WiHI: Indoor Human Identification with WiFi Signals

The rapid development of human identification technology based on WiFi sensing has demonstrated immense application potential in the fields of security and smart homes. However, existing WiFi sensing methods have certain limitations, including inadequate denoising, susceptibility of selected features to environmental influences, and low recognition accuracy. In this paper, we propose a WiFi-based Human Identification method called WiHI, which utilizes the Channel State Information (CSI) of WiFi signals to extract human gait information. Firstly, antenna and subcarrier selection, along with low-pass filtering, are employed to eliminate noise in the CSI that is unrelated to walking activities. To derive unique human walking patterns from CSI, we propose an effective feature extraction algorithm based on Discrete Wavelet Transform (DWT) and Principal Component Analysis (PCA) techniques. Based on the extracted features, we have designed a two-stage human identification system consisting of stranger detection followed by target human identification. We have carried out extensive experiments on the existing publicly available dataset Widar 3.0 for method evaluation. The results indicate that WiHI can effectively detect unknown users and achieve higher recognition accuracy compared to similar approaches.