Beamforming For Reconfigurable Intelligent Surface Assisted Millimeter Wave Communication System

Considering the high energy consumption,costly hardware and complex signal processing needed for traditional millimeter wave(mm Wave)MIMO systems,the reconfigurable intelligent surface(RIS)is con-sidered to be an emerging technology to improve the efficiency of wireless communication networks in the future.Beamforming for RIS-assisted mm Wave MIMO system is studied in this thesis,and the detailed work is concluded as follows.The beamforming for RIS-assisted mm Wave MIMO system is first studied both in single-user and multi-user scenarios when the air port channel matrix of the system is already obtained.In the single-user scenario,semidefinite relaxation and alternating minimization are used respectively to jointly design the beamforming at the base station(BS)and the reflection coefficients of the RIS,which minimize the transmit power of BS under the premise of ensuring that the received signal to noise ratio of user is greater than a certain threshold.Different from the single-user scenario,the inter-user interference needs to be considered in the multi-user scenario.Alternating minimization is used in multi-user scenario to jointly design the beamforming at the BS and the reflection coefficients of the RIS,which minimize the transmit power of BS under the premise of ensuring that the received signal to interference plus noise ratio of all users are greater than a certain threshold.Simulation results show that the proposed schemes can effectively reduce the transmit power of the BS compared to the mm Wave MIMO system without RIS,or the RIS-assisted mm Wave MIMO system which randomly generates the reflection coefficients.Considering that obtaining the channel matrix is challenging in practical RIS-assisted mm Wave MIMO system,we optimize the beamforming at the BS and reflection coefficients at the RISs through beam training when the channel matrix is unknown,and propose the multicast training scheme and the hierarchical multicast training scheme in this thesis.The multicast training divides the beam training process into two stages,and the beam scanning of the BS and all RISs are performed simultaneously at the first stage.Specifically,when the BS performs beam scanning,it separates a part of the energy to form the beams pointing at each RIS,and the beam scanning of each RIS is performed by changing the phase coefficients.Meanwhile,the user performs beam measurement and selects the best beam scanning result.Considering that the beam scanning at the BS and each RIS are superimposed in the first stage,in the second stage it is necessary to perform link-by-link training based on the best beam scanning result obtained in the first stage,select the optimal communication link from the base station to the user and concentrate all the energy of the base station to form a beam pointing at that link.Based on the multicast training scheme,the hierarchical multicast training introduces a hierarchical codebook design,which significantly reduces the training overhead in the first stage of the multicast training.Simulation results verify that compared with the beam scanning scheme,the proposed two beam training schemes can significantly reduce the training overhead while ensuring certain performance.