Independent component analysis, a new concept?
Signal Processing - Special issue on higher order statistics
Adaptive filter theory (3rd ed.)
Adaptive filter theory (3rd ed.)
On-line Convolutive Blind Source Separation of Non-Stationary Signals
Journal of VLSI Signal Processing Systems
ICASSP '01 Proceedings of the Acoustics, Speech, and Signal Processing, 2001. on IEEE International Conference - Volume 05
A frequency domain blind signal separation method based ondecorrelation
IEEE Transactions on Signal Processing
Multichannel signal separation: methods and analysis
IEEE Transactions on Signal Processing
Signal separation by symmetric adaptive decorrelation: stability,convergence, and uniqueness
IEEE Transactions on Signal Processing
Source separation using a criterion based on second-orderstatistics
IEEE Transactions on Signal Processing
A globally convergent approach for blind MIMO adaptivedeconvolution
IEEE Transactions on Signal Processing
A maximum likelihood approach to blind multiuser interferencecancellation
IEEE Transactions on Signal Processing
Frequency domain blind MIMO system identification based on second and higher order statistics
IEEE Transactions on Signal Processing
Hi-index | 0.08 |
This paper mainly deals with the necessity and the sufficiency of decorrelation criterion for the separation of convolutive mixtures. Although this problem has been studied in frequency domain, it is still necessary to reformulate this problem in time domain. In this paper, it is proved that decorrelation is a sufficient condition for the separation of convolutive mixtures in time domain, and some new results are presented in this paper. First, separation filters do not always approximate the mixing filters (the responses of the channels in which the sources are transformed and mixed) in time domain, they much frequently approximate the mixing filters in the nonzero spectral band of sources in frequency domain, so the separation filters are usually not unique. Second, imposing some constraints on the equivalent mixing filters, the blind separation of convolutely mixed signals is proved to be equivalent to the optimum filtering problem on the basis of the backward separation system. Third, we propose the Double-LMS and -RLS algorithms for the separation of two convolutely mixed sources by means of the standard LMS and RLS algorithms. These algorithms are naturally generalized to multi-source separation problem, that is, the Multi-LMS and -RLS algorithms. The numerical experiments are presented to illustrate the validity of our algorithms.