基于PDA的非线性分布式跟踪算法

Distributed interacting Multiple Model Tracking Algorithm

研究目标不确定性量测、不确定性机动环境下的分布式多雷达跟踪．针对目标未知机动，把交互式多模型算法中的模型样本空间分成若干子集，分别在多个处理器上进行并行滤波，再在融合中心处理器上将各子处理器结果进行交互式处理，形成了分布交互式多模型算法，然后将它与概率数据关联ＰＤＡ（ＰｒｏｂａｂｉｌｉｓｔｉｃＤａｔａＡｓｓｏｃｉａｔｉｏｎ）相结合，得到了一种新的分布式鲁棒跟踪算法，数字仿真结果证明该算法计算量少，并且具有很强的自适应能力和良好的跟踪性能．

In this paper, we develop Chang's idea [5] further and propose DIMM (Distributed Interacting Multiple Model). Unlike Chang, we deal with the 'distributed' aspect at the model level. We study distributed multi-radar tracking algorithm under the environment of uncertain measurements and uncertain target maneuvers. For unknown target maneuver, we decompose the model space of Blom's IMM (interacting multiple model) [4] into several subsets and we are pleased to find that the mathematical technique of subsets is useful in describing model distribution; then we carry out parallel fibering on multiple local processors. By combining the results of local processors, we bbtain distributed IMM filtering algorithm. Like Chang [5], we make use of PDA; but unlike Chang, we use PDA in preprocessing to get an equivalent measurement for each target. By joining DIMM with PDA, a new distributed robust tracking algorithm is developed. Computer simulations show that the proposed new algorithm requires less computation and has strong adaptive ability and very good performance.