The performance analysis of a new-type neural network code-division multiple-access receiver

The problems of optimal as well as suboptimal separation for Code-Division Multiple-Access (CDMA) signals have been the focus of study in the recent past. In the paper, we present a new CDMA receiver: Gram Charlier Probabilistic Neural Network (GCPNN) Receiver. It is based on Gram-Charlier Series Expansion (GCSE) and the maximum likelihood detection theory. Instead of treating the multiuser interference as white Gaussian noise simply as the traditional matched filter CDMA receiver, the new method proposed here uses the GCPNN to estimate the likelihood function of the detecting signal. The performance of this receiver is evaluated and compared with the matched filter.