Blind separation for sub-/super-Gaussian sources with momentum term based on entropy maximization

Blind source separation consists in processing a set of observed mixed signals to separate them into a set of original components. In this paper, an adaptive blind source separation method based on the entropy maximization criterion is proposed. Momentum term is added into the updating rules to speed up the algorithm and improve the convergence property. Moreover, an adaptive estimation of the score function for both sub-Gaussian and super-Gaussian signals is proposed. Simulation results show that the proposed method can separate signals with different kurtosis.