Research On Beamforming Algorithms Based User Grouping And Antenna Selection Of Massive MIMO Systems

Large-scale MIMO systems,equiped with tens up to hundreds of antenna at base station,can support multiple users to access at the same time,and provide multi-user diversity gain and spatial multiplexing gain to improve the transmission reliability and efficiency.However,the number of users served simultanously to achieve the maximum channel capacity is conditioned by the number of antennas at BS.Infinitely increasing the number of antennas at BS will lead to more radio frequency chains,which adds cost of the system implementation,so the research on user grouping and antenna selection in massive MIMO system has important significance.This thesis first respectively introduces the basic concepts of massive MIMO and beamforming,system model,and conventional beamforming technologies.Then we analyze the reseach status of user grouping and antenna selection.Comprehensively considering the hardware complexity and system capacity,a beamforming algorithm based on user grouping and antenna selection for massive MIMO was proposed.In order to reduce the inverse calculation of the beamforming matrix in the iterative process,we delete the minimun effective channel gain to realize the user grouping by usering the concept of orthogonal sub-space orthogonal projection and effective channel gain,and then considered antenna selection from two aspects.One is configuring the limited radio-frequency chains to the antennas with relatively better performance selecting from the large scale antennas.Another one is allocating each of the transmitted date signals on all the antennas and selecting an optimal antenna subset for each data,which named k-regularity antenna selection.The zero-forcing beamforming algorithm based on user grouping and antenna selection will greatly reduces the hardware complexity by reducing the cost and power consumption of radio-frequency chains with a small performance loss,and achieve system performance and hardware complexity of the compromise.In addition,for the high computational complexity problem of matrix inversion in the zero-forcing beamforming based on user grouping and antenna selection,we firstly introduce the simplified method based on Neumann series expansion,truncated polynomial expansion and symmetric successive over-relaxation(SSOR),and then analyze their performance in term of the total system rate.Finally,in the case of extending the raw variables,the feasibility of the simplified algorithm for matrix inversion based on SSOR in massive MIMO is verified.The theoretical analysis and simulation results show that the joint user grouping and antenna selection algorithm proposed in this thesis can reduce the system hardware complexity at a relative small cost of channel capacity with low computational complexity.They may serve as highly feasible and salient candidate schemes to implement in practice.