基于稳健联合分块对角化的卷积盲分离

汤辉; 王殊

doi:10.3724/SP.J.1004.2013.01502

基于稳健联合分块对角化的卷积盲分离

doi: 10.3724/SP.J.1004.2013.01502

汤辉,
王殊

1.
华中科技大学电信系武汉 430074

详细信息

作者简介:
汤辉华中科技大学电信系博士研究生.2006年获得华中科技大学电信系学士学位.主要研究方向为扩频信号检测和盲信号处理.E-mail:balatanghui@163.com

计量
- 文章访问数: 1263
- HTML全文浏览量: 41
- PDF下载量: 1131
- 被引次数: 0
出版历程
- 收稿日期: 2012-04-25
- 修回日期: 2012-06-29
- 刊出日期: 2013-09-20

Robust Joint Block Diagonalization Based Convolutive Blind Source Separation

TANG Hui,
WANG Shu

1.
Department of Electronic and Information, Huazhong University of Science and Technology, Wuhan 430074

摘要

摘要: 针对卷积盲分离问题,提出一种新的矩阵联合分块对角化(Joint block diagonalization, JBD)算法. 现有的迭代非正交联合分块对角化算法都存在不收敛的情况,本文利用分离矩阵的特殊结构确保其可逆性,使得算法的迭代过程稳定. 在已知矩阵分块结构的条件下,首先,将卷积盲分离模型写成瞬时形式,并说明其满足联合分块对角化结构; 然后,提出联合分块对角化的代价函数,依据代价函数的最小化等价于矩阵中每个分块的范数最小化, 将整个分离矩阵的迭代更新转化成每个分块的迭代更新;最后,利用最小化条件得到迭代算法. 实数和复数两种情况下的算法都进行了推导.基本实验验证了新算法在不同条件下的性能; 仿真实验中对在时域和频域都重叠的信号的卷积混合进行盲分离,实验结果验证了新算法具有更好的分离性能和更稳定的分离能力.
- 卷积盲分离 /
- 联合分块对角化 /
- 稳健迭代 /
- 矩阵范数
Abstract: A new joint block diagonalization (JBD) algorithm is proposed to solve the convolutive blind source separation problem. The existing iterative algorithms may not converge to the correct solution. We use a special structured separation matrix which is always invertible to avoid the divergence of the algorithm. First, the convolutive mixture is rewritten as an instantaneous one which satisfies the joint block diagonalization model. Second, the cost function is built with the priori information of block structure. Then the whole matrix iteration is transformed into update of the every block sub-matrix in the sense that the minimization of the cost function is equivalent to the minimization of Frobenius norm of each block. The iterative algorithm is deduced both in the situations of real and complex models. Sanity check experiment has verified the good performance of the new algorithm in different conditions. In the application of blind separation of signals which are both overlaping in the time and frequency domains, simulation results demonstrate the better performance and more robust ability of the proposed algorithm, as compared to others.
- Convolutive blind source separation /
- joint block diagonalization (JBD) /
- robust iteration /
- matrix Frobenius norm

HTML全文

参考文献(14)

[1]	Yang Zhu-Qing, Li Yong, Hu De-Wen. Independent component analysis: a survey. Acta Automation Sinica, 2002, 28(5): 762-772(杨竹青, 李勇, 胡德文. 独立成分分析方法综述. 自动化学报, 2002, 28(5): 762-772)
[2]	Castella M, Bianchi P, Chevereuil A, Prequet J C. A blind source separation framework for detecting CPM sources mixed by a convolutive MIMO filter. Signal Processing, 2006, 86(8): 1950-1967
[3]	He Z S, Xie S L, Ding S X, Cichocki A. Convolutive blind source separation in the frequency domain based on sparse representation. IEEE Transactions on Audio, Speech, and Language Processing, 2007, 15(5): 1551-1563
[4]	Abed-Meraim K, Belouchrani A. Algorithms for joint block diagonalization. In: Proceedings of the 12th European Signal Processing Conference. Vienna, Austria: IEEE, 2004. 209-212
[5]	Maehara T, Murota K. Algorithm for error-controlled simultaneous block-diagonalization of matrices. SIAM Journal on Matrix Analysis and Applications, 2011, 32(2): 605-620
[6]	Ghennioui H, Fadaili E M, Thirion-Moreau N, Adib A, Moreau E. A nonunitary joint block diagonalization algorithm for blind separation of convolutive mixtures of sources. IEEE Signal Processing Letters, 2007, 14(11): 860-863
[7]	Ghennioui H, Thirion-Moreau N, Moreau E, Aboutajdine D. Gradient-based joint block diagonalization algorithms: application to blind separation of fir convolutive mixtures. Signal Processing, 2010, 90(6): 1836-1849
[8]	Xu X F, Feng D Z, Zheng W X, Zhang H. Convolutive blind source separation based on joint block Toeplitzation and block-inner diagonalization. Signal Processing, 2010, 90(1): 119-133
[9]	Nion D. A tensor framework for nonunitary joint block diagonalization. IEEE Transactions on Signal Processing, 2011, 59(10): 4585-4594
[10]	Tichavský P, Yeredor A, Koldovský Z. On computation of approximate joint block-diagonalization using ordinary AJD. In: Proceedings of the 10th International Conference on Latent Variable Analysis and Signal Separation. Telaviv, Israel: Springer, 2012. 163-171
[11]	Lahat D, Cardoso J F, Messer H. Joint block diagonalization algorithms for optimal separation of multidimensional components. In: Proceedings of the 10th International Conference on Latent Variable Analysis and Signal Separation. Telaviv, Israel: Springer, 2012. 155-162
[12]	Ziehe A, Lascov P, Nolte G, Müller K R. A fast algorithm for joint diagonalization with nonorthogonal transformation and its application to blind source separation. Journal of Machine Learning Research, 2004, 5: 777-800
[13]	Zhang Xian-Da. Matrix Analysis and Applications. Beijing: Tsinghua University Press, 2004. 100-118(张贤达. 矩阵分析与应用. 北京: 清华大学出版社, 2004. 100-118)
[14]	Theis F J. Uniqueness of complex and multidimensional independent component analysis. Signal Processing, 2004, 84(5): 951-956