A class of asymptotically optimum iterated-decision multiuser detectors

A promising class of nonlinear multiuser detector is introduced for CDMA systems. These "iterated-decision" multiuser detectors use optimized multipass algorithms to successively cancel multiple-access interference (MAI) from received data and generate symbol decisions whose reliability increases monotonically with each iteration. They significantly outperform decorrelating detectors and linear minimum mean-square error (MMSE) multiuser detectors, but have the same order of computational complexity. When the ratio of the number of users to the spreading factor is below a certain threshold, iterated-decision multiuser detectors asymptotically achieve the performance of the "optimum" multiuser detector, ie, maximum-likelihood (ML) decoding.