Recent Advances in Neural Network Techniques for Channel Equalization: A Comprehensive Survey

Abstract

The channel equalization is a technique used in the digital communication system at the receiver side to mitigate the effect of inter-symbol interference in disruptive channels. In this paper a comprehensive survey of the latest research on the modeling of nonlinear phenomena of channel equalization by artificial neural networks (ANNs) is presented. The literature related to different neural network (NN)-based equalization techniques is presented. These NN techniques includes multilayer perceptron, Chebyshev neural network, functional link neural network, fuzzy neural network, radial basis functions and their variants. Feedback-based NN architectures are also described, such as recursive NN equalizers. Convergence and computational complexity comparison of training algorithms for nonlinear channel model is given in this paper. The limitations of recent research in this field and future research recommendations are also provided.