Asymptotic iterative multi-user detection of random CDMA

Abstract

Recent work has focused on designing efficient receivers for code-division multiple-access (CDMA) channels with forward error control (FEC) coding. In particular with the advent of turbo coding, new receiver structures employing iterative processing have been proposed. A variety of iterative receivers, where multi-user detectors with different complexities are used, have been investigated. In this paper we study such decoders from a large system perspective, where K, N /spl rarr/ /spl infin/. We introduce the variance transfer (VTR) functions which accurately describe the convergence of the complete decoder, and are used to predict performance behavior, such as the onset of the turbo cliff, and the maximal supportable user load /spl alpha/=K/N for a given FEC code and interference suppression method. Closed-form expressions for the simple interference canceller as well as the MMSE filter are used. Among the results we present the guidelines on the choice of error control codes, maximal system loads and accurate performance predictions for given system configurations.