Joint Space-Time Equalization and Multiuser Detection for High Data Rate Users in DS-CDMA Systems with Data Selective Adaptive Recurrent Neural Networks

Abstract

A space-time adaptive decision feedback (DF) multiuser receiver (MUD) for DS-CDMA systems using recurrent neural networks (RNN) is proposed for joint equalization and interference suppression. A space-time DF CDMA multiuser detection structure is presented with dynamically driven RNNs in the feedforward section and a finite impulse response (FIR) linear filter in the feedback section for performing interference cancellation. A data selective stochastic gradient (SG) algorithm, based upon the set-membership design criterion, is developed for estimating the parameters of the proposed receiver structure. Simulation results show that the proposed techniques achieve significant performance gains over existing ones