A semiblind approach to optimum multiuser detection

A novel multiuser semiblind demodulation scheme is proposed. Channel estimation is carried out according to the Maximum Likelihood (ML) principle and using the Expectation Maximization (EM) algorithm. Afterwards, the channel state information is used to implement the joint optimum detector with a dynamic programming algorithm (e.g. the Viterbi algorithm). The resulting demodulator is termed semiblind because the statistical features of the received signals are exploited together with the a priori knowledge of a very small number of transmitted symbols. Computer simulations show that the proposed scheme leads to practically optimum performance in Code Division Multiple Access (CDMA) time dispersive channels.