Harmony search heuristics for quasi-asynchronous CDMA detection with M-PAM signalling

Focusing on CDMA (Code Division Multiple Access) uplink communications, this paper addresses the application of heuristic techniques to the multiple user detection problem when dealing with asynchrony between transmitters and bandwidth-limited PAM (Pulse AmplitudeModulation) signals. In such systems it is known that, even for the simplest case of binary modulated signals with perfectly synchronous transmitters, simple Single-User Detection (SUD) techniques (e.g. Rake receiver) are outperformed by Multiple-User Detection (MUD) schemes (based on the Maximum-Likelihood - ML - criteria), at a computational cost exponentially increasing with the number of users. Consequently, Genetic Algorithms (GA) have been extensively studied during the last decade as a means to alleviate the computational complexity of CDMA MUD detectors while incurring, at the same time, in a negligible error rate penalty. In this manuscript, a novel heuristic approach inspired in the recent Harmony Search algorithm will be shown to provide a faster convergence and a better error rate performance than conventional GA's in presence of inter-user asynchrony in bandwidth-limited CDMA communications, specially when the complexity of the scenario increases.