首页 | 本学科首页   官方微博 | 高级检索  
     

基于滑窗递归广义特征分解的盲源分离方法
引用本文:周祎,贾朝文,杨光. 基于滑窗递归广义特征分解的盲源分离方法[J]. 电子信息对抗技术, 2011, 26(1): 5-8,50
作者姓名:周祎  贾朝文  杨光
作者单位:电子信息控制重点实验室,成都,610036
摘    要:针对混迭矩阵发生突变的盲源分离(BSS)问题,提出一种基于二阶统计量的滑窗递归BSS方法。首先滑窗递归更新两个不同时延的自相关矩阵,然后结合幂迭代法和压缩处理在线计算矩阵束的广义特征向量以求得分离矩阵。与已有递归型盲源分离算法相比,该方法收敛速度快,追踪性能好,仿真结果验证了所提算法的有效性。

关 键 词:盲源分离  递归广义特征分解  滑窗  幂迭代  压缩处理

A Sliding Window Recursive Generalized Eigen-Decomposition Algorithm for Blind Source Separation
ZHOU Yi,JIA Chao-wen,YANG Guang. A Sliding Window Recursive Generalized Eigen-Decomposition Algorithm for Blind Source Separation[J]. , 2011, 26(1): 5-8,50
Authors:ZHOU Yi  JIA Chao-wen  YANG Guang
Affiliation:ZHOU Yi,JIA Chao-wen,YANG Guang(Science and Technology on Electronic Information Control Laboratory,Chengdu 610036,China)
Abstract:A sliding window recursive blind source separation(BSS) algorithm for mixing matrix changed abruptly is presented.Firstly,two auto-correlation matrices with different time delay are estimated recursively.Secondly,the on-line calculation of the generalized eigenvectors,which is the column vectors of the desired separating matrix,is carried out by the power method and deflation procedure.Computer simulation results show that the proposed algorithm has faster convergence and better tracking performance than ex...
Keywords:blind source separation  recursive generalized eigen-decomposition  sliding window  power iterative  deflation procedure  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号