A multipath Gaussian mixture PHD filter algorithm for OTHR multi-target tracking

Considering the fact that target tracking of the sky wave Over-The-Horizon Radar (OTHR) inevitably faces the problems of multipath, low detection probabilities et al., we make use of the advantage of avoiding data association and low computational cost for Gaussian mixture probability hypothesis density (GM-PHD) filter and propose a multipath GM-PHD (MP-GM-PHD) filter framework. Such framework is expected to avoid the target number over-estimation problem by partitioning the radar measurement into sub-measurement set as a PHD updating cell, process the nonlinear problem with utilizing the third-degree spherical-radial cubature rule, and improve the probability of capture for missed targets by supplementing the Gaussian terms from history track information. Simulation results show that under the conditions of different detection probability, the MP-GM-PHD filter can gain a more accurate estimate of the target state and target number, capture the missed targets timely, and alleviate the problems of target number over-estimation and high computational cost by the direct use of GM-PHD filter.