时间异步无线传感器网络的分布式目标跟踪

Distributed target tracking in asynchronous wireless sensor networks

为了降低无线传感器网络在目标跟踪过程中的网络能耗,提出了一种时间异步条件下的分布式目标跟踪方法。首先,依据节点到目标的距离进行动态成簇,以跟踪簇为时间的计算单元,由簇头完成簇内跟踪时间计算及簇间贯序传递,然后引入并行粒子滤波(PPF)算法将粒子集分为多个子集,在子节点处并行采样、计算权重和重采样,最后,簇头节点收集各子节点上传的结果并完成目标的局部状态估计。仿真结果表明,PPF算法具有较好的跟踪精度,且相比于集中式粒子滤波(CPF)算法,可降低约38%的通信量。

To reduce the energy consumption of wireless sensor networks during their target tracking, a distributed target tracking method suited to the condition of time asynchronous was proposed. Firstly, dynamic clusters were established according to the distance between each node and the target. The tracking cluster was used as the calculation unit of time. The tracking time calculation in one cluster and sequential transferring among different clusters were implemented by cluster headers. Then, the particle set was separated into some subsets by the parallel particle filter （PPF） algorithm, which were sampled, weighed and resampled in several nodes. Finally, the estimation of local states was implemented by cluster headers through gathering the results uploaded from nodes. The simulation results show that the PPF algorithm has a good tracking accuracy and can reduce communication traffic about 38% compared with the center particle filter （CPF） algorithm.