A nonlinear function for gradient based BSS

Blind source separation (BSS) deals with separating independent signals form their linear mixtures observed at different sensors. In this paper a nonlinear function based on the cost function as a Kullback-Leibler divergence between the joint probability density function of the source vector and its parametric model, is proposed. This cost function is equivalent to maximization of the information transfer between inputs and outputs, and minimization of the mutual information between components of the output vector. Derivation process becomes extremely simple due to a simple approximation. Simulations with communication signals indicate that the proposed algorithm provides better accuracy