摘要
针对无线传感器网络环境下的机动目标跟踪问题,提出了一种描述目标机动加速度的目标状态空间模型,以此模型为基础开发出基于粒子滤波的单目标和多目标跟踪算法.基本思想是:在状态空间中通过寻找一组传播的随机样本来获得近似后验概率分布,并以样本均值代替积分运算,从而求得最小状态方差估计.仿真结果表明,所提算法可以较好地解决无线传感器网络环境下的机动目标跟踪问题,速度跟踪精度、机动加速度跟踪精度均较经典分布式粒子滤波算法分别提高20%、27%.
Focusing on the maneuvering target tracking problem in wireless sensor networks, a state space model for describing the maneuvering target acceleration is proposed. On the basis of the model, single and multiple target tracking algorithms based on particle filtering are developed, in which the approximate posterior probability distribution is acquired through searching a set of transmitted random samples in the state space, and the integral operation is replaced by sample＇s average value so as to obtain the minimum variance estimation. Simulation results show that in wireless sensor network environment, the maneuvering target tracking problem can be solved better by the proposed algorithm. The precision of velocity tracking and maneuvering tracking acceleration is increased by about 27 % and 20% respectively compared to the traditional particle filtering algorithms.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2007年第8期912-916,共5页
Journal of Xi'an Jiaotong University
基金
国家高技术研究发展计划资助项目(2005AA121130)
关键词
传感器网络
目标跟踪
粒子滤波算法
sensor network
target tracing
particle filter algorithm
