基于阵列接收模型的宽带盲源分离算法

针对传统盲分离算法对宽带信号不适用的问题，提出了一种基于阵列接收模型的宽带盲源分离算法。该算法以子带分解的方法实现了瞬时复值盲分离方法在宽带情形下的扩展。针对扩展过程中可能出现的子带间次序模糊及子带内幅度模糊的问题，利用阵列接收情况下分离矩阵与混合矩阵的特点，提出了一种基于波达方向(direction of arrival, DOA)估计的次序调整及幅度去模糊方法。仿真结果表明，该算法不仅能有效地分离宽带信号，而且可准确地恢复出信号幅度。

Broadband blind source separation algorithm based on array model

Aiming at the problem that conventional blind source separation algorithms are not suitable for the broadband mixed signals, a new blind source separation algorithm based on the array model is proposed. The method of sub band decomposition is adopted to realize the expansion of instantaneous complex valued blind source separation methods under the broadband condition. Using the characters of the separating matrix and mixing matrix in array model, a new method based on direction of arrival (DOA) estimation for solving the sca ling and permutation ambiguities is proposed. The simulation results show that the algorithm can not only separate broadband mixed signals effectively, but also recover the signals’ amplitude correctly.