Study On Hybrid Beamforming Algorithm In Large Scale MIMO

Mobile Internet has developed rapidly in the past decade.With the emergence of high-traffic applications such as ultra-high-definition video and emerging business scenarios such as industrial automation,unmanned driving,and smart home,the demand for mobile data has exploded sharply.The development of the fifth-generation mobile communication systems is progressing rapidly.As of September 2020,a total of 690,000 5G base stations had been built and opened,which is more than 75% of the total number of global(5G white paper).With the commercialization of 5G,people are also beginning to explore the 6G mobile communication systems.6G will be a whole connected world which includes land,sea and air communications.It will not only be fast,but will also achieve global coverage and interconnection of everything.As one of the key technologies of 5g and 6G,hybrid beamforming technology has been widely concerned and studied.In this thesis,the hybrid beamforming technology in large-scale MIMO system is studied.First,the conception and application background of millimeter wave communication systems,massive MIMO technology,beamforming technology,and the theoretical knowledge involved in this thesis,such as millimeter wave channel model,MIMO system model,and traditional digital beamforming algorithm are introduced in this thesis.At the same time,the structure and model of the hybrid beamforming system are also summarized.Then,the hybrid beamforming algorithms for point-to-point single user massive MIMO system have been studied,which include the orthogonal matching pursuit algorithm(SSP)based on sparse space,the orthogonal matching pursuit algorithm based on codebook(OMP),the alternating iteration minimization algorithm based on phase extraction(PE-Altmin),and the HBF algorithm based on iterative matrix decomposition(IMD).Combined with the improved MSE criterion,the OMP algorithm based on MSE(MSEOMP)is proposed in this thesis.Finally,these algorithms are simulated and analyzed.After that,the hybrid beamforming algorithm for multi-user massive MIMO systems have been studied,and the HBF algorithm based on the signal-to-leakage-plus-noise ratio(SLNR)and the improved HBF algorithm based on the block diagonalization of AOD(Projected BD HBF)as well as the improved block diagonalization algorithm based on null-space projection(Projected BD HBF)were proposed in this thesis.Among them,the SLNR algorithm is better than the traditional fulldigital block diagonal algorithm in multi-stream transmission,and the SLNR algorithm has no rank constraint on the joint channel,so it can be used as an alternative for the block diagonalization algorithm when the rank constraint condition is not satisfied.In addition,the improved Projected BD algorithm proposed in this thesis uses QR decomposition to obtain the characteristic direction of the channel,and uses an iterative method to project the precoding matrix of the subsequent user onto the null space of the previous user channel,which has lower complexity compared with the SP-AOD-BD algorithm;according to the proposed improved Projected BD algorithm,an improved Projected BD HBF algorithm is further designed.The improved Projected BD HBF algorithm ignores the error IMD algorithm,and the equivalent channel matrix is processed by the maximum combining ratio to obtain the unconstrained combining matrix,and then the IMD algorithm is used to obtain the digital-analog combining matrix.It is demonstrated that the performance of these two improved algorithms is better than that of the full-digital block diagonalization in subsequent simulations.At the same time,the phase-based zero-forcing algorithm(PZF)which is suitable for single-antenna users,the HBF algorithm based on block diagonalization and the HBF algorithm based on SVD decomposition are also simulated and analyzed in this thesis.Finally,several sub-optimization scheduling algorithms for full-digital beamforming systems are introduced when the channel rank constraints are not satisfied: sub-optimization scheduling algorithms based on the system sum rate and the chord-distance as well as the angle between subspace.After that,this thesis apply them in the hybrid beamforming system.Simultaneously,these algorithms are simulated and compared with random scheduling algorithm.Accroding simuation,it is found that these sub-optimized scheduling algorithms have better performance than random scheduling algorithm;the performance of sub-optimized scheduling algorithms based on angle between subspace is between the sub-optimized scheduling algorithms based on system sum rate and chord-distance;the sub-optimized scheduling algorithm based on chord-distance has the lowest complexity among them,so we can choose the best among these three algorithms according to the actual communication conditions,calculation cost and performance requirements.