A new evolutionary algorithm for blind linear channel inversion

This paper proposes a new genetic algorithm based on abstract data types for blind source separation applied to linear channel inversion. Such system allows the inverse filter coefficients to be adjusted by this algorithm considering some criterions, such as, the nongaussianity by kurtosis, measuring mutual information through approximation by cumulant and measuring negentropy through approximation by cumulant in application of blind deconvolution. This paper aims to show the efficiency of a new genetic algorithm in the characterization of a combined solution for complex optimization problems, like the blind inversion of linear channel.