Semiblind decision-feedback multiuser interference cancellation based on the maximum likelihood principle

This paper addresses the problem of interference suppression in Direct Sequence Code Division Multiple Access (DS CDMA) systems. We propose a novel semiblind Decision Feedback (DF) receiver based on the Maximum Likelihood (ML) principle that simultaneously exploits the transmission of training sequences and the statistical information concerning the unknown transmitted symbols. The Space Alternating Generalized Expectation Maximization (SAGE) algorithm allows an efficient iterative implementation of the receiver. Computer simulations show that the resulting multiuser detector attains practically the same performance as the theoretical DF Minimum Mean Square Error (MMSE) receiver.