多核并行粒子滤波算法设计与实现

Design and Implementation of Multi-core Parallel Particle Filter Algorithm

粒子滤波算法由于需要采样大量粒子才能较好地逼近后验概率,故使得系统实时性较差。以一个简单的离散系统为例实现多种粒子滤波并行算法。实验结果表明,并行算法可以有效提高计算效率,基于OpenMP和MPI的并行算法加速效果较好,在一定情况下可以达到超线性加速,此外,当粒子数达到一定数量时,MPI结合OpenMP的并行方法加速效果更佳。

The particle filter algorithm requires sampling a large number of particles to better approximate the posterior probability,so the real-time performance of the system is poor.This paper uses a simple discrete system as an example to implement a variety of parallel particle filter algorithms.Experimental results show that the parallel algorithm can effectively improve the computational efficiency,the parallel algorithm based on OpenMP and MPI has a good acceleration effect,and can achieve superlinear acceleration under certain circumstances,in addition,when the number of particles reaches a certain number,the parallel method of MPI combined with OpenMP accelerates better.