Human Movement Detection and Gait Periodicity Analysis via Channel State Information

Abstract

In recent years, movement detection and gait recognition methods using different techniques emerge in an endless stream. On the one hand, wearable sensors need be worn by the detecting target and the method based on camera requires line of sight. On the other hand, radio frequency signals are easy to be impaired. In this paper, we propose a novel multi-layer filter of channel state information (CSI) to capture moving individuals in dynamic environments and analyze his/her gait periodicity. We design and evaluate an efficient CSI subcarrier feature difference to the multi-layer filtering method leveraging principal component analysis (PCA) and discrete wavelet transform (DWT) to eliminate the noises. Furthermore, we propose the profile matching mechanism for movement detection and the gait periodicity analysis mechanism for human gait. Experimental results in different environments indicate that our approach performs identification with an average accuracy of 94%.