ADPF Algorithm for Target Tracking in WSN

Abstract

Particle filtering (PF) has been widely used in solving nonlinear/nonGaussian filtering problems. Inferring to the target tracking ina wireless sensor network (WSN), distributed PF (DPF) was used dueto the limitation of nodes�?? computing capacity. In this paper,a novel filtering method�??asynchronous DPF (ADPF) for target trackingin WSN is proposed. There are two keys in the proposed algorithm.Firstly, instead of transferring value and weight of particles, Gaussianmixture model (GMM) is used to approximate the poste- riori distribution,and only GMM parameters need to be transferred which can reduce thebandwidth and power consumption. Secondly, in order to use samplinginformation effectively, when target moving to the next cluster headregion, the GMM parameters are transfer to the next cluster head,and combine with the new local GMM parameters to compose the newGMM parameters incrementally. The ADPF can also deal with the situationof different number of nodes in different cluster when using thedynamic cluster structure. The proposed ADPF is compared to someother DPF for WSN target tracking, and the experimental results showthat not only the precision is improved, but also the bandwidth andpower is reduced.