Research On Device-Free Target Tracking Based On Wireless Sensor Networks

In recent years, tracking technology has become more and more important in applications such as security monitoring, fire and rescue, military and anti-terrorism. The rise of wireless sensor networks broadened the application range. Note that existing tracking system based on wireless sensor networks all require the targets to be tracked to carry transceiver devices, therefore their applications are limited. Device-free passive target tracking does not need the targets to carry transceiver device, and therefore has a wide range of applications. However, existing passive target tracking are often limited to a single target or a fixed known number of targets, there hasn’t been effective ways to deal with the scene of variable number targets.In this thesis, we first had a thorough study of existing localization and tracking method based on wireless sensor networks. In the meantime, we studied the measurement model and motion model which are highly related to the tracking problem, then we proposed a novel multi-target localization method. The method can estimate the total number of the targets and their respective coordinates. Compared with existing localization method, our method solved the problem of determining the number of the targets and accuracy has also to some extent been improved. Experiments showed that in both indoor and outdoor environment, the localization accuracy was improved to about0.07m.Then, we apply the probability hypothesis density particle filter to solve multi-target tracking. In the meantime, we proposed the improved Quasi-Monte Carlo to further improve the tracking accuracy. The method can successfully handle the entry and exit of the real tracking scene. It is a good solution to the multi-target tracking in conjunction with the multi-target localization method mentioned above. Experiments showed that in indoor environment, the tracking accuracy was about0.167m.