Person identification using Wi-Fi CSI focusing on door opening and closing motions based on per-channel similarity estimation

Abstract

Identification of persons entering a room or building will be beneficial to provide personalized smart home services. Conventional identification scheme based on image analysis involve additional actions to persons for identification and also there could be privacy concerns with unintended persons appearing in images. In this study, we propose a scheme to identify an individual from door opening and closing motions by analyzing Wi-Fi Channel State Information (CSI) time series. The motions to be analyzed appear in our daily lives and the scheme does not pose additional actions to persons. Also, privacy concerns will be negligible when analyzing WiFi signal. The proposed scheme estimates similarity between obtained CSI data of the target person and that in the reference data obtained in advance based on cross correlation and dynamic time warping and adopts voting principle per channel basis. Through evaluation experiments with 10 subjects, the proposed scheme achieves average F value of 0.94.