二进制无线传感器网络中的分布式自适应粒子滤波目标跟踪算法

Target Tracking Based Distribute Adaptive Particle Filter in Binary Wireless Sensor Networks

为了提高无线传感器网络目标跟踪的实时性,减少通信量,提出了一种二进制无线传感器网络的分布式自适应粒子滤波算法,该算法在簇头更换时,簇头之间只需要传送滤波值和误差方差,而无需传递大量粒子,同时该算法根据滤波方差在线调整粒子数,从而降低了算法的计算量。从算法耗时、均方根误差(跟踪精度)以及通信量等方面进行了仿真研究。仿真结果表明,分布式自适应粒子滤波算法的耗时、通信量要明显少于集中式粒子滤波和分布式粒子滤波;同时其均方根误差的变化幅度受粒子数的影响非常小,具有更好的跟踪性能。

In order to improve real-time and reduce communication costs in wireless sensor network tracking, distributed adaptive particle filtering（DAPF） in binary wireless sensor network was proposed,in which only filtering value and er- ror variance need to be transmitted between cluster heads, meanwhile number of particles is online adjusted according to filtering variance to decrease computation amount. Simulation research was done in aspects of time-consuming, root mean square error（tracking precision） and communication amount. The results indicate that DAPF has obviously less time-consuming and communication amount than centralized particle filter and distributed particle filter,and the rate of change of root mean square error is tinily affected by number of particle.