This book covers cutting edge advancements on self-powered Internet of Things, where sensing devices can be energy-positive while capturing context from the physical world. It provides new mechanisms for activity recognition without the need of conventional inertial sensors, which demand significant energy during their operation and thus quickly deplete the batteries of internet-of-things (IoT) devices. The book offers new solutions by employing energy harvesters as activity sensors as well as power sources to enable the autonomous and self-powered operation of IoT devices without the need of human intervention. It provides useful content for graduate students as well as researchers to understand the nascent technologies of human activity, fitness and health monitoring using autonomous sensors. In particular, this book is very useful for people working on pervasive computing, activity recognition, wearable IoT, fitness/healthcare and autonomous systems.
This book covers a broad range of topics related to self-powered activity recognition. The main topics of this book include wearables, IoT, energy harvesting, energy harvesters as sensors, activity recognition and self-powered operation of IoT devices. This book starts with the introduction of wearable IoT devices and activity recognition and then highlights the conventional activity recognition mechanisms. After that, it describes the use of energy harvesters to power the IoT devices. Later, it explores the use of various energy harvesters as activity sensors. It also proposes the use of energy harvesters as simultaneous source of energy and context information and defines the emerging concept of energy-positive sensing compared to conventional energy-negative sensing. Finally, it explores sensor/signal fusion to enhance the performance using multiple energy harvesters and charts a way forward for future research in this area.
This book covers all important and emerging topics that have significance in the design and implementation of autonomous wearable IoT devices. We believe that this book will lay the foundation for designing self-powered IoT devices which can ultimately replace the conventional wearable IoT devices which need regular recharging and replacement.
Author(s): Muhammad Moid Sandhu, Sara Khalifa, Marius Portmann, Raja Jurdak
Series: Green Energy and Technology
Publisher: Springer
Year: 2023
Language: English
Pages: 174
City: Cham
Foreword by Sajal K. Das
Foreword by Mahbub Hassan
Preface
Contents
About the Authors
Acronyms
Part I Overview of IoT and Activity Recognition
1 Introduction
1.1 Types of IoT Devices for HAR
1.1.1 Implantable
1.1.2 Wearable
1.1.3 Environmental
1.2 Energy Challenges in the use of IoT for HAR
1.3 Motivation
1.4 Book Organisation
References
2 Activity Recognition in IoT
2.1 Activity Recognition Mechanisms
2.2 Wearable Sensors for HAR
2.3 Activity Recognition Using Machine Learning
2.3.1 Data Acquisition and Preprocessing
2.3.2 Segmentation
2.3.3 Feature Extraction
2.3.4 Model Training (Learning)
2.3.5 Model Testing
2.3.6 Evaluation Metrics
2.4 Datasets for Developing and Evaluating HAR Algorithms
2.5 Challenges in Current Activity Recognition Mechanisms
References
Part II Energy Harvesting
3 Using Ambient Energy to Power IoT Sensors
3.1 Energy Harvesting Modes
3.2 Solar
3.3 Kinetic
3.4 Thermal
3.5 RF Waves
3.6 Kinetic Energy Harvesting
3.7 Kinetic Energy Harvesting Circuits
3.8 Operation of keh Transducer at MPP
3.8.1 mpp of the keh Transducer
3.8.2 Harvested Power Stored in the Capacitor
3.8.3 Impact of Threshold Voltage of DC-DC Converterpg on the Harvested Power
3.8.4 Power Consumption of DC-DC Converter
3.9 Solar Energy Harvesting
3.10 Solar Energy Harvesting Circuits
3.11 Operation of seh Transducer at MPP
3.12 Discussion
References
4 Energy Harvester as an Information Source
4.1 KEH as a Sensor
4.1.1 Step Count
4.1.2 Audio Signal Detection
4.1.3 Activity Recognition
4.1.4 Transport Mode Detection
4.1.5 Other Applications
4.2 SEH as a Sensor
4.3 TEH as a Sensor
4.4 RFEH as a Sensor
4.5 Discussion
References
Part III Self-Powered IoT
5 Simultaneous Sensing and Energy Harvesting
5.1 Challenges in Simultaneous Sensing and Energy Harvesting
5.2 System Architecture for Simultaneous Sensingpg and Energy Harvesting
5.2.1 Sensing and Energy Harvesting
5.2.2 Energy-Positive Sensing
5.2.3 Exploring Multiple Sensing Points
5.3 System Design for Simultaneous Sensing and Energy Harvesting
5.3.1 Hardware Designs for keh Sensing and Energy Harvesting
5.3.2 Experimental Setup
5.3.3 The Interference Problem at Different Sensing Points
5.4 Transport Mode Detection: A Case Study
5.4.1 Data Collection
5.4.2 System Model
5.5 Performance Evaluation
5.5.1 Detection Accuracy of keh-Based Sensing Signals
5.5.2 Energy Harvesting
5.5.3 Energy Consumption and System Costs
5.5.4 Energy-Positive Sensing: Discussion and Analysis
5.6 Discussion
References
6 Solar Cell Based Activity Recognition
6.1 Background
6.1.1 Previous har Mechanisms
6.2 Human Activity Recognition Using Solar Cell
6.3 SolAR: System Model and Implementation
6.3.1 Measurement Setup
6.3.2 Solar Cell as a Novel Human Activity Sensor
6.3.3 Implementation of SolAR
6.4 Performance Evaluation
6.4.1 Classification Accuracy
6.4.2 Variability Analysis of Human Activities
6.4.3 Varying Window Sizes
6.4.4 Varying Signal Sampling Frequency
6.4.5 Robustness to User Variance
6.4.6 Environment-Agnostic Analysis
6.5 Energy-Positive HAR
6.5.1 SolAR Harvested Power
6.5.2 SolAR Power Consumption
6.5.3 Energy-Positive har
6.6 Discussion
References
7 Fusion-Based Activity Recognition
7.1 Background
7.1.1 Accelerometer-Based har
7.1.2 keh-Based har
7.1.3 seh-Based har
7.1.4 Limitations and Challenges
7.2 Fusing Solar and Kinetic Energy Signals
7.2.1 Architecture
7.2.2 Measurement Setup
7.2.3 Human Activity Recognition
7.3 Performance Evaluation
7.3.1 Classification Accuracy
7.3.2 Varying Window Sizes
7.3.3 Varying Sampling Frequency of the Signal
7.3.4 Robustness to User Variance
7.3.5 Robustness to Diverse Lighting Conditions
7.3.6 Robustness to Environment-Agnostic and Environment-Preserving Scenarios
7.4 Analysis of Harvested and Consumed Power
7.4.1 Harvested Power
7.4.2 Power Consumption
7.4.3 Energy-Positive har
7.5 Discussion
References
8 Energy-Positive Activity Recognition: Future Directions
8.1 Energy-Efficient Communication Using Energy Harvesters
8.2 Deep Learning
8.3 Federated Learning
8.4 Personalised AI Models
8.5 Real-Time Activity Recognition
8.6 Multi-source Energy Harvesters
8.7 Hardware Implementation on the Edge Device
8.8 Batteryless Operation
8.9 Security and Privacy
8.10 Reducing the System Cost
8.11 Exploring Other Applications
References
Author Index
Appendix Subject Index
Index
