A Bayesian receiver for asynchronous code division multiple access communications

An adaptive Bayesian receiver structure for use in asynchronous multi-user code division multiple access (CDMA) systems is presented. The structure consists of a bank of eigenfilters and a radial basis function (RBF) network. This receiver can be trained to form the Bayesian non-linear decision boundary. Simulations demonstrate that it outperforms conventional matched filter detectors. Bit-error-rate curves are presented which show that it is robust, near-far resistant and able to cope with multipath. Its performance is similar to that of optimal maximum likelihood detectors with the advantage that it is less complex and inherently adaptive