Reduced-complexity clustering techniques for nonlinear channel equalization

In this paper a novel reduced complexity 1-dimentional implementation of the clustering based sequence equalizer is given. The technique is appropriate for nonlinear channels and leads to robust performance. No explicit channel modeling is required. A detailed comparison, with respect to the performance and computational complexity, with an optimum symbol by the symbol Bayesian equalizer with decision feedback is provided. Simulations are carried out both for 2-PAM and 4-PBM modulation schemes