Distributed decoding in a cellular multiple-access channel

This paper considers the problem of joint detection in the uplink of cellular multiaccess networks with base-station cooperation. Distributed multiuser detection algorithms with local message passing among neighbor base stations are proposed and compared in terms of computational complexity required in the base stations, the amount of serial communications among them, error rate performance, and convergence speed. The algorithms based on the belief propagation algorithm result in complexity and delay per base station which do not grow as the network size increases. In addition, it is observed that these algorithms have near single-user error rate performance for the fading channels considered. Thus it is illustrated that using the belief propagation algorithm, it is possible to have full frequency re-use and achieve near-optimal performance with moderate computational complexity and a limited amount of message passing between base stations of adjacent cells.