关键词: 单通道盲分离, 基于M算法的逐幸存路径处理算法（M-PSP）, 多路递归最小二乘（RLS）, 基于RLS算法的降复杂度PSP算法（RLS-RPSP）, 降低复杂度

Abstract: Aiming at the single-channel blind separation of the multi-channel co-frequency modulated linear mixed signal, a new algorithm called RLS-RPSP is proposed which can reduce the computation complexity. The proposed algorithm is according to the idea of adaptive tracking channel, the Per-Survivor Processing（PSP） algorithm, and the framework of M-PSP algorithm based on LMS algorithm. The survivor path can be reserved by piecewise setting the number of survivor path employing the fast convergence of multi-channel RLS algorithm. Therefore, the proposed algorithm not only can synchronously track the multiple channels, but also does not need any estimation of the initial response. The theoretical and simulation results show that, compared with the M-PSP algorithm based on LMS, the proposed algorithm can reduce at least half of the computational complexity when the bit error rates are approximately the same.

Key words: single-channel blind separation, Per-Survivor Processing algorithm based on M algorithm（M-PSP）, multi-channel Recursive Least Square（RLS）, complexity Reduced PSP algorithm based on RLS algorithm（RLS-RPSP）, computation complexity reduction