Shuffled Frog-Leaping Algorithm trained RBFNN Equalizer

Abstract

Ability of Artificial Neural Networks (ANN) in mapping between the variables attracts its application in channel equalization. Single hidden layer of Radial Basis function Neural Networks (RBFNN) makes it most popular equalizers to mitigate the channel distortions. Most challenging problem associated with design of RBFNN Equalizer is the traditional hit and trial method. Ability of evolutionary algorithms in solving complex problems in finding global optimal solutions attracted this paper for training of RBFNN equalizer using a recently proposed population based optimization, Shuffled Frog-Leaping Algorithm (SFLA) and three of its modified forms. It is found from the simulation results that performances of different forms of SFLA for the training of RBFNN equalizers are superior as compared to existing equalizers.