Research On Digital/Analog Hybrid Beamforming Optimization Algorithm In Millimeter-wave Massive Mimo System

With the development of wireless communication,millimeter wave massive multiple input multiple output(Massive MIMO)technology has become one of the current research hotspots,which can significantly improve system performance and capacity.In traditional MIMO systems,the number of antennas is small,and full digital beamforming is usually used.However,in large-scale MIMO systems,due to the large number of antennas,the number of corresponding RF links is also huge.Hybrid beamforming techniques are used to reduce the RF chain.The number of ways provides a new idea for solving this problem.The use of hybrid beamforming optimization algorithms is conducive to reducing computational complexity and energy consumption,and can also effectively improve the spectral efficiency of massive MIMO antenna arrays.Before the system is manufactured,the design of the hybrid beamforming scheme can ensure the optimization of the system performance at the link level.In the context of the project "Optimization of Resource Allocation for In-band Wireless Access and Backhaul Based on Millimeter Wave Massive MIMO in 5G Ultra-Dense Heterogeneous Networks",this paper designs a hybrid beamforming optimization algorithm for massive MIMO,and the computational complexity and spectral efficiency were simulated and evaluated.The specific work is divided into the following three parts:The hybrid beamforming optimization algorithm in single-user massive MIMO system is studied,and the problems existing in the existing algorithms are analyzed.A hybrid beamforming algorithm based on orthogonal matching tracking(OMP)and manifold(MO)optimization is proposed.improve proposals.Based on the OMP hybrid beamforming optimization algorithm,there is a problem of high computational complexity in the process of iteratively solving the minimum residual of the optimal beamformer and the digital and analog beamformers.This paper uses the orthogonality of the digital beamformer to perform the algorithm.Improvement: Extracting the phase of the analog beamformer from the phases of the equivalent beamformer determined by the digital beamformer and the unconstrained optimal beamformer can simplify the design of the analog beamformer.Compared with the existing OMP hybrid beamforming algorithm,the improved algorithm greatly reduces the algorithm complexity.Aiming at the manifold optimization algorithm in the existing single-user massive MIMO system,and aiming at the local best omission problem of the linear search in the design process of the analog beamformer,an improved linear optimization algorithm based on the Riemannian manifold was proposed.The improved optimization algorithm uses the Wolfe-Powell criterion for linear optimization.Compared with the original Armijo-Goldstein criterion,it effectively solves the problem that the original method excludes the minimum point out of the effective range,and improves the system's bit error performance.Simulation results show that the error rate and spectral efficiency of the proposed hybrid beamforming algorithm based on improved Riemann manifold optimization are closer to the all-digital beamforming algorithm,and the computational complexity is also lower than the existing manifold optimization algorithms.The hybrid beamforming optimization algorithm in a multi-user massive MIMO system is studied.Aiming at the problems of existing noise enhancement error and high computational complexity of the existing Breuden-Fletcher-Goldfarb-Shannon(BFGS)algorithm,an improved L-BFGS algorithm was proposed.Compared with the existing algorithm,this method changes the iterative method,reduces the storage space and computational complexity;it can effectively eliminate the problem of noise enhancement error in the original algorithm,and improve the spectral efficiency of the system.Simulation results show that the performance of the proposed hybrid beamforming system based on the L-BFGS algorithm is closer to the all-digital beamforming design,and the computational complexity is also significantly lower than the existing BFGS algorithm.