| In the past three decades,with the rapid growth of wireless communication traffic,5G emerges as the times require.Millimeter wave,Massive MIMO and beamforming are the three core technologies of 5G.Due to the DBF can achieve the best performance,but the hardware cost and energy consumption are high,and ABF has low hardware cost and poor performance,HBF technology can achieve the tradeoff between the ideal performance and hardware cost,but its processing in the analog domain has constant modulus constraint,so the system can not obtain all the antenna gain and has performance loss.Secondly,the traditional OMA technology in mobile communication has the problem that a resource block is monopolized by a single user.Therefore,combining with millimeter wave communication,Massive MIMO technology and beamforming technology,this paper studies the hybrid beamforming algorithms for millimeter wave Massive MIMO system,aiming at the performance loss of hybrid beamforming and the problem that a resource block in OMA technology is monopolized by a single user.Aiming at the problem of performance loss of hybrid beamforming,this paper proposes a hybrid beamforming algorithm based on partially-connected Massive MIMO system with millimeter wave to maximize the system spectral efficiency.Firstly,the optimal DBF is obtained based on singular value decomposition.Then,the matrix characteristics are used to update the analog and digital beamforming alternately.The nonconvex problem is transformed into a convex optimization subproblem,and the element phase increment in the analog domain is solved in a small range.The simulation results show that the average spectral efficiency of the proposed algorithm in the considered SNR range is 108.8%,168.1%,98.7%and 97.6%of the classical algorithm with partially-connected architecture,full analog beamforming,classical algorithm with fully-connected architecture and full digital beamforming,respectively.In other words,the performance of the algorithm is better than that of the classical algorithm with partially-connected architecture and the full analog beamforming,and is close to that of optimal full digital beamforming and that of the and classical algorithms with fully-connected architecture,but the hardware complexity and power consumption are lower.Aiming at the problem that a resource block in OMA technology is monopolized by a single user,this paper introduces NOMA technology and proposes two schemes of user grouping,power allocation and hybrid beamforming.Firstly,based on the K-means clustering algorithm,a user grouping scheme is proposed according to the channel correlation between users.On this basis,a joint optimization problem of power allocation and hybrid beamforming is proposed.Then,taking SLNR as the performance objective,the two variables of the original problem can be decoupled and solved iteratively.Then,based on the initial beamforming matrix,the power allocation problem is transformed into a convex problem by introducing an auxiliary-positive real variable,and the optimal closed-form solution of power allocation is obtained based on KKT condition and Lagrange multiplier method.Finally,the optimal closed-form solution of optimal full digital beamforming can be obtained based on GED.Based on this matrix,two hybrid beamforming algorithms can be designed according to the different optimization methods in analog domain and digital domain.The simulation results show that the average spectral efficiency and average energy efficiency of the classic NOMA scheme in the considered SNR range are 24.8%and 24.7%of the proposed scheme 2,and 32.8%and 32.7%of the proposed scheme 1,respectively.The spectral efficiency and energy efficiency of the classical OMA scheme are 0.45%and 1.17%of the proposed scheme 2,and 0.61%and 1.55%of the proposed scheme 1,respectively.In other words,the spectrum efficiency and energy efficiency of the proposed joint schemes of user grouping,power allocation and hybrid beamforming are better than those of the classic NOMA schemes and OMA schemes. |
PDF Full Text Request