, researchers are using particle swarm optimization to optimize the placement of
renewable energy sources in cities.
258 Nature-inspired optimization algorithms and soft computing

Other algorithms: Cat optimization, flower pollination optimization, lion and ant
optimization, etc. are nature-inspired algorithms that are based on some virtues of
animals. For example, cat optimization uses the virtue of the cat being attentive while
being in sleep mode for a large amount of time. These algorithms are used in smart
city planning to optimize power consumption, network management resources, etc.
Overall, nature-inspired computing is an exciting and rapidly evolving field
that is helping to drive innovation in smart city planning. By leveraging the prin-
ciples of natural systems, researchers and planners are developing new algorithms
and techniques that can help cities become more sustainable, efficient, and livable.
Although there are a number of algorithms in the IoT. IoT is a network of
interconnected devices that collect and exchange data. IoT applications can range
from smart homes and wearables to industrial monitoring and smart cities. To
process and analyze the vast amount of data generated by these devices, efficient
and effective algorithms are essential. Here are some of the most commonly used
algorithms in IoT applications:
Machine learning algorithms: Machine learning algorithms are widely used in
IoT applications for predicting, classifying, and clustering data. These algorithms
use statistical models to analyze the data generated by IoT devices, enabling them
to make accurate predictions and decisions.
Data compression algorithms: As IoT devices generate vast amounts of data, it is
often necessary to compress this data to reduce storage and transmission costs. Data
compression algorithms such as Huffman coding and Lempel-Ziv-Welch (LZW)
compression are commonly used in IoT applications.
Encryption algorithms: With the increasing amount of sensitive data generated by
IoT devices, encryption algorithms such as advanced encryption standard (AES)
and RSA are crucial for securing data and protecting privacy.
Optimization algorithms: Optimization algorithms are used to improve the per-
formance of IoT devices and systems. These algorithms aim to maximize efficiency
while minimizing energy consumption and other resources.
Sensor data fusion algorithms: IoT applications often involve data from multiple
sensors. Sensor data fusion algorithms are used to combine the data from these
sensors and provide a more accurate representation of the physical environment.
Time series analysis algorithms: IoT applications often generate time series data,
which can be analyzed using time series analysis algorithms. These algorithms can
detect patterns, anomalies, and trends in the data, enabling more accurate predic-
tions and decisions.
Overall, the algorithms used in IoT applications vary depending on the specific
use case, but the ones mentioned above are some of the most commonly used.

10.5 Future perspectives of nature-inspired computing

Nature-inspired computing has already shown great promise in solving complex
problems in various fields, including computer science, engineering, and biology.
 Conclusion 259

As technology continues to advance, we can expect to see further developments

and applications of this field. Some potential future perspectives of nature-inspired
computings are:
Integration with AI and machine learning: Nature-inspired computing techni-
ques, such as neural networks and genetic algorithms, can be combined with AI and
machine learning to create more efficient and effective systems. This integration
could lead to the creation of more intelligent machines capable of learning and
adapting to new situations.
Advancements in robotics: Nature-inspired computing could also be used to
improve the functionality of robots. For example, biomimicry could be used to
create robots that move and behave like animals, making them better suited for
tasks such as search and rescue missions.
Applications in healthcare: Nature-inspired computing could be used to develop
new medical treatments and devices. For example, genetic algorithms could be
used to optimize drug dosages for individual patients, while swarm intelligence
could be used to develop algorithms for analyzing medical images. Early detection
of disease is also possible using bio-inspired algorithms.
Sustainable development: Nature-inspired computing could also be used to help
address environmental issues, such as climate change and pollution. For example,
evolutionary algorithms could be used to optimize renewable energy systems, while
swarm intelligence could be used to develop efficient transportation networks.
Cybersecurity: Nature-inspired computing could also be used to improve cyber-
security. For example, immune system-inspired algorithms could be used to detect
and respond to cyber-attacks. An identity-based encryption with equality test
ensures user’s privacy [47].
Nature-inspired optimization techniques have become increasingly popular in
the field of IoT due to their ability to optimize complex systems with large amounts
of data. These techniques are based on the principles of natural selection, genetic
algorithms, and swarm intelligence. They can be used to improve the efficiency and
performance of IoT systems by optimizing various parameters such as energy
consumption, network traffic, and data processing.
One example of nature-inspired optimization in IoT is swarm intelligence, which
involves the use of algorithms inspired by the behavior of social insects, such as ants
and bees. Swarm intelligence algorithms can be used to optimize the routing of data
packets in IoT networks, resulting in more efficient and reliable data transfer.
Overall, the future of nature-inspired computing is exciting and holds great
potential for solving some of the most pressing problems such as latency mini-
mization for resilient responses are facing our world today.

10.6 Bio-inspired heuristic algorithms

Bio-inspired heuristic algorithms are computational techniques that draw inspira-
tion from the principles of natural systems such as biological evolution, swarm
260 Nature-inspired optimization algorithms and soft computing

intelligence, and neural networks to solve optimization problems. These algorithms

have been widely used in various fields, including IoT systems, due to their ability
to handle large and complex datasets.
In IoT systems, bio-inspired heuristic algorithms can be used for a wide range
of applications, including network routing, resource allocation, and energy man-
agement. Here are some examples of bio-inspired heuristic algorithms used in IoT:
Ant colony optimization (ACO): ACO is a bio-inspired optimization algorithm
that mimics the behavior of ants in finding the shortest path between their nest and
food sources. In IoT, ACO can be used to optimize the routing path of data packets
in wireless sensor networks.
Particle swarm optimization (PSO): PSO is a bio-inspired optimization algorithm
that simulates the social behavior of a flock of birds or a swarm of bees. In IoT,
PSO can be used to optimize the allocation of resources such as bandwidth,
memory, and processing power among IoT devices.
Genetic algorithms (GA): GA is a bio-inspired optimization algorithm that
mimics the processes of natural selection and evolution. In the IoT, GA can be used
to optimize the configuration of IoT devices and sensors, such as the selection of
the most suitable sensor parameters or the optimal placement of sensors.
Artificial neural networks (ANN): ANN is a bio-inspired algorithm that simulates
the function of the human brain in processing information. In IoT, ANN can be
used to predict the behavior of IoT devices and sensors based on historical data,
which can help in optimizing energy consumption and reduce maintenance costs.
In conclusion, bio-inspired heuristic algorithms have shown great potential in
solving optimization problems in IoT systems. As the complexity and size of IoT
systems continue to grow, the use of bio-inspired heuristic algorithms will become
even more important in ensuring the efficiency and reliability of these systems.

10.7 Probable future directions

There are several probable future directions in IoT optimization, some of which include:
Edge computing: IoT devices generate massive amounts of data, which can create
network congestion and latency issues. Edge computing involves processing data
closer to the source, reducing network traffic, and improving response times. In the
future, IoT devices will increasingly incorporate edge computing capabilities to
improve overall system performance.
Artificial intelligence and machine learning: AI and ML are already being used
to optimize IoT systems by analyzing data patterns and identifying anomalies. In
the future, AI and ML will become more sophisticated, enabling IoT systems to
learn from data and adapt to changing conditions in real-time.
5G networks: The next generation of wireless networks, 5G, promises faster
speeds and lower latency, which will significantly improve the performance of IoT
devices. 5G will enable real-time data processing and support a much larger
number of connected devices, making IoT systems more efficient and effective.
 Conclusion 261

Blockchain: The distributed ledger technology behind cryptocurrencies like bit-

coin, blockchain can be used to secure IoT data and transactions. In the future,
blockchain will likely be used to create more secure and decentralized IoT systems,
reducing the risk of cyber-attacks and data breaches.
Energy efficiency: IoT devices consume a significant amount of energy, which can
be expensive and harmful to the environment. In the future, IoT devices will be
designed to be more energy-efficient, incorporating features like low-power sensors
and optimized communication protocols to reduce power consumption.
Overall, these trends suggest that the future of IoT optimization will be focused
on making devices more intelligent, efficient, and secure, while also improving the
performance and reliability of the overall system.
Bio-medical: Nowadays many people are suffering from various diseases. Some
diseases result in sudden attack. An early detection of these diseases may be
helpful. So it is a need of an hour to design a system with quick response, accurate
with usage of minimum limited resources.
Localization of sensors: In smart cities, everything relates to IoT and to connect
physical devices with IoT or to automate the processes sensors need to measure all
kinds of data. During measurement, the position of the sensor plays an important in
accurate measurement.

References

[1] B. A. Draper, K. Baek, M. S. Bartlett, and J. R. Beveridge, “Recognizing

faces with PCA and ICA,” Comput. Vis. Image Underst., vol. 91, no. 1–2,
pp. 115–137, 2003, doi: 10.1016/S1077-3142(03)00077-8.
[2] R. M. Morais and J. Pedro, “Machine learning models for estimating quality
of transmission in DWDM networks,” J. Opt. Commun. Netw., vol. 10, no.
10, pp. D84–D99, 2018, doi: 10.1364/JOCN.10.000D84.
[3] M. Khajehzadeh, M. R. Taha, A. El-Shafie, and M. Eslami, “A survey on
meta-heuristic global optimization algorithms,” Res. J. Appl. Sci. Eng.
Technol., vol. 3, no. 6, pp. 569–578, 2011.
[4] M. R. McCall, T. Mehta, C. W. Leathers, and D. M. Foster, “Psyllium husk
II: effect on the metabolism of apolipoprotein B in African green monkeys,”
Am. J. Clin. Nutr., vol. 56, no. 2, pp. 385–393, 1992, doi: 10.1093/ajcn/
56.2.385.
[5] P. Cunningham and S. J. Delany, “K-Nearest neighbour classifiers: a tutor-
ial,” ACM Comput. Surv., vol. 54, no. 6, 2021, doi: 10.1145/3459665.
[6] M. Naeem, S. T. H. Rizvi, and A. Coronato, “A gentle introduction to
reinforcement learning and its application in different fields,” IEEE Access,
vol. 8, pp. 209320–209344, 2020, doi: 10.1109/ACCESS.2020.3038605.
[7] J. Zambrano, Gaussian Mixture Models – Method and Applications,
Presentation, FUDIPO project Machine Learning course, Oct-Dec 2017. doi:
10.13140/RG.2.2.32667.77602. https://www.researchgate.net/publication/
321245699_Gaussian_Mixture_Model_-_method_and_application.
262 Nature-inspired optimization algorithms and soft computing

[8] R. Dastres and M. Soori, “Artificial neural network systems,” Int. J. Imaging
Robot., vol. 2021, no. 2, pp. 13–25, 2021, www.ceserp.com/cp-jour
[9] L. Rabiner and B. Juang, “An introduction to hidden Markov models,”
in IEEE ASSP Magazine, vol. 3, no. 1, pp. 4–16, 1986. doi: 10.1109/
MASSP.1986.1165342.
[10] C. J. Peng, K. L. Lee, and G. M. Ingersoll, “An introduction to logistic
regression analysis and reporting,” The Journal of Educational Research, vol.
96, no. 1, pp. 3–14, 2002. doi: 10.1080/00220670209598786.
[11] T. Evgeniou and M. Pontil, “Support vector machines: theory and applica-
tions.” In: G. Paliouras, V. Karkaletsis, C. D. Spyropoulos (eds), Machine
Learning and Its Applications, ACAI 1999. Lecture Notes in Computer
Science, vol. 2049, 2001, Springer, Berlin, Heidelberg. https://doi.org/
10.1007/3-540-44673-7_12 10.1007/3-540-44673-7.
[12] G. Signoretti, M. Silva, P. Andrade, I. Silva, E. Sisinni, and P. Ferrari, “An
evolving TinyML compression algorithm for IoT environments based on
data eccentricity,” Sensors, vol. 21, no. 4153, pp. 1–25, 2021. doi: 10.3390/
s21124153.
[13] I. Kuzminykh, M. Yevdokymenko, and V. Sokolov, “Encryption algorithms
in IoT: security vs lifetime how long the device will encryption algorithms in
IoT: security vs lifetime,” preprint article, no. January 2022, 2021. https://
www.researchgate.net/publication/353237519_Encryption_Algorithms_in_
IoT_Security_vs_Lifetime _How_long_the_device_will_live.
[14] Z. Wang, C. Qin, B. Wan, and W. W. Song, “A comparative study of
common nature-inspired algorithms for continuous function optimization,”
Entropy, vol. 23(7), no. 874, pp. 1–40, 2021. doi: 10.3390/e23070874.
[15] F. G. Mohammadi, F. Shenavarmasouleh, K. Rasheed, T. Taha, M. H.
Amini, and H. R. Arabnia, “The application of evolutionary and nature
inspired algorithms in data science and data analytics,” in 2021 International
Conference on Computational Science and Computational Intelligence
(CSCI), 2021.
[16] A. Garg, R. Arya, and M. P. Singh, “An integrated approach for dual
resource optimization of relay-based mobile edge computing system,”
Concurr. Comput. Pract. Exp., no. February, pp. 1–16, March 2023, doi:
10.1002/cpe.7682.
[17] H. Li, J. Li, M. Liu, Z. Ding, and F. Gong, “Energy harvesting and resource
allocation for cache-enabled UAV based IoT NOMA networks,” IEEE
Trans. Veh. Technol., vol. 70, no. 9, pp. 9625–9630, 2021, doi: 10.1109/
TVT.2021.3098351.
[18] T. H. S. Li, C. Y. Liu, P. H. Kuo, et al., “A three-dimensional adaptive PSO-
based packing algorithm for an IoT-based automated e-fulfillment packaging
system,” IEEE Access, vol. 5, pp. 9188–9205, 2017, doi: 10.1109/
ACCESS.2017.2702715.
[19] X. Liu and E. Sanchez-Sinencio, “A highly efficient ultralow photovoltaic
power harvesting system with MPPT for Internet of Things smart nodes,”
 Conclusion 263

IEEE Trans. Very Large Scale Integr. Syst., vol. 23, no. 12, pp. 3065–3075,
2015, doi: 10.1109/TVLSI.2014.2387167.
[20] J. Zhang, P. Liu, F. Zhang, H. Iwabuchi, A. A. D. H. E. A. De Moura, and V.
H. C. De Albuquerque, “Ensemble meteorological cloud classification meets
internet of dependable and controllable things,” IEEE Internet Things J.,
vol. 8, no. 5, pp. 3323–3330, 2021, doi: 10.1109/JIOT.2020.3043289.
[21] C. C. Lin, W. Y. Liu, and Y. W. Lu, “Three-dimensional Internet-of-Things
deployment with optimal management service benefits for smart tourism
services in forest recreation parks,” IEEE Access, vol. 7, pp. 182366–
182380, 2019, doi: 10.1109/ACCESS.2019.2960212.
[22] A. Erturk, M. K. Gullu, D. Cesmeci, D. Gercek, and S. Erturk, “Spatial
resolution enhancement of hyperspectral images using unmixing and binary
particle swarm optimization,” IEEE Geosci. Remote Sens. Lett., vol. 11, no.
12, pp. 2100–2104, 2014, doi: 10.1109/LGRS.2014.2320135.
[23] J. Liu, D. Yang, M. Lian, and M. Li, “Research on intrusion detection based
on particle swarm optimization in IoT,” IEEE Access, vol. 9, pp. 38254–
38268, 2021, doi: 10.1109/ACCESS.2021.3063671.
[24] Q. Chen, J. Sun, and V. Palade, “Distributed contribution-based quantum-
behaved particle swarm optimization with controlled diversity for large-
scale global optimization problems,” IEEE Access, vol. 7, pp. 150093–
150104, 2019, doi: 10.1109/ACCESS.2019.2944196.
[25] M. Z. Hasan and H. Al-Rizzo, “Optimization of sensor deployment for
industrial internet of things using a multiswarm algorithm,” IEEE Internet
Things J., vol. 6, no. 6, pp. 10344–10362, 2019, doi: 10.1109/JIOT.2019.
2938486.
[26] Z. Zhu, J. Zhou, Z. Ji, and Y. H. Shi, “DNA sequence compression using
adaptive particle swarm optimization-based memetic algorithm,” IEEE
Trans. Evol. Comput., vol. 15, no. 5, pp. 643–658, 2011, doi: 10.1109/TEVC.
2011.2160399.
[27] Y. F. Khan, B. Kaushik, M. Khalid Imam Rahmani, and M. E. Ahmed, “HSI-
LFS-BERT: novel hybrid swarm intelligence based linguistics feature
selection and computational intelligent model for Alzheimer’s prediction
using audio transcript,” IEEE Access, vol. 10, no. November, pp. 126990–
127004, 2022, doi: 10.1109/ACCESS.2022.3223681.
[28] Q. Zhao and C. Li, “Two-stage multi-swarm particle swarm optimizer for
unconstrained and constrained global optimization,” IEEE Access, vol. 8, no.
1, pp. 124905–124927, 2020, doi: 10.1109/ACCESS.2020.3007743.
[29] F. Chen, Y. Lu, L. Liu, and Q. Zhu, “Route optimization of customized
buses based on optimistic and pessimistic values,” IEEE Access, vol. 11, no.
January, pp. 11016–11023, 2023, doi: 10.1109/ACCESS.2023.3241235.
[30] A. M. Ahmed, T. A. Rashid, and S. A. M. Saeed, “Cat swarm optimization
algorithm: a survey and performance evaluation,” Comput. Intell. Neurosci.,
vol. 2020, Article ID 4854895, 2020, doi: 10.1155/2020/4854895.
264 Nature-inspired optimization algorithms and soft computing

[31] L. C. C. By, S. Date, P. Date, C. Verma, C. S. Algorithm, and O. Problems, A

Survey on Cuckoo Search Algorithm for Optimization Problems, March,
pp. 0–16, 2021, doi: 10.36227/techrxiv.14199221.
[32] A. Sadollah, A. Bahreininejad, H. Eskandar, and M. Hamdi, “Mine blast
algorithm: a new population based algorithm for solving constrained engi-
neering optimization problems,” Appl. Soft Comput. J., vol. 13, no. 5,
pp. 2592–2612, 2013, doi: 10.1016/j.asoc.2012.11.026.
[33] H. Eskandar, A. Sadollah, A. Bahreininejad, and M. Hamdi, “Water cycle
algorithm – a novel metaheuristic optimization method for solving constrained
engineering optimization problems,” Comput. Struct., vol. 110–111, no.
September 2012, pp. 151–166, 2012, doi: 10.1016/j.compstruc.2012.07.010.
[34] G. Al-Rawashdeh, R. Mamat, and N. Hafhizah Binti Abd Rahim, “Hybrid
water cycle optimization algorithm with simulated annealing for spam e-
mail detection,” IEEE Access, vol. 7, pp. 143721–143734, 2019, doi:
10.1109/ACCESS.2019.2944089.
[35] A. Ahmadi-Javid, “Anarchic society optimization: a human-inspired
method,” in 2011 IEEE Congress of Evolutionary Computation (CEC), no.
March, pp. 2586–2592, 2011, doi:10.1109/CEC.2011.5949940.
[36] D. Wu, S. Wang, Q. Liu, L. Abualigah, and H. Jia, “An improved teaching-
learning-based optimization algorithm with reinforcement learning strategy
for solving optimization problems,” Comput. Intell. Neurosci., vol. 2022,
2022, doi: 10.1155/2022/1535957.
[37] A. A. Heidari, H. Faris, S. Mirjalili, I. Aljarah, and M. Mafarja, “Ant lion
optimizer: theory, literature review, and application in multi-layer percep-
tron neural networks,” Stud. Comput. Intell., vol. 811, no. August 2021,
pp. 23–46, 2020, doi: 10.1007/978-3-030-12127-3_3.
[38] Q. Cheng, H. Huang, and M. Chen, “A novel crow search algorithm based on
improved flower pollination,” Math. Probl. Eng., vol. 2021, pp. 1–26, 2021,
doi: 10.1155/2021/1048879.
[39] X. Zhu, W. Qu, T. Qiu, L. Zhao, M. Atiquzzaman, and D. O. Wu, “Indoor
intelligent fingerprint-based localization: principles, approaches and chal-
lenges,” IEEE Commun. Surv. Tutorials, vol. 22, no. 4, pp. 2634–2657,
2020, doi: 10.1109/COMST.2020.3014304.
[40] K. Wang, Z. Yang, B. Liang, and W. Ji, “An intelligence optimization
method based on crowd intelligence for IoT devices,” Int. J. Crowd Sci.,
vol. 5, no. 3, pp. 218–227, 2021, doi: 10.1108/IJCS-03-2021-0007.
[41] L. Liu, A. Wang, G. Sun, and J. Li, “Multiobjective optimization for
improving throughput and energy efficiency in UAV-enabled IoT,” IEEE
Internet Things J., vol. 9, no. 20, pp. 20763–20777, 2022, doi: 10.1109/
JIOT.2022.3175712.
[42] K. Dev, P. K. R. Maddikunta, T. R. Gadekallu, S. Bhattacharya, P. Hegde,
and S. Singh, “Energy optimization for green communication in IoT using
Harris Hawks optimization,” IEEE Trans. Green Commun. Netw., vol. 6, no.
2, pp. 685–694, 2022, doi: 10.1109/TGCN.2022.3143991.
 Conclusion 265

[43] E. Chen, J. Chen, A. W. Mohamed, B. Wang, Z. Wang, and Y. Chen,

“Swarm intelligence application to UAV aided IoT data acquisition
deployment optimization,” IEEE Access, vol. 8, pp. 175660–175668, 2020,
doi: 10.1109/ACCESS.2020.3025409.
[44] M. Sato, Y. Fukuyama, T. Iizaka, and T. Matsui, “Total optimization of
energy networks in a smart city by multi-population global-best modified
brain storm optimization with migration,” Algorithms, vol. 12, no. 1,
pp. 2186–2200, 2019, doi: 10.3390/a12010015.
[45] M. M. Islam, M. A. Razzaque, M. M. Hassan, W. N. Ismail, and B. Song,
“Mobile cloud-based big healthcare data processing in smart cities,” IEEE
Access, vol. 5, pp. 11887–11899, 2017, doi: 10.1109/ACCESS.2017.2707439.
[46] A. S. Shah, H. Nasir, M. Fayaz, A. Lajis, I. Ullah, and A. Shah, “Dynamic
user preference parameters selection and energy consumption optimization
for smart homes using deep extreme learning machine and bat algorithm,”
IEEE Access, vol. 8, pp. 204744–204762, 2020, doi: 10.1109/
ACCESS.2020.3037081.
[47] L. Wu, Y. Zhang, K. K. R. Choo, and D. He, “Efficient identity-based
encryption scheme with equality test in smart city,” IEEE Trans. Sustain.
Comput., vol. 3, no. 1, pp. 44–55, 2018, doi: 10.1109/TSUSC.2017.2734110.
[48] L. Kouhalvandi, L. Matekovits, and I. Peter, “Magic of 5G technology and
optimization methods applied to biomedical devices: a survey,” Appl. Sci.,
vol. 12, no. 14, 2022, doi: 10.3390/app12147096.
[49] X. Zhu, J. Xiong, and Q. Liang, “Fault diagnosis of rotation machinery based
on support vector machine optimized by quantum genetic algorithm,” IEEE
Access, vol. 6, pp. 33583–33588, 2018, doi: 10.1109/ACCESS.2018.2789933.
[50] A. Balamurali Gunji, B. B. B. V. L. Deepak, C. M. V. A. Raju
Bahubalendruni, and D. Bibhuti Bhushan Biswal, “An optimal robotic
assembly sequence planning by assembly subsets detection method using
teaching learning-based optimization algorithm,” IEEE Trans. Autom. Sci.
Eng., vol. 15, no. 3, pp. 1369–1385, 2018, doi: 10.1109/TASE.2018.2791665.
This page intentionally left blank
 Index

accuracy 166, 208 automated inventory management

active set methods 69 206, 208
adaptability 163 automated quality control 211
principle 95 automatic data quality control
adaptation 155 technology 211
adaptive PSO 104 automation technology 201
agriculture and environmental
optimization 224 bacterial foraging (BF) algorithm 5,
53, 75
aluminum alloys (AAs) 37
bacterial foraging optimization (BFO)
Alzheimer’s disease 178
162
anarchic society optimization
bandwidth factor (BF) 233
algorithms (ASO) 125–6, 253–4
bandwidth optimization 150
anisotropic diffusion 59
barrier-based hybrid constraint
ant colony optimization (ACO) 3,
handling 69
9–10, 15, 26–7, 32, 53, 56, 73,
96, 154, 158, 161, 176, 177, bat algorithm (BA) 15–17, 21, 53, 98,
220–1, 260 105, 222
ant lion optimizer (ALO) 126–8 bed capacity planning 176
ants 158 bee algorithm (BA) 155
approximation coefficients 182 bees 158
artificial bee colony algorithm (ABC) big bang big crunch (BBBC) 53
9–10, 16, 53, 56, 158, 161, big-bang weighted aggregation (BWA)
220–1, 254 100
optimization 24 big data 175, 187
artificial ecosystem-based optimization binary PSO algorithm 104
(AEBO) 222 bioinformatics 54, 178, 224
artificial immune system (AIS) 98 bio-inspired algorithms 15, 20, 167–8,
artificial intelligence (AI) 54, 89, 119, 252
173, 201, 260 bio-inspired but not swarm
artificial neural network (ANN) 105, intelligence based 29
178, 251, 257, 260 FPA 29–31
augmented Lagrangian method 67 classes of 20–1
268 Nature-inspired optimization algorithms and soft computing

swarm intelligence-based chromosomes 154

bio-inspired algorithms 20–2 civil engineering 54
ABC optimization 24 Classical Roulette Wheel selection
ACO 26–7 101
chicken swarm optimization 25–6 cloud computing 3, 175
CSA 28–9 Cloud of Things (CoT) 189
FFA 27–8 cluster head (CH) detection 32
PSO 23–4 clustering 36, 95
variants of swarm algorithms 29 colony optimization 219
bio-inspired heuristic algorithms 148, comparative analysis 166–8
259–60 complex diffusion 59
bio-inspired optimization algorithms complex interactions 149
(BIOAs) 53 complexity 166
biologically inspired optimization 160 computational complexity 156
biomedical engineering 224, 261 computed tomography (CT) 59, 179
biomedical image processing 179 computer-aided diagnosis (CAD) 178
biomedical informatics 178–9 computer vision (CV) 173
bird swarm algorithm (BSA) 187 computing optimized threshold levels
blockchain 261 233
Bluetooth 150 connected enterprise (CE) 209
brain cancer 174 constrained optimization 65
brain disorders 178 constraint handling methods 66
breast cancer 174 penalty function methods 65–6
Brent’s method 251 constraint handling techniques 68–9
Broyde Fletcher–Goldfarb–Shanno constraint programming (CP) 68
(BFGS) methods 11–12 constraints 55, 66, 243–5
constraint handling techniques 68–9
candidate solution 55
hybrid constraint handling
Canny edge detector 108 techniques 69–70
canonical PSO algorithm 88 linear penalty function 67–8
capsule endoscopy (CE) 179 penalty functions 67
cat swarm optimization algorithm quadratic penalty function 68
(CSO) 128, 222, 253
constraint satisfaction problems (CSPs)
chaotic distribution 29 68
chemical engineering 224 construction and project management
chemistry-based algorithms 252 224
chicken swarm optimization (CSO) 25–6 conventional weighted aggregation
Chimp optimization algorithm (CWA) 100
(CHOA) 186 convergence 106–7
 Index 269

criteria 55, 63–4 difficulty of tuning parameters 156

rate 160 digital image processing 58, 173–4
convolutional neural networks (CNNs) digital transformation 201
61, 178, 245 direct metal laser sintering (DMLS)
cost function 55, 227 203
cost optimization 150 disaster management and emergency
cost savings 166 response 224
Counts of Dimension to Change (CDC) discrete cosine transform 251
128 discrete PSO algorithm 103
COVID-19 178 discrete wavelet transform (DWT)
craziness 101 182, 251
crossover rate 63 distributed computing 164
crow search algorithm (CSA) 15, 129, diverse response principle 96
254 diversity 256
cuckoo optimization algorithm 253 dominant tree 101, 102
cuckoo search (CS) 221 doodlebugs 126
cuckoo search algorithm (CSA) 16, dragonfly algorithm (DA) 222
21, 28–9, 53, 57, 96, 129, 220 dynamic resource allocation 164
cutting-edge technology 201 dynamic task scheduling algorithm 3
cybersecurity 259
e-commerce 206
Darwin’s theory 9, 120 edge computing 148, 164, 260
data aggregation algorithms 151 edge detection 108
data analysis 162 edge linking 108
data collection 148 efficient exploration 156
data compression algorithms 258 emit behavior 21
data mining 219, 224 encryption algorithms 258
data privacy and security 149, 160 energy curve 226, 229
data reduction 164 equation of 229
decision-making 63, 176 energy efficiency 261
decision tree 61 energy-efficient ant-based routing
decision variables 55 (EEABR) protocol 32
deep belief network (DBN) 186 energy management 224
deep learning (DL) 54, 174–5, 245 energy optimization 147, 150, 224
deterministic algorithms 2 energy systems 223
device placement 162 engineering design optimization 223
diabetes 178 enterprise resource planning (ERP)
differential evolution (DE) 20, 53, 56, 202
105, 158, 220, 221 ethical considerations 157
270 Nature-inspired optimization algorithms and soft computing

evolutionary algorithms (EAs) 15, genetic algorithms (GA) 4, 9, 16, 19,

157, 167, 252 32, 53, 56, 75, 105, 148, 153–4,
evolutionary optimization algorithms 158, 161, 175, 219–21, 254,
119 257, 260
external irregularity index (EI) 126 feature selection based on 70–1
external repository 99 genomics 224
global optimization 156, 159
factory automation 202–3 global optimum 55
factory digitalization 201 glowworm algorithm (GSO) 16
fault diagnostics 174 gradient 55
fault tolerance 155 gradient-based optimization 2
feasible solution 55 grasshopper optimization algorithm
feature extraction 174 (GOA) 222
features classification 174 gravitational search (GS) 53
feature selection 107–8, 174, 184, gravitational search algorithm (GSA)
245–6 98
finance 223 gray level co-occurrence matrix
financial optimization 224 (GLCM) 182
fire fly algorithm (FFA) 16, 21, 27–8, gray wolf optimizer (GWO) 109, 222
53, 57, 98, 105, 161, 220–1 Greedy tours 7
Fisher’s Linear Discriminant 185 Grey Wolf Optimizer 127
fish swarm algorithm (FSA) 98 grid search 64
fitness function 55 gross domestic product (GDP) 175
5G networks 260 guaranteed convergence PSO
Fletcher–Reeves methods 11–12 (GCPSO) 104
flexibility 156, 159, 166 guided local search (GLS) 8–9
flower pollination algorithm (FPA) 17,
29–31, 222 Harlicks features 181
fourth order partial differential harmony memory (HM) 230
equation 59 and harmonic vector 232–3
frequency-division multiple access initialization 232
(FDMA) 151 harmony-memory consideration rate
friction stir welding (FSW) 37 (HMCR) 232
fuzzy-c-means (FCM) 186 harmony search algorithm
Fuzzy Delphi 177 (HSA) 221
harmony search optimization algorithm
Gabor filter 182 (HAS) 225
game theory and optimization 224 image segmentation with multilevel
thresholding 226–30
Gaussian distribution 29
Hawk optimization algorithm 254
Gaussian mixture model 251
 Index 271

healthcare accuracy 208

applications in 259 quality of delivery 209
systems 175–7 scalability 208
heart disorders 174 synchronization 208–9
heterogeneous devices 149 time efficiency 207–8
heuristic algorithms 18 logistics 215–16
heuristic optimization algorithms 6 packaging optimization 213–15
constructive heuristic optimization product flow monitoring 203
algorithms 7 creating applications with
Hidden Markov model 251 monitoring in mind from the
hill climbing method 3 start 204
histogram of oriented gradients (HOG) organize products into several
245 categories 204–5
histogram range 181 real-time tracking technologies
honey bee algorithm 5 205
Huber penalty function 67 quality control 211–13
human resource (HR) planning 177 safety and security 209
hunting search (HS) algorithm 98 supply chain optimization 215–16
hybrid ant colony optimization 69 industrial digitization 201
hybrid constraint handling techniques Industrial Internet of Things (IIoT)
(HCHTs) 69–70 201
hybrid genetic algorithm 69 infeasible solution 55
hybrid particle swarm optimization 69, Institute for Supply Management
104 (ISM) 205
hybrid system 178 intelligent optimization algorithms 16
Hybrid Whale (HW) Algorithm 106 interior-point methods 69
Internet of Medical Things (IoMT)
image and signal processing 223 188
image classification 60–1 Internet of Things (IoT) 55, 147, 206,
252, 254
image compression 108
application of optimization in 148–9
image denoising 58–9
bio-inspired heuristic algorithms
image enhancement 59, 173
160, 255, 259–60
image processing 54, 57
advantages of 163
image segmentation 59–60, 107
for biomedical applications 256–7
improvised algorithm (IFA) 226
challenges 255
individuals/entities 56
limitations of 163–4
industrial automation
potentials 256
factory digitalization 201
role of 162–3
inventory management 205
types of 161–2
272 Nature-inspired optimization algorithms and soft computing

comparative analysis 166–8 Krill Herd (KH) algorithm 98

evolutionary optimization in 157 Kuhn–Tucker conditions 12
advantages of 159–60
algorithms of 158 lack of generalizability 157
disadvantages of 160 lack of interpretability 160
role of 159 lack of transparency 157
load optimization in cognitive 164 Lagrange multiplier methods 12, 69
advantages of 165–6 Laplacian operator 108
disadvantages of 166 LEACH Protocol 37
uses of 165 least mean square algorithm (LMS)
nature-inspired computing 179
future perspectives of 258–9 Lempel-Ziv-Welch (LZW) 251
smart cities planning based on Levy’s distribution 29
257–8 linear discriminant analysis 60
nature-inspired optimization in 153 linear penalty function 67–8
advantages of 156 linear predictive coding (LPC) 183
algorithms of 153–5 linear programming 2
disadvantages of 156–7 liver diseases 178
role of 155 load forecasting 174
network optimization in 149 local optimum 55
advantages of 152 logistic regression method 185, 251
algorithms of 150–1 logistics 215–16
disadvantages of 152–3 logistics optimization 215
types of 150 LoRaWAN 150
interpretability 255 loss function 55
inventory management 205, 206 lung cancer 174
accuracy 208
quality of delivery 209 machine learning (ML) 8, 54, 159,
174, 219, 223, 224, 258–60
scalability 208
magnetic resonance imaging (MRI)
synchronization 208–9
59, 173, 179
time efficiency 207–8
mammography 173, 181–2
Island Migration approach 101
Mant 127
iterated local search (ILS) 8–9
manual tuning 64
Kapur’s methods 225, 234 Markov random field (MRF) 107
multilevel thresholding with 228–30 mathematical optimization
techniques 2
Kapurs’s entropy functions 226
mean square error (MSE) 179
key performance indicators (KPIs) 215
median absolute deviation (MAD) 181
K-nearest neighbor (KNN) 61, 185, 251
 Index 273

medical diagnosis 177–8 bio-inspired algorithms 20–31

MeI frequency cepstral coefficients classes of optimization algorithm 18
(MFCC) 183 on constraints engineering problem
metaheuristics optimization algorithms 31–45
2, 4–5, 7, 64, 153 evolutionary algorithms 19
implications of 5–6 differential evolution 20
population-based metaheuristic genetic algorithm 19
algorithms 9–11 extensive review of 16–17
trajectory-based metaheuristic general classification of 18–19
algorithms 8–9
optimization techniques 125
Metropolis-Hastings algorithm 57
anarchic society optimization
micro biopsy 59 125–6
micrometer 204 ant lion optimizer 126–8
mine blast optimization algorithm cat swarm optimization 128
(MBA) 130–1, 253
crow search algorithm 129
mixed integer programming (MIP)
cuckoo search 129
177
mine blast algorithm 130–1
monkey algorithm (MA) 222
water cycle algorithm 131–2
moth flame optimization (MFO) 222
physics- and chemistry-based
multilayer perceptron (MLP) 175, 178
algorithm 31
multimodal optimization 256
search for ideal algorithm 16
multi-objective optimization 224
nature-inspired optimization
problems 99 algorithms (NIOAs) 219
multi-objective PSO (MOPSO) 101–2 ant colony optimization 220, 221
multiplicative noise 59 application domains of 222
multispecies PSO 101 harmony search optimization
mutation rate (MR) 63 algorithm 225–33
mutual information (MI) 60 implementation 225
applications of 54
naive Bayes (NB) classifier 185 artificial bee colony algorithm 220,
natural inspiration 163 221
natural-inspired optimization artificial ecosystem-based
algorithms 148 optimization 222
natural language processing (NLP) bat algorithm 222
174 cat swarm optimization 222
natural science inspired optimization constrained and unconstrained
algorithms (NSIOAs) 53 optimization 241–3
nature-inspired algorithms 15, 120–5, constrained optimization 65
167
constraint handling methods 66
analysis of 17
274 Nature-inspired optimization algorithms and soft computing

penalty function methods 65–6 parameter tuning 63–4

constraints 66, 243–5 random search 64
constraint handling techniques monkey algorithm 222
68–9 moth flame optimization 222
hybrid constraint handling optimization
techniques 69–70 in image classification 60–1
linear penalty function 67–8 in image denoising 58–9
penalty functions 67 in image enhancement 59
quadratic penalty function 68 in image feature extraction and
on constraints engineering problem selection 60
31 in image segmentation 59–60
applied in solid-state wielding optimization techniques in
37–45 engineering applications 246–8
based clustering routing protocols parameter tanning 239–41
31–6
particle swarm optimization 220
implementation of WCA on leach
practical engineering applications
routing protocol 36–7
73–5
cuckoo search algorithm 220–1
unconstrained optimization 66
definition of 53
population-based methods 66
differential evolution 220, 221
single-point methods 66
dragonfly algorithm 222
whale optimization
feature selection 70, 245–6 algorithm 222
based on ABC 71 near-optimal solutions 159
based on ACO 71 network constraints 149
based on CS 71–3 network flow theory 2
based on FF 73 network optimization 147–8, 159, 167
based on GA 70–1 network resource allocation 148
firefly algorithm 220–1 network slicing 150
flower pollination algorithm 222 network virtualization 150
genetic algorithm 220, 221 neural networks (NN) 185
glowworm swarm optimization 221 Newton’s algorithm 11
grasshopper optimization algorithm Newton’s method 251
222
no-free-lunch (NFL) principle 226
gray wolf optimizer 222
nonlinear programming 2
harmony search algorithm 221
nonparametric techniques 225
implementation 61
normalized gray level variance (Nvar)
grid search 64 182
manual tuning 64 novel bee algorithm 5
metaheuristic optimization 64 NP-hard problems 252
 Index 275

objective function (OF) 55 needs for 3

operating rooms (OR) 176 search for optimality 2–3
optimization 95, 155 theoretical analysis 11
optimization algorithm (OA) 2, 16, 53, water cycle algorithm 131–2
175, 258 Orcas algorithm (OA) 98
based intelligent detection of heart Otsu’s methods 225–8, 234
disorders 186–7 overhead 166
in biomedical image processing 179
in biomedical informatics 178–9 packaging optimization 213–15
for ECG classification 179 parameter tanning 239–41
feature extraction and classification parameter tuning 61, 63–4, 159
180 parametric techniques 225
classification 185 Pareto optimality 99
by discrete wavelet transform 182 particle diversity 102
feature selection 184 particle swarm optimization (PSO)
by Gabor filter 182 3–4, 9–10, 15, 21, 23–4, 33, 53,
in mammography 181–2 56–7, 73, 85, 96, 154–5, 158,
in speech and pattern recognition 161, 177, 219, 220, 252, 257,
183 260
in healthcare systems 175–7 advantages and disadvantages 88–9
in medical diagnosis 177–8 algorithm 86–8
predictive analytics in healthcare applications 89
187 artificial life cycle 85
for smart healthcare 188–9 computational experiments 106
systematic approach for selection of convergence 106–7
11–12 different approaches of multi-
optimization problems 8 objective PSO 99
optimization techniques 1–2, 125 algorithms based on pareto
anarchic society optimization 125–6 dominance 101–3
ant lion optimizer 126–8 non-Pareto, vector-evaluated
cat swarm optimization 128 approach 100–1
crow search algorithm 129 objective function aggregation
approach 100
cuckoo search 129
objective function ordering
difference between metaheuristics
approach 100
optimization and heuristic
optimization 5 elements 86
heuristic optimization algorithms 6 evolutionary algorithm 85
metaheuristics optimization 4–5 feature selection based on 71
mine blast algorithm 130–1 in hybrid environment 105–6
276 Nature-inspired optimization algorithms and soft computing

on image processing problems principal component analysis 60, 251

107–8 protocol selection 150
for multi-objective problem 99 proximity principle 95
on optimum route-finding problems
108–9 quadratic penalty function 68
for single-objective problem 98 quality control 211–13, 224
standard pseudo code 88 quality of delivery (QoD) 209
swarm intelligence 89 quality of service (QoS) 150
division of work 95 quality principle 96
general swarm principles 95–8 Quick-Borůvka tours 7
self-organization 89–95
types of 97 random forest 61
variants of 103 random search techniques 5, 64
adaptive PSO 104 real-time constraints 255
binary PSO 104 real-time inventory data 208
discrete PSO 103 real-time processing 149
hybrid PSO 104 recurrent fuzzy neural network
immunity-enhanced PSO 104–5 (RFNN) 186
neighborhood-guaranteed recursive feature elimination 60
convergence PSO 104 region of interests (ROIs) 58, 60
neighborhood search strategies reinforcement learning 251
PSO 104 resource allocation algorithms 151,
PASCAL 176 159, 162, 164
pattern recognition 54 resource constraints 166
penalty-based hybrid constraint resource optimization 147
handling 69 risk management 210
penalty functions 67 robotics 54, 223, 259
perceptron algorithm 185 robustness 156, 159, 163, 180, 255
physical algorithms 252 route planning 108
PISTIL 29 routing 95
pitch adjusting rate (PAR) 232 algorithms 151
point-of-sale (POS) 206 optimization 147, 150
population size (PS) 63
positron emission tomography (PET) salp swarm algorithm (SSA) 186
59 scalability 156, 159, 160, 163, 208,
power system optimization 223 255
predictive algorithms 187 scale-invariant feature transform
predictive analytics algorithms 151 (SIFT) 60, 245
premature convergence 160 scheduling 95
 Index 277

Scout bee 24 standard score 181

Secant method 251 Steffensen’s method 251
security 155 stochastic algorithms 2, 252
algorithms 151 supply chain optimization 215–16, 224
optimization 150 support vector machine (SVM) 61,
Seeking Memory Pool (SMP) 128 178, 251
Selected Dimension (SRD) 128 support vector neural network (SVNN)
selection strategy 63 186
self-organization 21 surgery scheduling 186
Self-Position Considering (SPC) 128 sustainable development 259
sensitivity to initial conditions 157 swarm-based algorithms 252
sensor data fusion algorithms 258 swarm diversity 102
sensors swarm intelligence (SI) algorithm 89,
177, 257
localization of 261
division of work 95
placement 149, 159
general swarm principles 95–8
sequential quadratic programming
(SQP) 69 self-organization 89–95
sequential unconstrained minimization types of 97
techniques (SUMT) 11 swarm leaders 103
signal processing 54 synchronization 208–9
simplified swarm optimization (SSO)
algorithm 108 Tabu search 8, 176
simulated annealing (SA) 53, 57 Taylor-based bird swarm method
Sine Cosine Algorithm (SCA) 105 (Taylor-BSA) 186
single-objective optimization problem Taylor series 187
98–9 teaching learning-based optimization
smart grid optimization 224 (TLBO) algorithm 96
smart healthcare 188–9 3-D printing 203
and wearables 189 time-division multiple access (TDMA)
151
smart technologies 201
time efficiency 207–8
social media 54
time series analysis algorithms 258
social network analysis and
optimization 224 total variation (TV) 59
speech and pattern recognition 183 traffic shaping and management 150
speech processing 54 trajectory-based algorithms 6
speeded up robust features (SURF) transportation system optimization 223
245 traveling salesman problem 7
stability principle 95 tunicate swarm algorithm (TSA) 98
278 Nature-inspired optimization algorithms and soft computing

ulcers 179 water cycle algorithm (WCA) 15, 31,

ultrasound 179 131–2, 253
unconstrained optimization on leach routing protocol 36–7
population-based methods 66 wavelet ratio 182
single-point methods 66 whale optimization algorithm (WOA)
underwater friction stir welding 222
(UFSW) process 37–8 Wi-Fi 150
wireless sensor network (WSN) 32
vector-based evolutionary
algorithm 4 X-ray 179
vector-evaluated genetic algorithm
(VEGA) 100 ZigBee 150
vector-evaluated PSO (VEPSO) 100 z-score 181