机载空间光通信平台的交互多模型粒子滤波跟踪算法

Airborne platform′s tracking algorithm for free space optical communication based on IMMPF methods

机载空间激光通信是实现未来超大容量空间通信的主要途径,高精度实时性的机载平台跟踪问题一直是其研究的难点问题之一。因为机载通信平台机动形式的多样性,用单一的模型很难准确描述机载平台的运动状态,因此需要采用多个模型进行描述,即交互多模型方法。粒子滤波能够处理非线性/非高斯问题,其与交互式多模型结合可用来获得更好的跟踪性能。针对粒子滤波算法计算量大、实时性差的问题,给出了粒子滤波算法的并行处理方法。通过计算机仿真,交互多模型粒子滤波算法与其他算法相比,其跟踪性能有较大的提高,忽略通信代价,并行处理的粒子滤波时间开销显著下降。

Free space optical communication based on airborne platform has become the hot research, real time high precision tracking of airborne platform has been already the one of the most difficult problem. Because of diversity of maneuvering forms, it is impossible to accurately describe the movement of airborne platform with the simple model. Thus, it is necessary to adopt interacting multiple model （IMM） methods for tracking system. The particle filter（PF） algorithm can deal with nonlinear/non-Guassian problems, it can be introduced to the algorithm of IMM for nonlinear tracking model of airborne platforms. In order to decrease inherent complex and computational intensity of particle filter, the design of parallel particle filter based on task division was proposed. Through the computer simulation, the result shows that the tracking performance with improved IMMPF methods is better, computational time of parallel particle filter is greatly decreased if communication time is ignored.