FreeSense: human-behavior understanding using Wi-Fi signals

Abstract

Device-free passive human behavior understanding plays an important role in human–computer interaction and public safety management. Especially, human detection and human identification are two key enablers for a wide range of indoor location-based services such as asset security, emergency response and personalized service. In this paper, we proposed a new method for human detection with high robustness and a novel approach for indoor human identification based on Wi-Fi Channel State Information (CSI) signals. The former utilizes the phenomenon that when a person moves, phase differences will appear between the waveforms of different receiving antennas. It can be used to deal with the effect of multipath and noises. The latter is based on the observation that each person has specific influence patterns to the surrounding Wi-Fi signals while moving, regarding their body shape characteristics and motion patterns. We use a combination of Principal Component Analysis (PCA), Discrete Wavelet Transform (DWT) and Dynamic Time Warping (DTW) techniques to capture this specific influence patterns. We implemented our human detection method in two typical indoor environments (i.e., a meeting room and a bedroom) and the results demonstrate an average false positive (FP) of 0.58% and an average false negative (FN) of 1.20%. We also implemented our human identification system in a home environment and recruited 9 users for data collection and evaluation. Experimental results indicate that the identification accuracy is about 88.9–94.5% as the size of the candidate user set changes from 6 to 2, showing that the proposed method is effective in domestic environments.