基于概率假设密度的无线传感器网络多目标跟踪算法

A PHD Filter for Multi-target Tracking in Wireless Sensor Networks

摘要:

针对无线传感器网络多目标跟踪过程中杂波难以去除以及由数据关联复杂带来的计算复杂度高的问题,将概率假设密度滤波器应用于无线传感器网络,以更好地对多目标状态信息进行融合估计;首先,建立簇-树型无线传感器网络模型,并运用随机有限集理论对目标状态模型和传感器观测模型进行描述；然后,根据目标与节点之间的距离设置观测阈值,当传感器节点测量值小于观测阈值时,概率假设密度滤波器将实时对该组测量数据进行处理,从而实现传感器网络对目标状态的联合检测与跟踪;仿真结果表明,在无线传感器网络的多目标跟踪应用中,该算法比粒子滤波算法具有更高的跟踪效率和精度。

Abstract:

For the multi-target tracking in clutter is difficult to remove and the problem of high complexity by the data association, the probability hypothesis density (PHD) filtering algorithm is used in a wireless sensor networks (WSN) in order to get the appropriate target state information fusion estimation. First of all, the target state model and sensor observation model are described by random finite sets theory in a wireless sensor networks (WSN) with cluster-tree topology; then, the PHD filter processes the set of real-time data, when the measured value is less than the threshold, which is set up according to the distance between target and node, so as to realize the detection and tracking of targets. The simulation results show that PHD filtering algorithm used in the wireless sensor networks has higher tracking accuracy and efficiency than the particle filtering algorithm.