Target tracking in sensor networks using statistical graphical models

Recent advancement in sensor networks provides a platform for applications that requires in-network data fusion and parallel algorithms. However, processing data in parallel while propagating at low latency is very challenging. Also, implementation of these algorithms is limited by various constraints including energy, computation costs and complex network topology. In this paper, a statistical graphical model based algorithm is developed for in-network processing, which can be applied to tracking problems in the sensor networks. This algorithm represents the complex topology of a sensor network with a simple clique tree. It further utilizes the message passing algorithms to effectively make accurate inferences about the target location. The simulation shows that the algorithm can accurately track the target in a large scale random distributed sensor field with low complexity and low cost. The algorithm is also proved to be robust, as the simulation random disabled some sensors during the tracking phase.