Doppler Based Detection of Multiple Targets in Passive WiFi Radar Using Underdetermined Blind Source Separation

Device-free approaches for detecting and localizing people in wireless environments have attracted significant attention in recent years. However, within indoor environments the large numbers of people moving in close proximity to each other often impedes the utility of such approaches. In this paper we present a new method for identifying multiple targets in Wi-Fi passive radar systems using only a single receive channel to detect Doppler returns. The technique is based on tree-structure sparse underdetermined blind source separation and utilizes proximal alternating methods in a convex optimization field. Firstly, we show proof-of-principle simulation results for two targets moving within a typical indoor scenario and compare the results with those from the well-known independent component analysis (ICA). Secondly, we validate the simulation outputs using real-world experimental data. The results demonstrate the effectiveness of the technique for device-free detection of multiple targets in the urban wireless landscape.