一种GPS多径时延估计的改进粒子滤波算法

An Improved Particle Filtering Approach to GPS Multipath Time Delay Estimation

在众多影响GPS卫星导航定位精度的误差源中,多径效应产生的误差成为最重要的误差源。基于贝叶斯估计的粒子滤波对解决非线性非高斯问题提供了一个较灵活的框架,为在多径估计基础上的多径抑制提供了一种重要解决途径。提出将遗传算法引入粒子滤波的多径信号时延估计方法,解决标准粒子滤波中的粒子退化问题。此外,采用卡尔曼滤波估计多径信号幅值。对提出的算法进行了仿真实验,并对标准的粒子滤波(Particle Filter,PF)算法与改进的遗传算法粒子滤波(Genetic Algorithm Particle Filter,GAPF)进行了比较。仿真结果表明,改进算法能够提高对多径信号时延的估计精度。

Among various error sources of GPS,the error caused by the multi-path effect has become the most significant one.The particle filtering based on Bayesian estimation provides a flexible architecture for solving the non-linear and non-Gaussian issues,and an approach to multipath mitigation based on the multipath parameters estimation.A genetic algorithm is employed to combine with the par-ticle filtering for estimating the time delays of the multipath signals. This improved particle filtering technique is capable of combating the particle degeneracy phenomenon for the standard particle filtering,and also maintaining particle diversity.In addition,a Kalman filte-ring algorithm is applied to estimate the magnitudes of the multipath signals.The performance of the improved genetic algorithm particle filtering（GAPF）is compared with that of the standard particle filtering.The simulation results show that the improved particle filtering is superior to the standard particle filtering in estimation accuracy of multipath time delays.