Research On Beamforming Technologies In Massive MIMO Systems

Due to extraordinary performance of energy efficiency(EE)and spec-trum efficiency(SE),the massive multiple-input multiple-output(MIMO)tech-nology(also called the large-scale antennas system)has been recognized as a promising technique for the fifth generation mobile communication systems.However,there are still some bottlenecks faced by the massive MIMO itself,such as practical deployment difficulties,expensive cost and high power con-sumption,in addition to its integration with some promising techniques(e.g.,coordinated multi-point(CoMP)transmission,full-duplex relay(FDR)and ul-tra dense/heterogeneous networks etc.).Specifically,the three-dimensional(3D)space channel model,hybrid antenna architecture,imperfect channel state information(CSI)and the larger energy efficiency as well as other issues pose new challenges to the beamforming design in massive MIMO.This dissertation studies massive MIMO based energy efficient and spectrum efficient transmit beamforming problems and provides some solutions to these problems.Firstly,the implementation of the massive MIMO under 3D-MIMO chan-nels is studied.Two 3D dynamic beamforming schemes based on codebook and singular value decomposition(SVD)are proposed to realize full spatial resolution of the elevation as well as the traditional azimuth.Furthermore,a limited feedback precoding scheme is also proposed to realize multi-user MIMO(MU-MIMO)multiplexing gain.System-level simulation results are presented to demonstrate that the proposed 3D beamforming schemes signif-icantly outperform the traditional fixed downtilt beamforming in terms of the cell-average,cell-edge SE performance and cell throughput.The SVD based scheme can overcome the system performance loss due to codebook quantiza-tion error in codebook based scheme.Moreover,the system performance of the proposed limited precoding scheme is much superior to the classical maximum-ratio transmission(MRT)precoding.As the number of multiplexed users in-creasing in the cell,the cell throughput increase significantly due to large MU-MIMO multiplexing gains.For example,compared to two multiplexed users,there is 61%gain in terms of cell average throughput for 8 multiplexed users.Secondly,considering the fully-connected hybrid antenna architecture uti-lized by the coordinated multi-cell cellular networks where base stations equipp-ed with massive MIMO,we propose a hybrid beamforming design to reduce base station's hardware cost and energy consumption.To this end,a weighted sum energy efficiency maximization problem is formulated.In order to reduce the computation complexity,a two-stage design is proposed.At the first stage,a relaxed fractional problem is reformulated by changing the product of analog beamforming matrix and digital beamforming vector into a new vector vari-able,and then its optimal solutions can be derived from the proposed two-level iterative algorithm.Then,two effective alternating minimization algorithms based on manifold optimization and an empirical assumption are proposed for the hybrid beamforming design.Simulation results are presented to show its superior over full-digital design in both base station's implementation cost and EE performance.Next,a massive MIMO enabled heterogeneous network framework,con-sisting of one macrocell base station(MBS)equipped with switch mode hybrid antenna architecture is considered.In the presence of Gaussian CSI errors,an outage constrained robust hybrid coordinated beamforming(HyCoBF)prob-lem is formulated by minimizing the total transmit power under the preassigned outage probability constraints for users' target SINR.This nonconvex combina-tional optimization problem is solved by the proposed two-stage method.The analog beamforming mechanism at MBS is a newly devised low-complexity beam selection scheme.Then a coordinated beamforming(CoBF)problem based on effective digital channels is reformulated into an semidefinite pro-gram(SDP)by semidefinite relaxation(SDR)and an extended Bernstein-type inequality.Furthermore,a distributed implementation for the obtained CoBF solution using alternating direction method of multipliers(ADMM)is also pro-posed.Simulation results are then provided to show that the proposed HyCoBF design meets systems' design requirement and is robust against CSI errors with total power performance comparable to the full-digital design.Finally,for a mutli-pair(source node and destination node pair)FDR-based massive MIMO system,where the FDR is equipped with large-scale an-tenna arrays,the self-interference cancellation(by considering large number of antennas)at the FDR is analyzed under imperfect Rician fading channels.By random matrix theory,we derive the closed-form achievable rate expressions for the FDR transceiver using maximum-ratio combining/MRT(MRC/MRT)and those using zero-forcing(ZF)processing.In addition,the system spectral efficiency under different transmit power allocations is also derived.Theoreti-cal analysis and numerical results show that when the number of antennas at the FDR is sufficiently large,the self-interference,inter-pair interference and noise can be canceled out completely,and that the system spectral efficiency is double of half-duplex mode.Furthermore,in the presence of imperfect CSI,inter-relay interference,multi-user interference and self-interference caused by FDRs,an ADMM-based distributed worst-case robust design of FDR beamforming and total power minimization for downlink transmission is proposed for a FDR-aided wireless communication system consisting of one base station equipped with large-scale antennas and multiple MIMO FDRs.Some simulation results are provided to demonstrate that the proposed distributed algorithm meets sys-tems' design requirement,and is robust against CSI errors in the meantime.