Neural networks for blind equalization

This paper presents general guidelines for applying neural networks (NNs) to blind equalization. Firstly, basic concepts and practical issues related to NNs and traditional equalization techniques are discussed. Then, the main neural-network-based solutions for equalization are reviewed and classified in four groups, depending on the type of equalization (supervised or blind) and the use of NNs (nonlinear filter or classifier). Finally, based on conclusions drawn from the analysis of the considered papers, a new effective neural solution is proposed for blind equalization.