Minimum Mean-Square Error and Maximum Likelihood Multiuser Detection: Statistical Properties and Applications

Although multiuser detection (MUD) has received intensive research since its invention by Verdu in 1983, however, the statistical properties in MUD have not received enough attention in research and application. In this contribution, we first study the statistical properties of minimum mean-square error (MMSE)-MUD, maximum likelihood (ML)-MUD and hybrid MMSE/ML-MUD from different aspects in the context of direct-sequence code-division multiple-access (DS-CDMA) communicating over additive white Gaussian noise (AWGN) channels. Both the amplitude statistics and ratio statistics are investigated. Then, the applications of the statistics of MUD are discussed. Our study shows that the statistics embedded in the MUD schemes may be exploited for design of high- efficiency search algorithms in ML-MUD, for design of good spreading sequences, for post-MUD processing enhancement, for design of novel MUD schemes, etc.