The Second-Order Blind Identification (SOBI) algorithm (Belouchrani et al., 1997) is aimed at blind separation of static mixtures of stationary source signals with distinct spectra. It uses approximate joint diagonalization of empirical correlation matrices to estimate the mixing matrix. We show that SOBI's performance can be improved by transforming the joint diagonalization into a properly weighted nonlinear Least Squares problem. In the case of Gaussian sources, the optimal weights can be estimated consistently from the empirical correlation matrices. We demonstrate the substantial improvement by analysis and simulations.
|Journal||European Signal Processing Conference|
|State||Published - 31 Mar 2000|
|Event||2000 10th European Signal Processing Conference, EUSIPCO 2000 - Tampere, Finland|
Duration: 4 Sep 2000 → 8 Sep 2000