On end-to-end capacity of MIMO nonregenerative relay networks via time scheduling and subchannel pairing

This paper considers a multi-input multi-output (MIMO) wireless relay network, where the source S communicates wirelessly to the destination D via the help of a nonregenerative relay R. Previous studies are almost exclusively based on the assumptions of equal time-duration phases for S-R and R-D transmission and equal number of multiple antennas for every node, and the results concerning arbitrary time scheduling and arbitrary antenna arrays are not yet known. To achieve the best end-to-end data rate, this paper considers joint source and relay beamforming design combined with time scheduling and subchannel pairing, in accordance to respective link conditions. We first propose a practical beamforming framework that is inspired by the optimal MIMO relay beamforming structure. It is shown that in this framework, after subchannel pairing, the joint beamforming design problem can be transformed to a joint optimal power allocation problem. Since the overall optimization problem is nevertheless very difficult, involving both continuous and integer variables as well as a non-convex objective function make, we solve it by enumerating integer variables, and with each given integer values, performing alternative optimization and decomposing the problem into convex subproblems. Numerical results confirm the effectiveness of our approach, and it is shown that relaxing the equal time-duration constraint and pairing subchannels optimally can significantly improve the end-to-end capacity.