Massive multiple-input-multiple-output(MIMO) technology can improve system performance without extra bandwidth and power consumption. It can serve more users on the same frequency resources and possess lower power consumption compared to the small-scale MIMO. However, these performance improvements increase system hardware costs. For large-scale MIMO systems, antenna selection is an efficient ap-proach to overcome this problem. Therefore, this paper focuses on antenna selection technique for massive MIMO systems.Firstly, the thesis analyzes the conventional antenna selection techniques. Dis-cuss the influences of different selected antennas and fading channel with different power allocations.Then, after calculating the complexity of different selection technolo-gies, we come to the conclusion that the norm based method is more suitable for single user Massive MIMO systems.This is based on the trade-off of complexity and capacity performance.Analyze the k-regular beamforming scheme for massive MIMO to over-come the rate loss of the simple antenna selection method,while keeping the hardware complexity far less than that required full optimal eigen-beamforming. Finally, this thesis studies antenna selection for massive multiuser MIMO system.However, the techniques used in single-user communication cannot directly be applied because of multiuser interference. So combine precoding and selection techniques could be a good solution. We compare two linear precoding techniques matched filter-ing(MF) and zero-forcing(ZF) from different SNR, selected antennas, transmitting an-tennas and number of users perspective.In particular, analyze the optimal numbers of users for ZF according to the characteristics of the performance curve. Mean-while, focus on the cell-boundary users, from the simulation results we can draw a con-clusion that MF precoding is more suitable for cell-boundary users when the number of active users is larger than the intersection. Antenna selection technique result in inter-section shifting to the right.For precoding technique,discuss MF and ZF from normali-zation perspective.Through analytical and numerical results,we confirm that matrix mormalization is better for MF while vector normalization is better for ZF. |