Iterative differential decorrelation for nonstationary blind source separation
P. Ubaidulla, N. Ravikiran · 2005
In this work we address the problem of blind separation for the case where the sources are second-order non-stationary. The concept of differential correlation is used to capture the information contained in the time variation of the second order statistics of the source signals. We develop an iterative least squares algorithm based on the principle of correlation matching and the concept of differential correlation.