Abstract:Wireless sensing has recently attracted significant interest due to its potential to support a wide range of immersive human-machine interactive applications without requiring any extra devices to be carried out by human users. However, previous studies have shown that wireless sensing performance can be significantly degraded if the relative locations of the transmitters, human users, and receivers are non-ideal and/or the distances between the user and receivers are large. These constraints hinder the wide applications of wireless sensing in many practical scenarios. A promising approach for wireless sensing is through the use of reconfigurable intelligent surfaces (RISs) that can control the propagation environment to create a customizable wireless environment for wireless sensing. To this end, in this paper, the use of RIS-assisted wireless sensing for enhanced human gesture recognition is investigated. A novel RIS-assisted distributed wireless sensing framework that utilizes federated learning (FL) is proposed to enable collaborative model training among decentralized receivers. Then, a novel metric, called human-influencing signal-to-interference ratio (HSIR), is introduced to characterize the quality of locally recorded data as well as its impact on the performance of wireless sensing. To alleviate the model draft problem of FL-assisted wireless sensing, caused by spatial heterogeneity of the quality of wireless sensing data at different receivers, the optimal amplitudes and phases of the RIS are derived so as to improve the HSIR of a set of low-performance receivers located at non-ideal locations. Simulation results show that the proposed RIS-assisted system can significantly improve wireless sensing accuracy by up to 20.1% compared to traditional distributed system.

Learn to Sense: a Meta-learning Based Distributed Sensing Algorithm in Wireless Sensor Networks

Learn to Sense: A Meta-Learning-Based Sensing and Fusion Framework for Wireless Sensor Networks

Networked Sensing with AI-Empowered Interference Management: Exploiting Macro-Diversity and Array Gain in Perceptive Mobile Networks

Learning to Sense: Deep Learning for Wireless Sensing with Less Training Efforts

Meta-Learning for Human-Centered Wireless Sensing: Architecture, Applications, and Challenges

Co-sense: a learning-based collaborative wireless sensing framework

Distributed Least Squares Estimation Algorithms Design for Sensor Networks

A Learning-Based Collaborative Wireless Sensing Framework

Distributed Compressed Sensing in Wireless Local Area Networks.

A Survey Of Machine Learning In Wireless Sensor Netoworks

A Survey of Machine Learning in Wireless Sensor Netoworks from Networking and Application Perspectives

Deep Learning Based Energy Efficiency Optimization for Distributed Cooperative Spectrum Sensing

Sensor Deployment for Wireless Sensor Networks: A Conjugate Learning Automata-Based Energy-Efficient Approach

An Adaptive Distributed Localisation in Wireless Sensor Network

Threshold-Learning In Local Spectrum Sensing Of Cognitive Radio

Adaptive Data Acquisition with Energy Efficiency and Critical-Sensing Guarantee for Wireless Sensor Networks.

Learnable WSN Deployment of Evidential Collaborative Sensing Model

MetaSensing: Intelligent Metasurface Assisted RF 3D Sensing by Deep Reinforcement Learning

E3: Towards energy-efficient distributed least squares estimation in sensor networks

Optimizing Reconfigurable Intelligent Surface-Assisted Distributed Wireless Sensing

Multi-scale Learning Based Data Processing Mechanism for Wireless Body Area Network