New Issues on the Criterion Properties and Algorithm Convergence of the Generalized CM Approach

A. O. Neves, M. D. Miranda, J. M. T. Romano

DOI: 10.14209/its.2002.314

Evento: 2002 International Telecommunications Symposium (ITS2002)

Keywords:

Abstract

"The first aim of this work is to extend some recent and important results on the equivalence between Shalvi-Weinstein and Constant Modulus methods, by including the so called Generalized Constant Modulus criterion and its corresponding stochastic algorithm. Such method avoid the power restriction on the optimization process, based on the idea that any constant modulus signal at the output of the equalizer represents the achievement of equalization. We show the equivalence between this alternative criterion and the Shalvi-Weinstein one. As far as the corresponding algorithm is concerned, we introduce a study on an alternative stochastic approximation, which leads to an increased performance in terms of convergence rate. The behavior is analyzed in terms of the stochastic gradient vector and evaluated as a function of the algorithm parameters. The dependence of the algorithm in terms of its initial conditions is also analyzed and a modified center-spike procedure is proposed to attain an improved convergence behavior."

Download