Research On The Optimization Algorithms Of Amplify-and-forward MIMO Relay Systems

For its remarkable performance promise, MIMO wireless channels has been widely analyzed both on the theoretical and practical issues. This research was primarily focused on point-to-point(single-hop) communications. But with the evolution of the tendency of mobile network towards densification and heterogeneous, recently there has been more and more work on relay network. And the core of relay network is the optimization design of the tranceivers, expecially the design of the relay nodes. This dissertation focuses on the issues of optimization design of AF MIMO relay systems, which include optimization for interferent multi-user multi-relays MIMO relay systems.Firstly, an up-to-date overview of designing AF MIMO relay systems is provided. In this part of job, the optimization of many kinds of system models are discussed. Secondly, the optimization design of multi-user multi-realays AF MIMO relay systems without direct links is discussed. A minimum mean-squared error(MMSE) based iterative algorithm is proposed to jointly optimize the beamfromming matrices of sources nodes, the precoding matrices of relay nodes and the receiver matrices. To solved the nonconvex optimization problem, we depart the problem into three convex subproblem. In other word, we optimize one of the matrices one time by setting others to be fixed. After gotten the globle optimal results of all the matrices, a high-quality suboptimal result of the original problem is got. Thirdly, we extend the research toward the multi-user multi-relays AF MIMO relay systems with direck links. Two more iterative algorithms is proposed. The first algorithm uses the same idea of the algorithm used in systems without direct links, named as seperation algorithm. And the second algorithm views both all beamformming matrices of source nodes and all precoding matrices of relay nodes as a union and optimizes at the same time, named as union algorithm.According to the matlab simulation results, all algorithms proposed in this dissertation converge quickly after a few iterations and significantly outperform the existing algorithm based on minimum total power leakage in terms of the MSE and BER. The performance of both two proposed algorithms is significantly improved when the direct links are taken into account. On the other hand, the union algorithm do not provide much more performmance increasing comparing with seperation algorithm, even though it gets higher computational complexity.