Hybrid simplex genetic algorithm for blind equalization using RBF networks

The purpose of this paper is to derive a hybrid simplex genetic algorithm for nonlinear channel blind equalization using RBF networks. Most of the algorithms for blind equalization are focused on linear channel models because of their simplicity. However, most practical channels are better approximated by nonlinear models. In order to find an effective method for nonlinear channel blind equalization, here, the equalizer based on RBF networks which is constructed from channel output states instead of the channel parameters is considered. Using the Bayesian likelihood cost function defined as the accumulation of the natural logarithm of the Bayesian decision variable, the problem becomes to maximize the Bayesian likelihood cost function with the dataset which composes the RBF equalizer’s center. For this high dimensional complex optimal problem, the proposed hybrid simplex genetic algorithm solves it by incorporating the simplex operator with GA, and obtains a good convergence characteristic and satisfied equalization result.