用于宽带非平稳信源分离的一种盲波束算法  

Deterministic Blind Beamforming for Broadband Nonstationary Source Separation

作　　者：林静然[1] 彭启琮[1] 邵怀宗[1] 居太亮[1] 

机构地区：[1]电子科技大学通信与信息工程学院,成都610054

出　　处：《电子科技大学学报》2008年第3期343-346,共4页Journal of University of Electronic Science and Technology of China

基　　金：四川省科技基金(04GG21-020-02)

摘　　要：提出了一种用于分离宽带非平稳信源的盲波束算法。该算法以阵列旋转不变性为基础,利用宽带信源的非平稳特性,将传统的窄带盲波束算法扩展到频域执行;针对在扩展过程中可能出现的通道互换,利用相邻频点序列的互相关系数,对各频点的分离序列进行通道重排,以确保分离出的宽带信源在频域的一致性。该算法还消除了传统盲分离算法中普遍存在的幅度模糊问题,使得分离出的信号没有幅度失真。计算机仿真验证了算法的正确性和有效性。A novel approach of deterministic blind beamforming （DBBF） for broadband nonstationary source separation is presented in this paper. Utilizing the nonstationarity of broadband sources, the conventional narrowband DBBF, based on rotational invariance techniques, is extended to the broadband case, and implemented in frequency domain. A channel rearranging operation, exploring the correlation coefficients between the separated sequences of adjacent bins, is performed to cope with the channel swap. Then the frequency consistency of the recovered sources can be ensured. In addition, the problem of scale ambiguity is eliminated and the original sources can be recovered without scale distortion. Numerical examples confirm the excellent performance of the proposed approach.

关 键 词：宽带非平稳信源 相关系数 盲波束算法 旋转不变性 幅度模糊 

分 类 号：TN918[电子电信—通信与信息系统]