LAUSR

In this paper, we study the multiuser relay network in which each transmitter sends messages to its intended receiver with the help of a cluster of intelligent amplify-and-forward (AF) relays. We assume that each transmitter has the channel information of only the link to its corresponding receiver. With this assumption, we propose a joint optimization algorithm to achieve Pareto optimality. Achieving Pareto optimality of our relay network is more complicated than that of the one-hop interference channel since not only the beamforming vectors but also the processing matrices of the AF relays need to be optimized. With fixed relay processing matrices and minimum mean square error with successive interference cancelation (MMSE-SIC) receiver, we first find a sufficient condition for the transmission covariance to be Pareto optimal. Based on this sufficient condition, a transmit beamforming scheme is proposed. Then, by fixing transmit and receive beamforming vectors, we optimize the relay processing matrix and give a suboptimal algorithm to achieve the maximum sum rate and Pareto optimality. Finally, by optimizing transmit beamforming vectors and relay processing matrices alternatively, we obtain the joint optimization algorithm, which can be guaranteed to be convergent. Simulation results show that our joint algorithm achieves much better performance than those schemes compared.