Research On Multi-user Joint Beamforming And Discrete Phase Optimization

In recent years,with the higher requirements for wireless communication in the information age,more wireless communication technologies need to be studied.Among them,many wireless communication technologies involve the optimization problem of joint design of beamforming and discrete set variables in the physical layer signal processing.The hybrid beamforming technology currently widely used in millimeter wave communication involves such problem.In addition,intelligent reflecting surface technology also involves such problem.Therefore,in order to be able to use some current technologies to improve the performance of wireless communications,the joint design of beamforming and discrete set needs to be studied.This thesis focuses on the two application scenarios of joint design of beamforming and discrete set,namely hybrid beamforming technology and intelligent reflecting surface technology.The specific research content is as follows:Firstly,considering a wireless communication system where base station adopts hybrid beamforming technology combined with massive MIMO,the optimization problem with the goal of maximizing the minimum user signal to interference plus noise ratio(SINR)is studied.To abstract mathematical problem,two algorithm based on total decomposition and ZF-PENALTY alternating optimization are proposed respectively in this thesis.For the analog beamforming matrix,majorization minimization(MM)and projected gradient(PG)are used to iteratively obtain an approximate solution in this thesis;For the digital beamforming matrix,an approximate solution algorithm based on zero forcing and power allocation is proposed in this thesis.The simulation results show that the two proposed algorithms are superior to the existing methods.Secondly,considering the wireless communication system with the aid of intelligent reflecting surface,the optimization problem of maximizing the sum rates is studied.To solve this problem,the method based on the framework of the block coordinate descent is proposed in this thesis.And the solution of each block is based on the weighted minimum mean square error(WMMSE)and the GEMM algorithm respectively.The simulation results prove that the performance and complexity of the proposed algorithm are better than the existing methods.