Bayesian approach of multiuser separation and interference suppression in CDMA communication systems

Abstract

Based on Bayesian philosophy, we propose a simplified detection scheme for multiuser separation in CDMA communication systems. The proposed detector uses a multi-stage approach to achieve improved detection performance. In the first stage, we use a marginal Bayesian estimate to get a tentative estimate of the random signal vector. In the second stage, a refined estimate of the multiuser information bit is obtained by incorporating the correlation structure of the noise. This detector avoids the multidimensional integration associated with the general multi-dimensional Bayesian estimate. Simulation results show improved performance over existing detectors of comparable simplicity.