Blind Source Separation Based on Generalised Canonical Correlation Analysis and Its Adaptive Realization
Wei Liu, Danilo P. Mandic, Andrzej S Cichocki · 2008
The canonical correlation analysis (CCA) approach is generalised to accommodate the case with added white noise. It is then applied to the blind source separation (BSS) problem for noisy mixtures. An adaptive blind source extraction algorithm is derived based on this idea. A proof is provided that by this generalised CCA approach, the source signals can be recovered successfully, which is also supported by simulation results.