A New Method of Using Sensor Network for Solving Pursuit-Evasion Problem

Wireless sensor networks offer the potential to significantly improve the performance of pursuers in pursuit-evasion games. In this paper, we study several sensor network systems, their interaction with the pursuers, and the effect on pursuer performance. We propose a general framework to solve the pursuit-evasion problem and present new centralized as well as distributed methods. Specifically, we address three issues in the design of pursuers based on data provided by the sensor network : a) how to identify evader moving patterns, (b) how to predict the evader locations using different evader moving models, and c) how to choose the most efficient pursuit strategies. We propose efficient algorithms to solve these problems and show that they are effective in reducing the capturing time in our simulations. We also compare the distributed and centralized methods. Experimental results show that the distributed method is efficient and produces solutions close to the centralized method.