Abstract:

As adopting the central particle filter to solve the target tracking problem in binary sensor networks cannot do well in timeliness and energy conserving, a distributed particle filter based on dynamic clustering is proposed. The dynamic cluster is an information processing structure changing with the target’s movement, in which the cluster head collects the measurements from the cluster members through one hop communication and estimates the target states in real time; the other sensor nodes out of the cluster are turned into the sleep mode to save energy. Simulations indicate that when the cluster switching distance and the activation radius are properly set, the algorithm can have as a good tracking precision as the central particle filter. Besides, reducing the cluster switching distance and the activation radius will benefit the energy conserving for the network while does not affect the tracking effectiveness.