Relationships between instantaneous blind source separation and multichannel blind deconvolution

We present a general algebraic approach to an extended dynamic independent component analysis (EDICA) for multichannel blind signal separation/deconvolution. Precise algebraic equivalence and direct analogies between instantaneous blind source separation (BSS) and dispersive (dynamic) blind signal separation/deconvolution (referred to also as multichannel blind deconvolution, MBD) problems are shown, as well as, the equivalence of the problem in the time domain and the Z-transform domain. For circular convolution the equivalence (analogy) is precise for finite length time series, while for linear convolution such analogy is valid only in the asymptotic sense for infinite length series. Elegant and concise derivation of learning algorithms in the time domain is presented using the algebraic properties of the convolution operator and relationships between convolution and cross-correlation. Using this general concept, unsupervised learning algorithms (both batch and online algorithms) are developed for multichannel blind deconvolution/separation problems. Computer simulation experiments confirm validity and high performance of the proposed algorithms. The proposed approach and some automatic rules can be applied not only to the known, already existing algorithms for blind separation and extraction of sources but we hope could be also used for extension and generalisation of learning rules developed in future.

[1]  S. Amari,et al.  Fast-convergence filtered regressor algorithms for blind equalisation , 1996 .

[2]  Andrzej Cichocki,et al.  Robust neural networks with on-line learning for blind identification and blind separation of sources , 1996 .

[3]  Jean-François Cardoso,et al.  Equivariant adaptive source separation , 1996, IEEE Trans. Signal Process..

[4]  Terrence J. Sejnowski,et al.  An Information-Maximization Approach to Blind Separation and Blind Deconvolution , 1995, Neural Computation.

[5]  Andrzej Cichocki,et al.  Stability Analysis of Learning Algorithms for Blind Source Separation , 1997, Neural Networks.

[6]  Andrzej Cichocki,et al.  A New Learning Algorithm for Blind Signal Separation , 1995, NIPS.

[7]  A. Cichocki,et al.  Self-adaptive independent component analysis for sub-Gaussian and super-Gaussian mixtures with an un , 1997 .

[8]  Joseph J. Atick,et al.  Convergent Algorithm for Sensory Receptive Field Development , 1993, Neural Computation.

[9]  S.C. Douglas,et al.  Multichannel blind deconvolution and equalization using the natural gradient , 1997, First IEEE Signal Processing Workshop on Signal Processing Advances in Wireless Communications.

[10]  Shun-ichi Amari,et al.  Stability Analysis Of Adaptive Blind Source Separation , 1997 .

[11]  R. Lambert.  Multichannel blind deconvolution: FIR matrix algebra and separation of multipath mixtures , 1996 .

[12]  Andrzej Cichocki,et al.  Robust learning algorithm for blind separation of signals , 1994 .

[13]  E. Ferrara.  Fast implementations of LMS adaptive filters , 1980 .

[14]  T. Sato,et al.  On-line algorithms for blind deconvolution of multichannel linear time-invariant systems , 1997, Proceedings of the IEEE Signal Processing Workshop on Higher-Order Statistics.

[15]  Shun-ichi Amari,et al.  Neural Network Models for Blind Separation of Time Delayed and Convolved Signals , 1997 .

[16]  S.C. Douglas,et al.  On the relationship between blind deconvolution and blind source separation , 1997, Conference Record of the Thirty-First Asilomar Conference on Signals, Systems and Computers (Cat. No.97CB36136).