Co-Sense: Exploiting Cooperative Dark Pixels in Radio Sensing for Non-Stationary Target

Jinbo Chen,Dongheng Zhang,Ganlin Zhang,Haoyu Wang,Qibin Sun,Yan Chen

DOI: https://doi.org/10.1109/tmc.2024.3498048

IF: 6.075

2024-01-01

IEEE Transactions on Mobile Computing

Abstract:Radio sensing has emerged as a promising solution for monitoring vital signs in a contactless manner. However, most of the existing designs focus on stationary target and struggle with body motion interference. While some efforts have been made to address this issue, the lack of a physical explanation for the motion elimination principle makes them work as a blind signal separation way and thus leaves the body motion elimination problem still as an open challenge. In this paper, we reveal for the first time the existence of ”dark pixels”-specific points on the same rigid body parts that share the same body movement but exhibit varying physiological motions, with these variations still preserving the physiological rhythm. By exploiting the inherent relationship between the dark pixels, we propose a cooperative sensing framework, Co-Sense, that can achieve robust radio sensing for non-stationary targets in an explainable way. Through extensive experiments, Co-Sense demonstrates its superiority over existing methods, achieving effective motion cancellation and breath sensing with a median absolute respiratory rate (RR) error of 0.36 respiration per minute (RPM) and breath wave correlation of 0.61 under non-stationary scenarios. The results indicate the great potential of Co-Sense in enhancing the accuracy of vital sign sensing with radio signals, especially in real-world environments where targets are rarely stationary.