Neural networks for multiuser detector in asynchronous PC/SS communication system

Proposes a neural network for a multiuser detector in a parallel combinatory/spread spectrum (PC/SS) multiple access communication system. In the PC/SS system, M orthogonal sequences are assigned for one user, and q out of M (q/spl les/M) sequences are used for transmitting information. The combination of sequences can transmit more information than the conventional CDMA systems. In general, all of the users' signals overlap in time and frequency and cause mutual interference. Multiuser detection is a vital technique for combating mutual interference and near-far problems. The paper proposes a neural system for the multiuser detector in the PC/SS, which can output three values, +1, -1 and 0. It is compared with the conventional detector, and its performance evaluated.