Research On Transceiver Design In Massive MIMO System

In order to meet the next generation mobile communication system higher transmission rate and spectrum efficiency demand, Massive MIMO, as one of5G key technologies, has drawn extensive attention. Massive MIMO deploying much more antenna in transceiver compared with traditional MIMO, can serve more users in the same time-frequency resource. More antennas bring higher spatial multiplexing gain, spectrum efficiency and system robustness.This thesis covers Tx-Rx technologies of Multi-user Massive MIMO scenarios. Conventional MIMO usually adopts TDD mode and exploit channel reciprocity to lower CSI cost; but FDD mode less studied, the main reason is that with the increase of number of transmitter antennas, FDD mode needs much more CSI and consumes large amounts of system resources. Based on the above problems, this thesis focus on Multi-user MassiveMIMO) system in FDD mode, transceiver technologies and spatial correlation are studied, this thesis proposes a Massive MIMO downlink transmission scheme based on spatial correlation.The key ideas of the downlink transmission scheme proposed in this thesis is are to partition users into multiple groups with the approximately same channel spatial correlation and to scheduling users and multi-user precoding whin different user groups.The scheme divides transmission processing into two stages based on this group:The first stage is pre-processing designed by the channel’s second-order statistics information; the second stage is MU-MIMO pre-coding designed by the effective channel matrix which is given by the result from pre-processing stage. By using the pre-processing matrix before conventional MU-MIMO pre-coding matrix, the dimension of the effective channel matrix can be reduced, and the CSI needed in transmission can also be reduced significantly. In addition, with the number of antennas limited, we need to schedule users within different user groups, this thesis proposes a user selection scheme combining random beamforming and semi-orthogonal user scheduling. Finally, simulation results of proposed scheme performance analysis given with Matlab software.The results verify that the scheme proposed in this thesis reduce the amount of feedback information required greatly with small system capacity loss.