传感器网络中的分布式粒子滤波被动跟踪算法比较研究

Comparison of Decentralized Particle Filter Algorithms in Sensor Networks for Passive Tracking

为提高无线传感器网络(WSN)中的被动跟踪性能,并减少通信量,提出了两种分布式粒子滤波方法.在使用动态分簇结构的基础上,采用信息粒子滤波器(IPF)技术,以簇头作为簇的处理中心,接收来自子节点的观测量,形成本地估计,再将并行粒子滤波器(PPF)将粒子集被分成多个小的子集,分配到簇中的各子节点,完成并行进行粒子滤波过程.在通过计算机仿真的基础上,进行了跟踪和能耗的对比分析研究,结果表明IPF和PPF不仅提高了跟踪精度,而且减少了WSN中的通信能量开销.

Two decentralized particle filtering methods for improving the passive tracking performance and reducing communication amount in wireless sensor networks （WSN） （are） proposed. Based on dynamic clustering, the information particle filter receives the observations from children nodes and formulates the local estimate with the cluster head as the processing center. The parallel particle filter divides the particle set into several subsets, which are distributed to children nodes, and the particle filtering processing runs parallel. Finally, computer simulation is conducted to compare tracking performance and to analyze communication amount. Simulation results show not only the tracking performance is improved, hut also the communication costs are reduced in WSN.