Maximum likelihood sequence estimation of communication signals by a Hopfield neural network

The application of Hopfield´s neural networks for data sequence estimation in digital communication receivers is presented. The Hopfield neural networks are used to perform the maximum-likelihood sequence estimation (MESE), and robust architectures for VLSI realizations are presented. The Hopfield´s neural networks for MESE have several advantages over other applications in that the complexity is proportional to channel memory, the network provides a regularity in architecture, and the problem of vanishing diagonal elements can be relaxed. It has been shown that artificial neural networks have potential abilities to perform optimization problems which occur often in the area of electronic communications