Based on wireless sensor networks, the technique of device-free target localization is a novel, popular and very potential technique in recent years. The basic approach is described as follows. We deploy our wireless sensor network around the area of interest. The wireless nodes communicate with each other and measure the corresponding received signal strength of wireless link. The RSS of links will fade when the target travels through the links in the wireless sensor network. Based on this theory, we can implement target localizing and tracking work with RSS measurement model and appropriate algorithm.This paper presents a novel method of device-free multi-target detection and tracking with wireless sensor networks. The method consists of two major parts:one is background learning algorithm using kernel density estimation for multi-target detection; the other is probability hypothesis density particle filter algorithm for multi-target tracking. For well examining the stability, robustness and accuracy of our method, this paper conducts two different experiments. After analysis of the experiment data, we draw a conclusion that the OSPA accuracy of tracking multiple targets with our method is0.183m. |