The random set based Gaussian mixture probability hypothesis density filter algorithm is a typical multi-target tracking algorithm, it can track multiple targets when the number of the targets is unknown, but this algorithm requires the knowledge of the targets' starting positions. In many cases, the target's initial location information is not available. In this paper, a modified Gaussian mixture probability hypothesis density filter is developed for multi-target tracking problem because the traditional Gaussian mixture probability hypothesis density filter can not work well when where the targets will appear is unknown. And the proposed algorithm is applied to track the primary users in the cognitive radio systems. A double side prediction algorithm is adopted to solve this primary user tracking problem. First, the forward prediction algorithm is used to estimate the locations of the existed primary users, and then the backward prediction algorithm is used to search the new primary users. The proposed algorithm can be used when we do not know how many'primary users exist, when and where they will appear. The performance of the proposed algorithm is analyzed by simulation. Simulation results show that the proposed algorithm can track the primary users even in a high false detection environment.%基于随机集的高斯混合概率假设密度滤波算法是一种典型的多目标跟踪算法,可以在目标数目未知的情况下进行多目标跟踪,但是该算法要求已知目标的起始位置,在很多情况下,目标的起始位置信息是无法获得的.本文针对这一问题,提出了改进的高斯混合概率假设密度滤波算法,并将本文算法应用于认知无线电系统的主用户跟踪问题.该算法利用双向预测的方式对检测结果进行估计,即使用正向预测算法来估计现存主用户的位置,然后采用后向预测算法来搜索新生的主用户并估计出新生主用户的位置.本文算法的主要优点是在主用户的数目、出现的时间和起始位置均未知的情况下仍可以有效的跟踪目标.最后,通过仿真对本文算法的性能进行了分析.仿真结果表明,本文算法在误检率较高的情况下可以准确地跟踪主用户.
展开▼
