基于多步分解算法的解卷积混合盲源分离新方法

徐先峰; 冯大政

doi:10.3724/SP.J.1146.2008.01427

基于多步分解算法的解卷积混合盲源分离新方法

doi: 10.3724/SP.J.1146.2008.01427

基金项目:

国家自然科学基金(60672128)资助课题

计量
- 文章访问数: 3107
- HTML全文浏览量: 98
- PDF下载量: 788
- 被引次数: 0
出版历程
- 收稿日期: 2008-11-03
- 修回日期: 2009-03-23
- 刊出日期: 2009-10-19

A New Method Based on a Multi-Stage Algorithm for Blind Source Separation of Convolutive Mixtures

摘要

摘要: 该文提出一种基于二阶统计量的时域多步分解算法求解卷积混合盲源分离问题。引入白化处理，将混迭矩阵转变成酉矩阵，同时，根据源信号不同延时下相关矩阵所具有的块状对角结构，将酉矩阵分为不同的列块。针对各列块之间相互正交的特性，提出一种关于某一特定列块的最小二乘三二次代价函数。利用一种常规的基于梯度下降法的三迭代算法，交替估计代价函数中的3组待定参数，搜索其最小点，得到酉矩阵一个列块的估计。利用系统化的多步分解算法(MSA)，依次估计酉矩阵的每个列块，最终得到整个酉矩阵的估计，进而恢复出源信号。仿真结果表明，新方法性能优于经典的SUB方法及新近提出的JBD-NonU方法，可有效地解决卷积混合盲源分离问题。
- 盲源分离; 卷积混合; 多步分解算法; 三迭代算法
Abstract: A time-domain Multi-Stage Algorithm (MSA) based on the second order statistics for blind source separation of convolutive mixtures is proposed. Whitening procedure is adopted to transform the mixing matrix into a unitary matrix. The unitary matrix is expressed as a column-block matrix according to the block-diagonalization structure in autocorrelation matrices of source signals at different time delays. A novel least square tri-quadratic cost function with respect to a certain column block of the unitary matrix is proposed utilizing the orthogonality between each two different column blocks. Furthermore, a regular Triply Iterative Algorithm (TIA) following the gradient descent idea is used to seek the minimum point of the tri-quadratic cost function by alternately estimating one of the three independent variables parameter subsets, obtaining a column block of the unitary matrix. With each column block being got by using the systemic multi-stage algorithm, the unitary matrix can be estimated and then the source signals can be retrieved. Simulations results illustrate that, the new method outperforms the classic SUB method and the recently proposed JBD-NonU method, and can be efficiently applied to the blind source separation of convolutive mixtures.

HTML全文

参考文献(1)

Belouchrani A, Abed-Meraim K, and Cardoso J F, et al.. Ablind source separation technique using second-orderstatistics [J].IEEE Transactions on Signal Processing.1997,45(2):434-444[2]Feng D Z, Zhang X D, and Bao Z. An efficient multistagedecomposition approach for independent components [J].Signal Processing.2003, 83(1):181-197[3]Laar J, Moonen M, and Sommen P C W. MIMOinstantaneous blind identification based on second-ordertemporal structure [J].IEEE Transactions on SignalProcessing.2008, 56(9):4354-4364[4]Feng D Z, Zheng W X, and Cichocki A. Matrix-groupalgorithm via improved whitening process for extractingstatistically independent sources from array signals [J].IEEETransactions on Signal Processing.2007, 55(3):962-977[5]Buchner H, Aichner R, and Kellermann W. A generalizationof blind source separation algorithms for convolutivemixtures based on second-order statistics [J].IEEETransactions on Speech and Audio Processing.2005, 13(1):120-134[6]Gorokhov A and Loubaton P. Subspace-based techniques forblind separation of convolutive mixtures with temporallycorrelated sources [J].IEEE Transactions on Circuits andSystems.1997, 44(9):813-820[7]Ghennioui H, Fadaili E M, and Moreau N T, et al.. Anonunitary joint block diagonalization algorithm for blindseparation of convolutive mixtures of sources [J].IEEE SignalProcessing Letters.2007, 14(11):860-863[8]Sawada H, Mukai R, and Araki S, et al.. A robust and precisemethod for solving the permutation problem offrequency-domain blind source separation [J].IEEETransactions on Speech and Audio Processing.2004, 12(5):530-538[9]He Z S, Xie S L, and Ding S X, et al.. Convolutive blindsource separation in the frequency domain based on sparserepresentation [J].IEEE Transactions on Audio, Speech, andLanguage Processing.2007, 15(5):1551-1563[10]Castella M, Rhioui S, and Moreau E, et al.. Quadratic higherorder criteria for iterative blind separation of a MIMOconvolutive mixture of sources [J].IEEE Transactions onSignal Processing.2007, 55(1):218-232

施引文献

资源附件(0)

访问统计