非正交联合对角化盲分离算法的可辨识性研究

张延良; 楼顺天; 张伟涛

doi:10.3724/SP.J.1146.2009.00750

非正交联合对角化盲分离算法的可辨识性研究

doi: 10.3724/SP.J.1146.2009.00750

张延良^{①②;楼顺天,},
楼顺天,
张伟涛

计量
- 文章访问数: 4172
- HTML全文浏览量: 105
- PDF下载量: 996
- 被引次数: 0
出版历程
- 收稿日期: 2009-05-15
- 修回日期: 2009-12-01
- 刊出日期: 2010-05-19

A Study of Identifiability for Blind Signal Separation via Nonorthogonal Joint Diagonalization

摘要

摘要: 该文从非正交联合对角化的唯一性条件出发，研究了盲分离算法的可辨识性问题。由接收信号的二阶统计量和高阶累积量分别组成的目标矩阵具有可对角化的结构，因此可以用非正交联合对角化的方法解决盲分离问题。指出非正交联合对角化的唯一存在条件是：由对角矩阵中相同位置的对角元素所组成的向量两两线性无关。从该条件出发推导出基于二阶统计量的非正交联合对角化算法实现盲分离的充分必要条件是源信号自相关函数的形状不同，基于高阶累积量的算法实现盲分离的充分必要条件是源信号中没有高斯信号，从而为运用非正交联合对角化解决盲分离问题提供了理论指导。数值仿真试验验证了结论的正确性。
- 信号处理; 盲信源分离; 联合对角化; 可辨识性; 唯一存在条件; 高阶累积量
Abstract: Based on the uniqueness condition of the solution of Nonorthogonal Joint Diagonalization (NJD), the identifiability for Blind Signal Separation (BSS) is analyzed. Firstly, it is proved that the target matrices consisting of Second-Order Statistics (SOS) or higher-order cumulant are diagonalizable, so the problem of BSS can be solved by NJD. The uniqueness condition for NJD is that the vectors consisting of diagonal elements in the same position of diagonal matrix are pairwise linearly independent. From this proposition，the necessary and sufficient condition for BSS is deduced. For second-order statistics based BSS, the condition is that the source signals have not the identical autocorrelation shape. For higher-order cumulant, there is not Gaussian signal in sources. The above conclusion provides a mathematical foundation for the BSS methods based on the NJD. Numerical simulations confirm the conclusion in this paper.

HTML全文

参考文献(1)

Cardoso J F and Souloumiac A. Blind beamforming for non Gaussian signals[J].IEE Proceedings, Part F: Radar and Signal Processing.1993, 140(6):362-370[2]Belouchrani A, Meraim K, Cardoso J F, and Moulines E. A blind source separation technique using second-order statistics[J].IEEE Transactions on Signal Processing.1997, 45(2):434-444[3]Yeredor A. Non-orthogonal joint diagonalization in the least-squares sense with application in blind source separation[J].IEEE Transactions on Signal Processing.2002, 50(7):1545-1553[4]Ziehe A, Laskov P, Nolte G, and Mller K R.A fast algorithm for joint diagonalization with non-orthogonal transformations and its application to blind source separation[J]. Journal of Machine Learning Research, 2004, 5(12): 777-800.[5]Li X L and Zhang X D. Nonorthogonal joint diagonalization free of degenerate solution[J].IEEE Transactions on Signal Processing.2007, 55(5):1803-1814[6]Wang Fu-xiang, Liu Zhong-kan, and Zhang Jun. Nonorthogonal joint diagonalization algorithm based on trigonometric parameterization[J].IEEE Transactions on Signal Processing.2007, 55(11):5299-5308[7]Souloumiac A. Non-orthogonal joint diagonalization by combining givens and hyperbolic rotations[J].IEEE Transactions on Signal Processing.2009, 57(6):2222-2231[8]Zhang Hua, Feng Da-zheng, and Zheng Wei-xing. A study of identifiability for blind source separation via nonorthogonal joint diagonalization[C]. IEEE International Symposium on Circuits and Systems, Seattle, Washington, USA, 2008: 3230-3233.[9]Comon P. Canonical tensor decompositions[R]. Technology report, Laboratory of Information Signal System, French National Center for Scientific Research, June, 2004.[10]Giorgio T and Rasmus B. A comparison of algorithms for tting the PARAFAC model[J].Computational Statistics and Data Analysis.2006, 50(7):1700-1734[11]De Lathauwer L. A link between the canonical decomposition in multilinear algebra and simultaneous matrix diagonalization[J]. SIAM Journal on Matrix Analysis and Applications, 2006, (28): 642-666.[12]李细林. 盲信号分离中的联合对角化和相位恢复问题研究[D]. [博士论文]，清华大学，2008.[13]Li Xi-lin. Studies on joint diagonalization and phase recovery in blind soure separation[D]. [Ph.D. dissertation], Tsinghua University, 2008.[14]Ten Berge J M F and Sidiropoulos N D. On uniqueness in CANDECOMP/PARAFAC[J]. Psychometrika, 2002, (67): 399-409.[15]Cao Xi-ren and Liu Ruey-wen. General approach to blind source separation[J].IEEE Transactions on Signal Processing.1996, 44(3):562-571

施引文献

资源附件(0)

访问统计