Abstract:Existing human intrusion detection methods based on Wi-Fi signals lack of utilization on phase information, which limit both the scope of application and effective range of intrusion detection. This paper discusses the availability of phase difference with CSI and adopts a light-weight indicator to sense the changes of states in wireless environments. In order to detect stationary intruders, the recurrence quality analysis is adopted to select the most periodical subcarrier and discrete wavelet transformation is utilized to extract breathe signals in certain frequency band. Experimental results provide a true detection rate of around 90% for both moving and stationary intruders in two scenarios.