A multi-frame track-before-detect algorithm for maneuvering targets in radar system

This paper mainly addresses the maneuvering targets detection and tracking problem in pulse Doppler radar using Dynamic Programming (DP) based Track-Before-Detect (TBD). Generally, the DP-TBD algorithm implicitly embeds the tracking stage into its framework using a predefined target model to determine the admissible searching paths. However, its performance will suffer severe loss due to the mismatching of target model when maneuver occurs. In this paper, to handle the target maneuvers, an improved DP-TBD algorithm is proposed, wherein the hidden state variable of radial acceleration is recursively estimated to help scheme the DP searching paths in a more appropriate way. Meantime, by employing the Current Statistical (CS) model, the DP integration process is modified which can adaptively adjust the searching region to avoid the interference of many noisy states involved by some impractical paths. Simulation results show that the proposed algorithm can effectively improve the detection and tracking performance of DP-TBD.