A Bayesian framework for the development of suboptimal blind multiuser detectors for DS/CDMA signals in synchronous channels

In this paper, a new family of near-optimum blind Bayesian multiuser detectors (MUDs) for digital mobile communications is developed. First, a general framework for multiuser Bayesian detection is briefly presented. The extension to the blind case as well as both the study of the capability of blind bit error rate (BER) estimation and some implementation aspects are examined with further detail. Next, the core of the paper focuses on the development of several techniques that are applied to limit the number of hypotheses (possible sets of symbols transmitted by the users of the system) which are considered in the Bayesian approach. Special care must be taken so as to maintain a high diversity within the symbol hypotheses’ set related to the user of interest. Finally, the performance is evaluated and compared to that of some traditional and recent MUDs. Simulations show how the different proposed algorithms offer, each of them, a different balance between computational complexity and performance. The proposed methods exhibit good near–far resistance and estimation accuracy, constituting an attractive and feasible alternative to previous algorithms.