分布式并行粒子滤波算法结构分析与研究

Structure analysis and research on distributed parallel particle filter

粒子滤波器是解决非线性/非高斯系统状态估计的有效技术,广泛应用于目标跟踪、无线通信、自动控制等领域。但因其计算复杂、计算量庞大等缺陷,无法满足实时系统的应用需求。针对粒子滤波器计算量大、实时性差的问题,提出了一种基于MPI的分布式并行粒子滤波算法,给出了Master-Slave并行模式下任务分配、数据划分与负载平衡策略。实验结果表明,若忽略通信代价,加速比基本呈线性增长。

Particle filter is an effective technique for the state estimation in non-linear and/or non-Gaussian dynamic systems. However, its real-time application is limited due to their inherent complex and computational intensity. In order to decrease drawback on on-line filtering, design of distributed parallel particle filter algorithm based on MPI is proposed, a task division, data decomposition and dynamic load balance strategy under Master-Slave model are presented. The result shows that speedups is great improved if the communication time is ignored.