RLS Adaptive Noise Cancellation via QR Decomposition for Noisy ICA

Abstract:

This study addresses the independent component analysis (ICA) in the presence of additive noise via an approach of adaptive filtering. Recursive least squares (RLS) adaptive noise cancellation via QR decomposition (QRRLS) is introduced to reduce the bias in the mixing matrix caused by noise. To test performance of this approach, two kinds of experiments for speech signals are conducted by combining Fast-ICA algorithm with it, on the conditions of identical noise and correlational noises respectively. Moreover, in order to measure the performance availably, the least-squares method is adopted to calculate the signal to noise ratio (SNR) of recovery signals. By comparison, it shows that this approach outperforms the adaptive noise cancellation via least-mean-squares (LMS) algorithm.

• Periodical: Advanced Materials Research (Volumes 403-408)
• Pages: 1291-1296
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.403-408.1291
• Online since: November 2011
• Authors: Ming Liang Zhang, Shu Zhao Wang, Xin Yan Jia
• Keywords: Least-Square Method (LSM), Noisy Independent Component Analysis, QR-Based RLS Adaptive Noise Cancellation

References

[1] S. Amari, A. Hyv¨arinen, and S.Y. Lee, etc. Blind signal separation and independent component analysis. Neurocomputing, 49(2002): 1–5.

DOI: 10.1016/s0925-2312(02)00509-x

[2] S. Choi, A. Cichocki, and H.M. Park, etc. Blind Source Separation and Independent Component Analysis: A Review. Neural Information Processing, 6(2005): 1-57.

[3] A. Hyv¨arinen, J. Karhunen, and E. Oja. Independent Component Analysis. John Wiley, New York, (2001).

[4] A. Hyvarinen. Gaussian moments for noisy independent component analysis. IEEE Signal Processing Letters, 6(1999): 145-147.

DOI: 10.1109/97.763148

[5] Zhenwei Shi, Changshui Zhang. Gaussian moments for noisy complexity pursuit. Neurocomputing, 69(2006): 917–921.

DOI: 10.1016/j.neucom.2005.10.002

[6] A. Cichocki, S.C. Douglas, and S. Amari. Robust techniques for independent component analysis with noisy data. Neurocomputing, 22(1998): 113-129.

DOI: 10.1016/s0925-2312(98)00052-6

[7] M. G. Jafari, J. A. Chambers. Adaptive noise cancellation and blind source separation. 4th international symposium on independent component analysis and blind signal separation (ICA 2003), 2003, 627-632.

[8] Li-li Wu, Qing Cao, and Hui Li. Study of noised speech separation based on ICA and linear neural network. Computer Engineering and Applications, 46(2010): 143-146.

[9] Yulin Liu, Xiaojun Jing, and Gangbing Tan. Adaptive Filtering: Algorithms and Practice Implementation. Publishing House of Electronics Industry, 2004, 192-220.

[10] Wei Yang. The Method of Seismic Random Noise Blind Separation based on ICA. Chengdu university of technology, Chengdu, China, 2009, 38.

[11] Yumin Yang, Chonghui Guo. Gaussian moments for noisy unifying model. Neurocomputing, 71(2008): 3656– 3659.

DOI: 10.1016/j.neucom.2008.03.005

[12] A. Hyvarinen, Fast ICA for noisy data using Gaussian moments. Proceedings of the International Symposium on Circuits and Systems. Orlando, FL, 1999, 57–61.

DOI: 10.1109/iscas.1999.777510

[13] H.M. Parka, S.H. Ohb, and S.Y. Lee. A filter bank approach to independent component analysis and its application to adaptive noise canceling. Neurocomputing, 55(2003): 755 – 759.

DOI: 10.1016/s0925-2312(03)00438-7

[14] H.M. Parka, C. S. Dhir, S.H. Ohc, and S.Y. Lee. A filter bank approach to independent component analysis for convolved mixtures. Neurocomputing, 69(2006): 2065–(2077).

DOI: 10.1016/j.neucom.2005.09.016