Convergence Speed of Iterative Multi-user Detection for Turbo-Coded CDMA

In this paper, we analyze the convergence speed of the iterative multi-user detection turbo-coded CDMA system. The convergence speed is measured in terms of iterations. We are mainly interested in the influence of the system parameters on the convergence speed to the single-user performance. We use the variance exchange graph (VEG) to predict and visualize the convergence speed of the proposed system. The variance exchange graph provides us a detailed description of the iterative process, and enables us to investigate the convergence speed of the system from two angles: first, the influence of multiple access interference (MAI) and channel noise on the convergence speed of the system is studied; second, the convergence speed of the system with different turbo decoding iterations is investigated. Some design guidelines are obtained to choose an efficient iteration pattern for the system.