On the Nonlinear Prediction for Blind Equalization: Is that a Good Choice?

Charles C. Cavalcante, João Cesar M. Mota, Jugurta R. Montalvão Filho, Bernadette Dorizzi

DOI: 10.14209/sbrt.2000.5150129

Evento: XVIII Simpósio Brasileiro de Telecomunicações (SBrT2000)

Keywords:

Abstract

"In this paper we investigate the conditions of applicability of a nonlinear predictor based on neural networks to perform blind equalization using only one input instead of several ones as usual in linear strategies. Linear and nonlinear prediction concepts are revisited in order to clarify some believes. A new self-organized learning algorithm is proposed in order to improve the performance of the nonlinear equalizer. Computational simulations illustrate and compare the performance of the proposed approach with classical structures for equalization."

Download