A nonlinear multiuser receiver for UMTS communications

The capacity of direct-sequence code division multiple access (DS-CDMA), under which third generation (3G) cellular systems operate, is limited by multiple access interference (MAI). A DS-CDMA synchronous system with multipath propagation channel is being considered A DS-CDMA nonlinear adaptive receiver based on neural networks is herein proposed. This receiver has the ability to mitigate the near/far problem and at the same time, to minimize the effect of MAI in a multipath channel. A comparative study analysis of the conventional RAKE receiver and the proposed method is presented under 3G considerations. It has been observed that the neural network receiver, with an appropriate and stable architecture, allows fast learning. We show that the performance is appreciable in a near/far environment and is capable of suppressing both MAI and ISI. Unlike the minimum mean square error (MMSE) and interference cancellation (PIC and SIC) receivers, channel identification is not needed by the neural network receiver.