Cooperative uplink cellular systems with multi-antenna relays and heterogeneous users

This paper considers a multi-access cooperative system uplink where multiple users access a base station (BS) through a multi-antenna relay. For a fixed relay beamformer, the optimal user beamformers are determined. Then when user beamformers are fixed, the optimal relay beamformer is derived. An iterative algorithm that jointly optimizes the user and relay beamformers is then proposed and proven to converge to a fixed point. Numerical results reveal that (i) convergence typically requires about three iterations, and (ii) performance of the iterative algorithm is very close to that of the user-relay MIMO-MAC channel and relay-BS MIMO upper bounds. An extension is proposed where multiple relays, serving as access points, co-exist with directly accessing users. While not guaranteeing a globally optimum solution to this non-convex problem, numerical results show that the performance of this extended iterative algorithm is very close to the MIMO MAC upper bound.