一种快速实值分解Root-MUSIC方位估计算法  

A Fast Root-MUSIC Algorithm for Real-valued Egendecomposition

作　　者：王晓东[1] 贾继鹏[1] 于清津[1] 

机构地区：[1]中国人民解放军91245部队

出　　处：《电声技术》2009年第S1期38-40,45,共4页Audio Engineering

摘　　要：针对基阵输出信号的协方差矩阵具有中心-厄尔米特特性,提出了一种基于实值分解技术的Root-MUSIC算法,并给出了一种降阶快速求根方法,这种降阶算法可以降低求解多项式的次数。实值分解技术极大降低了特征分解时的计算量,可分辨多个高度相关或相干信号源。仿真实验结果证明,与传统算法相比,快速实值分解Root-MUSIC算法具有可以分辨相干信号、计算量小和性能好等优点。A fast algorithm for the Root-MUSIC with real-valued eigendecomposition is proposed for the covariance matrix of array output is centro-Hermitian.There is no spectral search in Root-MUSIC algorithm.The computational complexity is reduced by the real-valued eigendecompostion technique.Using the Root-MUSIC algorithm,DOAs(Direction-Of-Arrivals) of uncorrelated sources can be accurately estimated,and correlated or coherent signals are allowed to separate highly.A new fast way for the polynomial finding roots is proposed.Finally,the simulating results,compared with convention Root-MUSIC algorithm,have features of better performance and low computational complexity.

关 键 词：波达方向 实值特征分解 空间平滑 

分 类 号：TN911.7[电子电信—通信与信息系统]