一种按块递归的盲源分离方法

Block recursive blind source separation method

自然梯度算法比随机梯度算法有更好的收敛性能和数值稳定性,块递归算法需要较少的运算时间.结合这两者的优点,提出一种基于块递归的盲源分离算法.首先基于自然梯度和非线性主分量分析,构造出按块递归更新的矩阵方程,然后用QR分解和回代法逐块求解该矩阵方程得到最优分离矩阵.与已有递归型盲源分离算法相比,数值仿真实验表明本方法运行一次所需平均时间减少了65%,所求矩阵的正交性能指标改善了10 dB.

The natural gradient algorithm works more efficiently than the ordinary gradient algorithm, and the block recursive method is feasible to real-time processing. To benefit from the above advantages, a block recursive blind source separation （BSS） approach is presented. Firstly, based on natural gradient and nonlinear principle component analysis, a matrix equation is obtained by block recursive updating, and then the matrix equation is solved by using QR factorization and back substitution to obtain the optimal separating matrix. Compared with other existing recursive-type BSS methods, the proposed algorithm leads to 10 dB improvement in orthogonality performance index, and the average running time reduces by 65 %, which is verified by extensive numerical simulation experiments.