Research On Multi-User MIMO Beam Management Technology

The 5th Generation Mobile Communication System(5G)applies the millimeter wave wireless communication technology,and adopts the beamforming technology based on the multi-antenna array to realize signal transmission and reception in the base station and terminal.The directional beams at both transmitting and receiving ends need to match the user and beam to improve the system performance.Firstly,beam alignment between transmitter and receiver is the key problem to be solved in the whole communication process,and beam training in the initial access stage is the basis of directional communication.In addition,the problem of user beam tracking in mobile scenarios,and the design of beam selection strategy for Multi-User Multiple Input Multiple Output(MU-MIMO)system in which multiple beams work simultaneously,are also key challenges for multi-user millimeter wave MIMO wireless communication.Therefore,this paper focuses on a series of beam management problems in millimeter-wave largescale Multiple Input Multiple Output(MIMO)systems,including the following aspects:Firstly,aiming at the beam alignment problem required for directional beams,the beam scanning process simulation based on Synchronization Signal Block(SSB)in the initial access stage is completed,and the result is the best beam pair number selected,which verifies the effectiveness of exhaustive search beam scanning by repeatedly sending SSB.Then,in order to reduce the overhead of beam selection,a neural networkbased beam selection is designed.The Top-K accuracy simulation results show that when K=119,the accuracy has exceeded 90%,which verifies the effectiveness of the beam selection task using the trained neural network,and the average Reference Signal Received Power(RSRP)simulation results show that the average RSRP results generated by the neural network are close to the optimal exhaustive search performance.Then,aiming at the beam tracking problem in the moving scene,this paper proposes a beam tracking algorithm based on beam training,including two low-complexity local search algorithms.Compared with the beam training overhead of the three algorithms,it can be seen that the local search algorithm has fewer alternative beams and beam measurements,which can greatly reduce the overhead of beam tracking.The simulation results under the condition of no occlusion show that compared with the benchmark exhaustive search algorithm,the average spectrum efficiency performance loss generated by the local search 1 algorithm is about 0.38 bit/s/Hz,and the average performance loss generated by the local search 2 algorithm is about 0.16 bit/s/Hz,that is,the performance of the local search is very close to that of the exhaustive beam search.Finally,in view of the research needs of multi-beam simultaneous working scenarios,this paper designs the multi-user multi-beam selection strategy in MU-MIMO system.In addition,a multi-user grouping algorithm is proposed,and the average power allocation scheme and priority allocation scheme are designed to maximize the system achievable rate as the optimization goal.The simulation results show that the system achievable rate varies with the threshold of single-link Signal to Interference plus Noise Ratio(SINR),which verifies the validity of the multi-beam selection strategy.The beam-space MUMIMO system shows superior rate performance by comparing performance with other systems.