Bayesian and RBF structures for wireless communications detection

This work presents two different algorithms for multiuser detection in wireless DS/CDMA environments. First, a Bayesian detector which implements merging techniques, based on natural computation selection strategies, for complexity limitation, is analyzed, and, second, a low complexity radial basis function-based detector is presented. Both approaches share in common a low computational load and the capability to be implemented even with a high number of active users, since their complexity does not increase exponentially with it. Their performance and characteristics are compared with those of traditional multiuser detectors, such as the matched filter, the decorrelator and the MMSE detector, as well as with other low complexity detectors based on evolutionary computation methods.