基于贪婪量测划分的隐身目标跟踪MS-LMB 滤波器

For stealthy target tracking, we apply the greedy measurement partitioning mechanism to the multisensor labeled multi-Bernoulli (MS-LMB) filter to solve the problem of multi-radar tracking under low detection probability. Generally, Gibbs sampling is applied to solve the measurement partitioning problem in the traditional MS-LMB filter. However, when most radars in the radar network are in the state of missing detection due to low detection probability, the likelihood weight of the stealthy target will be too small to be easily obtained by Gibbs sampling. Thus, it’s difficult to accurately estimate the state of stealthy target. The greedy measurement mechanism can solve this problem since it separately considers the measurement set containing missing detection items. The simulation results show that the filtering performance of the MS-LMB filter with greedy measurement partitioning mechanism is obviously better than that of the MS-LMB filter with Gibbs sampling when tracking stealthy targets.