Multi-target Tracking Method of Non-cooperative Bistatic Radar System Based on Improved PHD Filter

The Non-cooperative Bistatic Radar (NBR) system, recognized for its low cost, robust battlefield survivability, and ability to detect stealth targets, has become a research hotspot in the radar field in recent years, playing a significant role in addressing the limitations of existing monostatic active radars. However, due to this "noncooperative" detection mechanism, from the perspective of multi-target tracking (MTT), NBR systems using radar as the external illuminator also face problems such as complex background, poor continuity and random intervals of plot data. In this context, traditional multi-target tracking methods based on data association often encounter the issue of combinatorial explosion in complex scenes. This paper introduces an improved Probability Hypothesis Density (PHD) filtering method based on the theory of Random Finite Sets (RFS), which can effectively handle non-linear/non-Gaussian models and achieve effective filtering estimation of real points in MTT under NBR scenarios. Simulation results demonstrate that the improved PHD filtering method outperforms its predecessor in tracking performance and real-time processing performance.