The Research Of Hybrid Beamforming Technology Based On Digital-analog Hybrid Architecture

In recent years,more and more attention from academia and industry has been paid to the research on large-scale antenna system(LSAS)and millimeter wave(mmWave)communication.LSAS could obtain the beamforming gain from antenna array,significantly improving spectral efficiency.Further,there are a large number of unallocated frequency band resources in millimeter wave band,which can greatly expand the bandwidth of system.Therefore,LSAS and mmWave communication are highly possibly becoming important technologies in 5G systems to solve communication congestion.With full digital beamforming LSAS,which could perform beamforming optimally,the antenna number of the system must be the same as the number of radio frequency(RF)chains to ensure that each antenna element has a dedicated RF chain and baseband processing unit,leading to high system complexity,high power consumption and high hardware costs.Therefore,in practical implements,digital-analog hybrid beamforming(HBF)structure would be a good choice for a trade-off between performance and cost.With the HBF architecture,only a limited number of RF chains are used,and the system complexity and hardware cost are reduced while the performance loss is acceptable,which provides a trade-off between system performance and hardware complexity plus cost.In this thesis,firstly,we present the basic model and the response vector of antenna array,and then introduce the channel model commonly used in mmWave communication.In order to further state the principle and significance of beamforming,we analyze conventional analog beamforming(ABF)algorithms based on beam steering,beam nulling and minimum variance distortionless response(MVDR),and digital beamforming(DBF)algorithms based on zero-forcing(ZF),minimum mean square error(MMSE),singular value decomposition(SVD),block diagonalization(BD)and codebook.Secondly,several HBF algorithms for downlink transmission,i.e.beamspace MIMO,alternating HBF methods,phased-ZF HBF algorithm and two-stage HBF scheme based on codebook feedback,are studied.The performance of HBF algorithms is simulated and compared to analyze the existing problems and limitations of these conventional schemes.With the HBF architecture,we proposes an improved HBF scheme based on signal to leakage and noise ratio(SLNR)criterion.Besides,due to the existing problem in FDD and TDD systems that perfect channel state information(CSI)cannot be obtained under HBF architecture,a HBF algorithm based on uplink training is proposed in this thesis,using the channel reciprocity existing in TDD systems.By uplink training,this algorithm could obtain the optimal joint beamforming matrix.Then we decomposed the optimal joint beamforming matrix into analog beamforming matrix and digital beamforming matrix.Based on designed analog beamforming algorithm,the best codebook vectors would be found in the codebook set,forming analog beamforming matrix.And digital beamforming matrix can be obtained directly in the form of least-square solution to complete the downlink hybrid beamforming algorithm design.