Time-varying channel neural equalisation using Gauss-Newton algorithm

Artificial neural network techniques have become very common as equalisation solutions in several types of communication channels. These neural networks are presented in many topologies. The suitable choice of a topology for equalisation purpose depends on different criteria such as: convergence rate, bit error rate, computational complexity, among many others. Reported is an investigation into the behaviour of a structure similar to a decision feedback equaliser employed to equalise time-varying channels. The structure, a single recurrent perceptron, is based on a simplified recurrent neural network. The Gauss-Newton algorithm has been used to estimate the synaptic weights of the perceptron during the training and testing phases. Despite the simplicity of implementation and low computational cost, it has been shown that the proposed topology presents some good comparative performances compared with more complex structures based on recurrent neural networks and multilayer perceptrons using Kalman filters.