Research On Massive MIMO And Hybrid Beamforming Technologies

Massive MIMO is considered as one of key techonologies of 5G since it can greatly increase spectral efficiency for future communication systems.Meanwhile,hybrid beamforming technology is one of key issues of Massive MIMO in practice.In this thesis,relevant research has been taken on Mas-sive MIMO and hybrid beamforming technologies.Our work not only enrich the theoretical basis of favorable propagation under massive MIMO,but also propose novel and effective schemes for hybrid beamforming used in pratical massive MIMO systems.In details,we make the following specific contribu-tions and achievements.Firstly,theory of favorable propagation under massive MIMO has been deeply and widely investigated in this thesis.Previous research usually focuses on whether large arrays in a specific channel model can approach favourable propagation under the condition of single-antenna user.In this thesis,effect on favorable propagation under massive MIMO with single-antenna user is firstly investigated in a general channel model of multiple antennas with common an-tenna arrays(i.e.,uniform linear arrays,uniform planar arrays and uniform cir-cular arrays),and general conclusions are made thereafter.Furthermore,the thesis extends the research of single-antenna user case to multi-antenna user case.By investigating Rayleigh channels and mm Wave channels,we derive the conclusion which is different from single-antenna user case.Thus,our work enriches the theory of favourable propagation under massive MIMO.Secondly,for pratical deployment of massive MIMO,especially when combined with mm Wave communication,it usually uses hybrid beamforming to reduce system complexity.Under this background,this thesis investigates the multi-user multi-stream hybrid beamforming design with fully-connected subarray formation.Different from the existing idea of designing analog and digital beamforming separately,we propose a novel joint-optimization criterion by exploiting the inherent relationship between analog and digital beamform-ing found with the help of information theory.Assuming perfect channel state information(CSI)is known prior,we respectively propose two theoretical opti-mal and near optimal close-form solutions with RF chains doubled the number of data streams and equal to the number of data streams,by exploiting asymp-totic orthogonality of different users in a massive MIMO regimen.Although the proposed algorithm is firstly designed to use in a massive MIMO regimen,the simulation thereafter confirms the effectiveness of the proposed schemes when there are a small number of base station(BS)antennas.Furthermore,low complexity scheme for massive MIMO system based on codebook searching in the analog stage is proposed by utilizing the idea of aforementioned theoreti-cal schemes in order to avoid the esitmation of high-dimensional analog chan-nel in practice.The simulation results show that the proposed codebook-based scheme achieves similar performance as theoretical solution.At last,this thesis further investigates a more practical hybrid beamform-ing system with sub-connected subarray formation when consider the state of the art.Different from the general assumption of best receiver in the existing literature,we attempt to find the optimal hybrid precoder and combiner that maximizes sum-rate of system,which can be regarded as a joint transeiver de-sign.By following the idea of information theory based two-stage optimiza-tion criterion,we unifies the analog design of single-user and multi-user com-munications by theoretical analysis.Under the assumption of perfect CSI,we firstly propose a theoretical optimization algorithm of jointly design combined with iterative approach.Then,an upper bound of the performance of hybrid beamforming system with sub-connected subarray formation is analyzed.The simulation thereafter validates that the proposed scheme achieves similar per-formance as the proposed upper bound.Additionally,the corresponding code-book searching based solution for practical system is further derived with the help of the idea of theoretical solution.The simulation results show that the proposed codebook-based scheme not only has extremely low searching com-plexity compared to other existing codebook searching based schemes,but also has close performance gap compared to the exhaustive searching scheme,which is an upper bound of achievable sum-rate.