Using Neural Networks for Adaptive Equalization

. Non linrea distortion introduced by communications channels increases the probability of error. Application of artificial neural network structures to the problem of channel equalizationin a digital communication system has been considered in this paper. The difficulties associated with channel non linearities can be overcome by equalizers employing diagonal recurrent neural network (DRNN). Because of nonlinear processing of signals in an DRNN, it is capable of producing arbitrarily complex decision regions. For this reason, the DRNN is proposed for channel equalization problem. The architecture of DRNN is a modified model of the fully connected recurrent neural network with one hidden layer, and the hidden layer is comprised of self-recurrent neurons. The performance of the proposed network will be compared to other neural networks (already used for channel equalization) through simulations. Key-Words: Adaptive channel equalization, Artificial neural networks, , M-PAM signals.