Combating Cross-Technology Interference For Robust Wireless Sensing With Cots Wifi

Yue Zheng,Zheng Yang,Junjie Yin,Chenshu Wu,Kun Qian,Fu Xiao,Yunhao Liu

DOI: https://doi.org/10.1109/ICCCN.2018.8487431

2018-01-01

Abstract:The past years have witnessed the rapid conceptualization and development of wireless sensing based on Channel State Information (CSI) with commodity WiFi devices. Many research efforts have been devoted to promote WiFi sensing by innovating applications, refining models and optimizing algorithms. A critical issue of Cross-Technology Interference (CTI), however, is surprisingly unnoticed and largely unexplored in the existing literature. In this paper, we demonstrate that CTI poses severe impacts on CSI measurements and further degrades the performance of CSI-based sensing. Based on in-depth understanding of such impacts, we present PERFIC to deal with CTI for CSI on commercial WiFi. We first exploit the inherent cyclostationarity property of different signals to detect CTI and further identify the specific distorted subcarriers on CSI. For each interfered CSI, we then propose to mitigate the impacts of CTI by amending the abnormal subcarriers. We conduct experiments on typical wireless sensing applications, including human detection and activity classification, using off-the-shelf WiFi devices. The results demonstrate that PERFIC yields a remarkable performance gain of >30% with high efficiency and outperforms existing robust classifiers. By providing interference-free CSI that is amendable to existing and emerging CSI-based sensing applications, PERFIC underpins new insights for improving the sensitivity and reliability of wireless sensing.