Antenna Selection And Beamforming Optimization For Reconfigurable Intelligent Surface Assisted Millimeter Wave Communication

Mm Wave communication has the advantages of large bandwidth,high spectral efficiency and high transmission rate.However,due to the high propagation loss,short wavelength,and high frequency of mm Wave,it is necessary to use a massive antenna array to compensate for the propagation loss at high frequency.The use of a massive antenna array can improve system capacity,counteract the loss during mm Wave transmission,and provide better anti-fading performance.However,as the number of transmitting antennas increases,hardware costs also increase.Therefore,antenna selection technology can reduce the use of radio frequency(RF)chains and reduce hardware circuit costs.Since there is some correlation between the massive antennas,and the use of these antennas does not increase much spectral efficiency,using a subset of optimized antennas can reduce system complexity while minimizing the loss of spectral efficiency.However,since only some antennas are active,antenna selection inevitably leads to certain performance losses.Reconfigurable intelligent surface(RIS)is a low-cost technology that can improve the spectral efficiency of wireless networks by reconfiguring the propagation environment,and can compensate for the performance loss caused by antenna selection.Therefore,this paper focuses on antenna selection and beam optimization of RIS-assisted mm Wave communication systems to improve system performance and reduce hardware costs.This thesis focuses on the joint optimization of antenna subsets,transmitter beamforming,and passive beamforming of RIS for both multiple input single output(MISO)and multiple input multiple output(MIMO)scenarios assisted by RIS,with the aim of maximizing the received signal-to-noise ratio(SNR)and reducing hardware costs.However,since the objective function and unit modulus constraints are nonconvex,the problem is also non-convex.To address this non-convex problem,this thesis proposes manifold optimization based subset selection(MOBSS)and semi-definite relaxation based subset selection(SDRBSS).Since these two algorithms require too much signal exchange overhead for channel estimation,three algorithms of lowcomplexity are proposed in this thesis.Among them,the base station and RIS independently adjust the antenna subset,transmitted beam coefficient and phase in an alternate way until convergence is reached.Simulation results show that compared with the benchmark,the proposed algorithm can significantly improve performance.Compared with the optimization schemes of MOBSS and SDRBSS,the low-complexity algorithms can greatly reduce the overhead of channel estimation and computational complexity.Compared with traditional links without RIS assistance,RIS can greatly improve the received SNR.