Device-free human activity recognition using CSI - Robotics Institute Carnegie Mellon University

Device-free human activity recognition using CSI

Wei Xi, Dong Huang, Kun Zhao, Yubo Yan, Yuanhang Cai, Rong Ma, and Deng Chen
Workshop Paper, 1st Workshop on Context Sensing and Activity Recognition (CSAR '15), pp. 31 - 36, November, 2015

Abstract

Activity recognition is an important component of pervasive computing applications. Device-free activity recognition has the advantage that it does not have the privacy concern of using cameras and the subjects do not have to carry a device on them. Recently, it has been shown that channel state information (CSI) can be used for device-free activity recognition. Their key limitation lies in the lack of universality. In this paper, we propose ARM, a wireless human Activity Recognition and Monitoring system. ARM investigate the correlation between CSI phase variation and human activity. We present an efficient carrier frequency offset (CFO) estimation algorithm for Wi-Fi devices and introduce Haar wavelet function to eliminate the noises. After these preprocessing, ARM uses the correlation as the profiling mechanism and recognizes a given activity by profile matching. We implemented ARM using both commercial Wi-Fi devices and USRP to evaluate it in different environments. Our result- s show that ARM achieves an average accuracy of greater than 75%.

BibTeX

@workshop{Xi-2015-122494,
author = {Wei Xi and Dong Huang and Kun Zhao and Yubo Yan and Yuanhang Cai and Rong Ma and Deng Chen},
title = {Device-free human activity recognition using CSI},
booktitle = {Proceedings of 1st Workshop on Context Sensing and Activity Recognition (CSAR '15)},
year = {2015},
month = {November},
pages = {31 - 36},
}