Resampling from the Niching Genetic Algorithm Applicated in Extended Kalman Particle Filter

Two serious problems existing in Particle filter (PF) are the degeneracy phenomenon and the sample impoverishment caused by simple random resampling. In this paper, based on the Extended Kalman particle filter (EKPF) which selects the importance distribution of PF by the Extended Kalman filter (EKF), we propose a new resampling method from niching genetic algorithm to inhibit the degeneracy phenomenon and avoid the sample impoverishment problem, and name the improved particle filtering algorithm as Niching genetic algorithm Extended Kalman particle filter (NGA-EKPF). According to the theoretical analysis and computer simulation of three algorithms in the Global positioning system (GPS), i.e. EKF, EKPF with simple random resampling and NGA-EKPF, the performance of the proposed algorithm has improvement compared with other algorithms not only in positioning accuracy, but also by Cramér-Rao low bound (CRLB) which provides a theoretical bound on the filtering performance.