Multi-user detection via iterative processing

Multi-user detection of CDMA signals is studied in the light of iterative processing. The factor graph representation of a CDMA system is used to develop a low-complexity joint detection algorithm. This algorithm replaces the large sums that appear at the multiple access nodes by simple averages and ignores the error control code network of the interfering users, assuming a fixed probability distribution on the output symbols of the interfering users. This is optimal in the case where the code network for the interfering users is unknown, and amounts to an error control decoder using a multi-user metric generator. Simplifications of the multi-user metric generation lead to linear iterative metric generators. Simulation and analytical results on the performance of these detectors are shown and illustrate complexity/performance trade-offs.