Device-free Localization and Tracking of Multiple Person in UWB Sensor Networks

In recent years, radio tomography image (RTI) becomes a hot topic in device-free localization, which needs target taking no electronic device in indoor environment. However, the state-of-the-art does not efficiently solve the problems of the localization accuracy introduced by the multipath interference, and cannot work well to locate multi-person. Due to ultra wide band (UWB) owns good multipath resolution, high robustness, and high stability. In this paper, we introduce UWB sensor networks into state-of-the-art RTI system to locate and track moving multi-person. In order to locate and track multi-person in real-time, we propose the constant false alarm rate detector (CFAR), the density clustering (DBSCAN) and the dynamic threshold to the peculiarities of RTI. In addition, considering multipath crossing when people are walking, we proposed to utilize joint probabilistic data association (JPDA) to solve the problem of multi-person tracking. The field experiment results show that our system is capable of accurately tracking multi-person in indoor environment, even when their trajectories intersect multiple times.