Smart Wireless Sensing : From IoT to AIoT

Perception of human beings has evolved from natural biosensor to powerful sensors and sensor networks. In sensor networks, trillions of devices are interconnected and sense a broad spectrum of contexts for human beings, laying the foundation of Internet of Things (IoT). However, sensor technologies have several limitations relating to deployment cost and usability, which render them unacceptable for practical use. Consequently, the pursuit of convenience in human perception necessitates a wireless, sensorless and contactless sensing paradigm. Recent decades have witnessed rapid developments in wireless sensing technologies, in which sensors detect wireless signals (such as acoustic, light, and radio frequency) originally designed for data transmission or lighting. By analyzing the signal measurements on the receiver end, channel characteristics can be obtained to convey the sensing results. Currently, significant effort is being devoted to employing the ambient Wi-Fi, RFID, Bluetooth, ZigBee, and television signals for smart wireless sensing, eliminating the need for dedicated sensors and promoting the prospect of the Artificial Intelligence of Things (AIoT). This book provides a comprehensive and in-depth discussion of wireless sensing technologies. Specifically, with a particular focus on Wi-Fi-based sensing for understanding human behavior, it adopts a top-down approach to introduce three key topics: human detection, localization, and activity recognition. Presenting the latest advances in smart wireless sensing based on an extensive review of state-of-the-art research, it promotes the further development of this area and also contributes to interdisciplinary research. Preface 6 Organization of the Book 7 Anticipated Audience 8 Acknowledgments 9 Contents 10 About the Authors 11 Acronyms 13 Part I The Background 15 1 Wireless Sensing Overview 16 1.1 What Is Wireless Sensing? 16 1.2 Applications of Wireless Sensing 17 1.3 Challenges of Wireless Sensing 18 1.3.1 Representative Signal Feature 18 1.3.2 Effective Recognition Model 20 1.3.3 Informative Dataset 21 References 21 2 Understanding of Channel State Information 23 2.1 Definition of CSI 23 2.2 Noise in CSI 26 2.3 Signal Features in CSI 27 2.3.1 Time of Flight 27 2.3.2 Angle of Arrival and Angle of Departure 28 2.3.3 Doppler Frequency Shift 30 2.4 CSI Data Collection 31 2.4.1 Intel 5300 NIC CSI Tool 31 2.4.2 Atheros-CSI-Tool 32 2.4.3 PicoScenes Platform 32 References 32 Part II Detection: Passive Human Detection with Wireless Signals 34 3 Passive Human Detection with Wi-Fi 35 3.1 Introduction 35 3.2 Preliminaries 37 3.2.1 The Omnidirectional Passive Human Detection Problem 38 3.2.2 Signal Power Features 39 3.3 Feature Extraction and Classification 40 3.3.1 Sensitivity to Human Presence 40 3.3.2 Resistance to Environmental Dynamics 42 3.3.3 Modeling CFR Amplitude Features 43 3.3.4 Signature Classification 44 3.4 Human Detection 45 3.5 Performance 46 3.5.1 Experiment Methodology 46 3.5.2 Static Detection Performance 48 3.5.3 The Impact of Window Size 50 3.5.4 Mobile Detection Performance 52 3.6 Related Work 53 3.7 Conclusion 54 References 54 4 Passive Detection of Moving Targets with Dynamic Speed Using Wi-Fi 56 4.1 Introduction 56 4.2 Related Works 58 4.3 Overview 59 4.4 Methodology 60 4.4.1 Data Preprocessing 61 4.4.1.1 Phase Sanitization 61 4.4.1.2 Outlier Filter 63 4.4.2 Feature Extraction 64 4.4.3 Motion Detection 66 4.4.4 Enhancement Via Multiple Antennas 67 4.5 Experiments and Evaluation 68 4.5.1 Experimental Setup 68 4.5.2 Performance Evaluation 70 4.5.2.1 Evaluation Metric 70 4.5.2.2 Overall Performance 70 4.5.2.3 Impacts of Sliding Window Size 72 4.5.2.4 Impacts of Number of Features 72 4.5.2.5 Impacts of Number of Antennas 73 4.5.2.6 Performance Against Dynamic Speed 74 4.6 Conclusion 74 References 75 5 Passive Detection of Moving and Stationary Human with Wi-Fi 76 5.1 Introduction 77 5.2 Related Works 79 5.3 Preliminary 80 5.4 System Design 81 5.4.1 Overview 81 5.4.2 Motion Interference Indicator 82 5.4.3 Moving Target Detection 85 5.5 Stationary Target Detection 87 5.5.1 Periodic Alterations from Breathing 87 5.5.2 Breathing Detection 89 5.5.2.1 Signal Filter 90 5.5.2.2 Sinusoidal Model 90 5.5.2.3 Parameter Estimation 91 5.5.3 Embracing Frequency Diversity 92 5.6 Experiments and Evaluation 95 5.6.1 Implementation 95 5.6.1.1 Experimental Environments 95 5.6.1.2 Experimental Methodology 96 5.6.1.3 Evaluation Metrics 97 5.6.2 Performance 97 5.6.2.1 Performance of Detecting Moving Human 97 5.6.2.2 Performance of Detecting Static Human 99 5.6.2.3 Putting It All Together 101 5.7 Discussions and Future Works 102 5.7.1 Monitoring Breathing Rate 102 5.7.2 Expanding Detection Coverage Via Space Diversity 103 5.7.3 Multiple Target Detection 104 5.7.4 Extending to Through-Wall Detection 104 5.8 Conclusions 104 References 105 Part III Localization: Passive Human Localization with Wireless Signals 107 6 Decimeter-Level Passive Human Tracking with Multiple Wi-Fi Links 108 6.1 Introduction 108 6.2 Preliminary 111 6.2.1 Channel State Information 111 6.2.2 From CSI to PLCR 111 6.2.3 Challenges for Tracking 113 6.3 Modeling of CSI-Mobility 113 6.3.1 The Ideal Model 114 6.3.2 The Real Model 115 6.4 PLCR Extraction 117 6.4.1 CSI Preprocessing 117 6.4.2 PLCR Extraction Algorithm 119 6.4.3 PLCR Sign Identification 119 6.5 Tracking Velocity and Location 120 6.5.1 Movement Detection 120 6.5.2 Initial Location Estimation 121 6.5.3 Successive Tracking 121 6.5.4 Trace Refinement 121 6.6 Evaluation 123 6.6.1 Experiment Methodology 123 6.6.2 Overall Performance 124 6.6.3 Parameter Study 126 6.7 Related Work 128 6.8 Conclusion 129 References 129 7 Decimeter-Level Passive Human Tracking with a Single Wi-Fi Link 131 7.1 Introduction 131 7.2 Overview 135 7.3 Motion in CSI 136 7.3.1 CSI-Motion Model 136 7.3.2 Joint Multiple Parameter Estimation 137 7.3.3 CSI Cleaning 139 7.4 Localization 142 7.4.1 Path Matching 142 7.4.2 Range Refinement 146 7.4.3 Localization Model 147 7.5 Evaluation 148 7.5.1 Experiment Methodology 148 7.5.2 System Performance 149 7.5.3 Parameter Study 152 7.6 Discussion 156 7.7 Related Work 157 7.8 Conclusion 158 References 158 Part IV Recognition: Passive Human Activity Recognition with Wireless Signals 161 8 Inferring Motion Direction with Wi-Fi 162 8.1 Introduction 162 8.2 System Overview 165 8.3 Doppler Effect in Wi-Fi 166 8.3.1 Doppler Effect 166 8.3.2 Doppler Effect in CSI 167 8.3.3 Doppler Effect by Multiple Antennas 168 8.3.4 Extraction of Doppler Effect 170 8.4 Motion Recognition 172 8.4.1 Player Reaction in Doppler Effect 172 8.4.2 Motion Recognition Workflow 173 8.5 Evaluation 175 8.5.1 Experiment Methodology 175 8.5.2 Performance 177 8.5.3 Parameter Study 180 8.6 Limitations and Discussion 184 8.7 Related Work 184 8.8 Conclusion 185 References 186 9 Human Gesture Recognition with Wi-Fi 188 9.1 Introduction 188 9.2 Motivation 191 9.3 System Overview 193 9.4 Body Coordinate Velocity Profile 195 9.4.1 Doppler Representation of CSI 195 9.4.2 From DFS to BVP 196 9.4.3 BVP Estimation 198 9.4.4 Location and Orientation Prerequisites 199 9.5 Recognition Mechanism 200 9.5.1 BVP Normalization 201 9.5.2 Spatial Feature Extraction 202 9.5.3 Temporal Modeling 202 9.6 Evaluation 203 9.6.1 Experiment Methodology 203 9.6.2 Overall Accuracy 205 9.6.3 Cross-Domain Evaluation 208 9.6.4 Method Comparison 210 9.6.5 Parameter Study 212 9.7 Discussions 215 9.8 Related Work 216 9.9 Conclusion 217 References 217 10 Human Gait Recognition with Wi-Fi 220 10.1 Introduction 221 10.2 Related Work and Motivation 223 10.3 System Design 224 10.3.1 GBVP Extraction 224 10.3.2 Recognition Mechanism 227 10.4 Evaluation 229 10.4.1 Experimental Methodology 229 10.4.2 Overall Performance 230 10.4.3 Generalizability Evaluation 231 10.4.4 Parameter Study 232 10.5 Conclusion 234 References 234 Part V Conclusions 236 11 Research Summary and Future Directions 237 11.1 Summary 237 11.2 Open Issues 238