Research On Hybrid Precoding Technology Based On Millimeter Wave Massive MIMO System

Millimeter wave band can provide rich spectrum resources for communication systems,and largescale Multiple Input Multiple Output(MIMO)technology can provide higher spectral efficiency and energy efficiency for the system,and the effective combination of both can be found in the technology for the Fifth Generation(5G).As a key signal processing technique in millimeter-wave large-scale MIMO systems,hybrid precoding replaces the traditional all-digital processing unit with a combination of analog and digital signal processing units,which can effectively reduce the system hardware cost and energy loss while ensuring spectral efficiency.In this thesis,we focus on the hybrid precoding technique for millimeter-wave large-scale MIMO systems,and propose corresponding hybrid precoding methods based on different architectures.Also,the important performance metrics of the system,such as spectral efficiency,hardware cost,computational complexity and energy consumption,are analyzed and compared.The main works are as follows:(1)A hybrid precoding algorithm based on gradient descent is proposed for a fully connected millimeter-wave large-scale MIMO system.At the transmitter side,the hybrid precoding algorithm is designed by the gradient descent method with the objective function of minimizing the expectation matrix and the Euclidean distance of the requested precoding matrix.In which,the expectation matrix is obtained by performing Singular Value Decomposition(SVD)on the channel matrix.Subsequently,the digital precoding matrix is designed using the least squares method.Finally,the receiver-side combiner is designed using the symmetry of both the transceiver and the transmitter.The difference is that the optimal combiner matrix is obtained by the Minimum Mean Squared Error(MMSE).The simulation analysis shows that the algorithm in this thesis has higher spectral efficiency and lower algorithm complexity compared with existing algorithms.(2)Two hybrid precoding algorithms based on Successive Interference Cancelation(SIC)are proposed for partially connected millimeter wave massive MIMO systems.The system architecture of fully connected at the transmitter side and partially connected at the receiver side is adopted considering the spectrum efficiency and energy efficiency trade-offs.The first algorithm adopts the same design idea as the fully connected structure,except that the expectation matrix is designed by the SIC algorithm and the joint analog and digital optimization is used in the design process.Simulation analysis shows that the algorithm has poor spectral efficiency when the number of antennas at both transceiver ends is too large.In order to improve the spectral efficiency of the system with a large number of antennas,we propose a second algorithm: the analog combiner matrix is designed using the SIC algorithm and the digital combiner matrix is designed by MMSE.The simulation analysis shows that the spectral efficiency is effectively improved in the system with any number of antennas,and the proposed algorithm has the best performance in the comparative analysis.In addition,the energy efficiency of the improved system is further improved,and the performance is optimal.