Adaptive wavelets neural networks based multiuser detector

Abstract

The capacity and performance of code division multiple access (CDMA) system are limited by multiple access interference (MAI) and "near-far" problem, the effect on receivers of which depends on the users' signatures and the actual detector in the receiver. An adaptive wavelets neural networks (AWNN) based multiuser detector is proposed for demodulation of direct sequence CDMA (DS-CDMA) signals in both synchronous and asynchronous Gaussian channels, the complexity of which only lies on that of AWNN. The performance analysis of the detector are carried out by Monte Carlo simulations. The results show it greatly exceeds the matched filter detector and the multilayer perceptron based multiuser detector. In addition, it approaches to the matched filters under single user scenario.