Research On Hybrid Precoding Technology In Millimeter Wave Large-Scale MIMO Systems

As the candidate core technologies of 5^th Generation?5G?mobile communication systems,both millimeter wave?mmwave?communications and large-scale multiple input multiple output?MIMO?technology can enlarge their advantages to further improve the system capacity.For mmwave communication systems,large-scale MIMO technology provides array gain that can effectively compensate for the loss of millimeter-wave signal in free space.At the same time,joint precoding technology can further improve the spectral efficiency of mmwave systems.However,Precoding are done digitally at baseband in traditional MIMO systems,and requires dedicated baseband and Radio Frequency?RF?hardware for each antenna element,which will improve the cost and power consuming of mmwave large-scale MIMO systems with more than hundreds of antennas.In order to solve those problems,hybrid precoding technology has been proposed and applied to mmwave systems,which divides the precoding process into digital precoding and analog RF precoding.However,the near-optimal hybrid precoding is still facing many problem due to analog RF precoding has constant modulus constraints.Therefore,the thesis makes an in-depth study on hybrid precoding technology in mmwave large-scale MIMO.The main contributions are listed as follows.1.Hybrid precoding scheme design based RF vector updates.Firstly,an initial RF precoding matrix was constructed based the singular value decomposition.Then,the first column in the RF precoding matrix is removed and the digital precoding matrix at this time is obtained,and an updated RF matrix was constructed by placing the vector which was constructed by the maximum left singular vectors of the residual matrix onto the last column of the initial RF matrix.By multiple iterations,the resultant RF precoding matrix can be obtained and the digital precoding matrix was designed based on the least square criterion.Theoretical analysis and simulation results show that the proposed scheme greatly outperforms the hybrid precoding method based on orthogonal matching pursuit?OMP?scheme with much reduced computational complexity.2.Hybrid precoding scheme design based on two-stage matrix factorization.The hybrid precoding is divided into two stages.Firstly,two matrices are found that can reconstruct the optimal digital precoding by gradient descent theory during the stage of relaxing the constraint of hybrid precoding scheme.Secondly,considering the constant modulus,an auxiliary unitary matrix is used to design the RF precoding matrix.The unitary matrix and RF matrix are designed,which are the local optimal solutions of the objective function by the alternating minimization.Finally the digital precoding matrix is constructed by combining the first stage.Theoretical analysis and simulation results show that the proposed scheme can design the near-optimal hybrid precoding matrix,and has better performance compared with existing hybrid precoding scheme.