Blind source separation based on non-Gaussianity and generalized complexity pursuit
Kui Xu · Journal of Circuits and Systems · 2010
One often solves the BSS problem by using the statistical properties of original sources,e.g.,non-Gaussianity or time-structure information.Nevertheless,real-life mixtures are likely to contain both non-Gaussianity and time-structure information,thus the algorithms which use only one statistical property would be fail.We address BSS problem when source signals have non-Gaussianity and nonlinear predictability.An objective function based on the two statistical characteristics of sources is proposed.Minimizing the objective function,a gradient descent blind source separation algorithm is proposed.The validity of the proposed algorithm is demonstrated by computer simulation.Moreover,comparisons with the existing algorithms indicate the better performance.