无线传感器网络中基于量化观测的粒子滤波状态估计

Particle Filter State Estimation Based on Quantized Observations in Wireless Sensor Networks

无线传感器网络环境下处理分布式状态估计问题,由于网络中的带宽限制,减少通信成本是非常重要的一个环节,需要将观测值量化后再传送。针对非线性系统的状态滤波问题,本文提出了一种基于量化观测的粒子滤波状态估计算法,并阐述了基于量化观测的状态估计过程。文中分别采用基于均匀量化(UQDPF)和非均匀量化(NUQDPF)观测的分布式粒子滤波算法进行状态估计,通过被动跟踪仿真实例,利用均方根误差(RMSE)比较了误差性能,并且比较了在不同量化级数下的非均匀量化算法的跟踪误差,仿真结果表明,基于非均匀量化观测的粒子滤波器具有更高的跟踪精度,是一种有效的非线性滤波算法。

To solve the problem of bandwidth limitation and reduce the cost of communication, measurements from wireless sensor networks are usually quantized before being sent out. Aiming at the nonlinear / filter problem, a distributed particle filter state estimation algorithm based on quantized observation in wireless sensor network is proposed. The progress of particle filter is discussed and illustrated. Using a passive tracking example, the performance and the root-mean-square error （RMSE） are analyzed by comparing the uniform quantization distributed particle filter （UQDPF） with the nonuniform quantization distributed particle filter （NUQDPF）. We also studied the the RMSE of the nonuniform quantization particle filter in different quantization levels. Simulation results show that the NUQDPF is more accurate than the UQDPF in passive aim tracking.