A novel cluster based MLSE equalizer for M-PAM signaling schemes

In this paper a new cluster based Maximum Likelihood Sequence Equalizer is presented. The novelty of the algorithm consists of a new technique for the estimation of all the centers around which the received observations are clustered. For a channel of order L and M-PAM signaling scheme, only L of the cluster centers need to be estimated, and the rest, M-L-L, are subsequently computed via simple operations. This has a two-fold advantage compared to previously proposed cluster based algorithms. It reduces dramatically both the computational complexity and the required length of the training sequence. The new method is compared with the standard LMS and RLS based MLSE and the Bayesian RBF equalizer. Moreover, the overmodeling and the undermodeling cases are also explored. The results are very favorable for the new technique, from the computational as well as the performance point of view. (C) 2003 Elsevier B.V. All rights reserved.