Study On Hybrid Beamforming And Multiport Network Related Technologies In Massive MIMO Systems

Massive MIMO wireless communication technology,which employs a large number of antennas at base stations is one of essential technologies for the fifth generation(5G)mobile communications.This paper focuses on some signal processing methods related to antenna arrays to improve system efficiency of Massive MIMO communication systems.Aiming at the near-field effects,with the multiport network technology,this paper studies the joint design of decoupling and matching networks(DMNs)for massive MIMO systems under near-field conditions.Beamforming technology which can improve the system spectrum efficiency is also an important signal processing technology in wireless communications.Due to the increasing number of antennas in Massive MIMO systems,it is difficult to implement full digital beamforming scheme in practice.In order to solve this problem,the hybrid beamforming technology applied in the massive MIMO system is studied detailedly both in SU-MIMO and MU-MIMO systems.Firstly,Massive MIMO technology,multiport network technology,and beamforming technology are studied,including MIMO system capacity,multiport network models,digital beamforming algorithms and hybrid beamforming system solutions,as well as narrowband clustered channel model for millimeter wave communication.Secondly,the design technology of decoupling and matching network for Massive MIMO system based on multi-port network is investigated.In view of the near-field effects of the massive MIMO system,the expressions of the MIMO system model based on the impedance matrix(Z matrix)and the scattering matrix(S matrix)are studied and provided by multiport network analysis technology.And then the decoupling and matching networks at the transmitter and receiver are jointly designed by the iterative algorithm of DMNs.The equivalent channel matrix H of the whole system is derived,based on which,the system performance is analyzed.Simulation results show that the designed DMNs considering the near-field effect can reduce the correlation between antenna elements,improve channel conditions,increase transmission efficiency,and enhance channel capacity.Thirdly,the hybrid beamforming technology of single-user millimeter-wave Massive MIMO system is studied,and the hybrid beamforming design is decoupled into matrix design problems at the transmitter and receiver separately.Hybrid precoding matrix design methods is provided based on orthogonal matching pursuit(OMP)algorithm and alternating minimum iteration(Alt Min)algorithm respectively.Based on OMP algorithm,two improvement schemes are studied and proposed in this thesis.First,The RSVs-based OMP algorithm(RSVs-OMP)is proposed which adopts the right singular value vector(RSVs)codebook based on the channel matrix instead of antenna array response vectors codebook which has no relation with the channel.Then,based on the RSVs codebook,RSVs-based parallel precoding algorithm(RSVs-PS)and direct selection algorithm(RSVs-DS)are also studied and proposed.The RSVs-PS and RSVs-DS algorithms remove the iterative operations and then reduce complexity of algorithms.Second,an improved SVD phase extraction iterative algorithm(SVD-PEI)based on the optimal precoding matrix and residual matrix is proposed,where no candidate codebook is required.Simulation results show that,compared with the OMP algorithm,the RSVs-OMP algorithm can obtain higher spectral efficiency,and the system performance obtained by the RSVs-PS algorithm and the RSVs-DS algorithm is very close to the RSVs-OMP algorithm.With a slightly higher algorithm complexity,the SVD-PEI algorithm can obtain the relatively closer spectral efficiency to the performance of optimal precoding method.All the proposed improved algorithms have stronger robustness compared with the OMP algorithm.Finally,the hybrid beamforming technology of the multi-user millimeter-wave Massive MIMO system is also studied.Multi-user precoding algorithms based on block diagonalization algorithm(BD)are provided,including BD-OMP,BD-Alt Min,BD-RSVs-OMP,BD-RSVs-PS and BD-RSVs-DS algorithms.And then the improved equal-gain transmission(EGT)-block diagonalization(BD)two-stage algorithm based on discrete Fourier transform orthogonal basis(DFT)is put forward.Simulation results show that BD-based multi-user precoding algorithms can obtain better system spectral efficiency.Compared with the BD-OMP algorithm,BD-RSVs-PS,BD-RSVs-DS can obtain good system performance while with much lower complexity.The improved DFT-based EGT-BD algorithm can keep the good system performance even in the scenario with more users.Compared with the BD algorithm above,the improved DFT-based EGT-BD algorithm is insensitive to the user number and is more suitable for MIMO systems with more users.