Evolutionary programming with cloning and adaptive cost function applied to multi-user DS-CDMA systems

Abstract

A new utilization form for the evolutionary programming (EP) algorithm applied to multi-user DS-CDMA (direct sequence code division multiple access) systems in a synchronous AWGN (additive white Gaussian noise) channel is proposed. In order to find the maximum likelihood (ML) solution, the variance matrix for the generation of new candidates was expressed as a function of the following parameters: E/sub b//N/sub 0/; near-far ratio (NFR); cost function (fitness value). A cloning strategy procedure was introduced where the best vectors with the bit-candidates were cloned, reducing drastically the required time for the algorithm convergence to the single user bound (SuB) performance. An improvement in the convergence process was obtained using a parallel interference canceller with soft decision (SD-PIC) for initial vector generation.