Hybrid Beamforming Design In Millimeter Wave MIMO System

The millimeter wave(mmWave)technology is the frontier for commercial high volume consumer wireless communication systems,and is the core tech-nology of current high-speed WLAN and 5G cellular systems.The main ben-efit of going to mmWave carrier frequencies is the huge bandwidth channels.However,due to the propagation characteristics of the high path loss and block-age of mmWave,as well as the hardware complexity and power cost of high frequency device design,the development of mmWave commercialization has been restricted for a long time.With the research and development of mmWave technology in academia and industry,mmWave MIMO system using hybrid digital/analog beamforming technology,which equipped with large-scale an-tennas and reduced number of RF chains,make mmWave commercial possible.Since the array gain of large-scale antennas can be fully utilized to compensate for poor link budgets in the analog domain,and the sparse characteristics of the mmWave channel can be fully utilized to achieve efficient data transmission in the digital domain.This thesis provides novel solutions for hybrid beamforming design in mmWave MIMO systems.On the one hand,based on the theory of mmWave radio propagation and antenna arrays,combined with advanced signal process-ing techniques,including massive MIMO,interference alignment(IA)as well as non-orthogonal multiple access(NOMA),etc.,constructive baseband pro-cessing algorithms are proposed.On the other hand,based on the commercial standards of IEEE 802.15.3c,IEEE 802.11ad and 3GPP 5G NR,consolidated with RF characteristics and baseband signal processing technology,correspond-ing practical and feasible designs are proposed.The main contributions of the thesis are as follows:Firstly,ideal hybrid beamforming designs in mm Wave MIMO systems are studied.For the single-user MIMO system,a novel blind hybrid beamforming design without CSI is proposed.Specifically,a stochastic gradient descent al-gorithm with the maximum received power as the measure is used to design the analog beamforming matrix,and a power iterative algorithm implemented on the equivalent baseband channel is used to obtain digital beamforming matrix.For the multi-user MIMO system,in order to avoid the synchronization error and phase noise caused by variable rate coding and multi-stream data trans-mission,a hybrid BD-GMD design using the same modulation coding scheme(MCS)is proposed.Starting from the BER optimal design,the influence of the analog design on the channel freedom is considered,as well as the digital do-main interference between users and the inter-stream interference processing in the user.The closed-form solution of hybrid BD-GMD is derived.In addition,a low complexity limited feedback design is also given.Secondly,antenna array codebook design and beam training in mmWave MIMO systems are studied.For the antenna array codebook design,a low-complexity two-level 3D codebook design is proposed,in which the main beam codebook guarantees the maximum beam gain by controlling the antenna phase shift,and the auxiliary beam codebook ensures the beam precision by control-ling the rotating factor.For the single beam training problem,a low complexity algorithm based on antenna selection to change the beamwidth is proposed.The selection of the multilevel beam codebook is realized by the correlation of array factors.For the multi-beam training problem,based on the antenna array and random matrix theory,an effective beam subset optimization method is de-rived,and it is proved that the minimum size of the optimized beam subset is the channel propagation path number.On this basis,according to the number of TIXRUs that are limited to participate in beam training,two methods of linear beam training are given,and the proof of convergence is given.Finally,interference suppression based on hybrid beamforming in mmWave MIMO systems are studied.A 60 GHz dense communication net-work and a 5G mmWave cellular network are considered typical scenarios.Combined with the interference problems in the actual system,a hybrid beam-forming design based on IA and NOMA is proposed,respectively.For the 60 GHz network,to deal with the global interference problem with low SNR,the feasible conditions of the hybrid design based on IA are analyzed,that is,the number of RF chains of each node is greater than or equal to twice of data streams.In addition,a joint max-SINK and min-MSE iterative algorithm based on channel reciprocity is proposed.Furthermore,a detailed analysis of the pro-posed algorithm for large scale system is also provided.For the 5G mmWave cellular network,considering the waste of resources caused by the tradition:al OMA processing multi-user interference,the NOMA-MIMO technology can effectively utilize the high path loss characteristics of the mmWave propagation.To deal with intra-cluster interference and inter-cluster interference in NOMA systems,detailed solutions are given from three perspectives:user pairing,hy-brid beamforming design and power allocation.