Blind separation of instantaneous mixed Gaussian sources via genetic algorithms
Guo Tongheng, Chundi Mu · 2003
A method for blind source separation (BSS) of an instantaneous mixture of colored sources is proposed. It is based on minimizing a Gaussian mutual information criterion, leading to a second-order procedure, which amounts to jointly reducing a set of forward prediction error. Separation is shown to be achievable (up to a scaling and a permutation). Efficient real number genetic algorithms for the joint minimization of mean-squared prediction error are described. Some simulations are performed to show good performance can be attained by a relative small prediction order.