Optimal Design Of Reconfigurable Intelligent Surfaces Enhanced Multi-User Communication Systems

With the development of mobile internet and its wide and deep application in our society,consumers and business users continue to create demands for higher data trans-mission rates.From the beginning of modern wireless communication to the current 5G era,the environment in which electromagnetic waves propagate has been considered as a random and uncontrollable entity,and the shadow fading and multi-path effect greatly af-fect the efficiency of mobile communication networks and the quality of service for users.However,the advent of reconfigurable intelligent surface(RIS),which consists of massive low-cost passive reflecting elements,makes wireless propagation being controllable be-comes true.The reflecting elements on RIS can independently modify the amplitude and phase of impinging signals,and collaboratively shape specific three-dimensional beam-forming to enhance signal power or cancel interference.Due to this capability,RIS can significantly improve the performance of wireless communication networks,by intelli-gently reconfiguring wireless propagation,and it has been considered as a revolutionary technology in beyond 5G and 6G.In this thesis,two problems related to the optimization of RIS-enhanced multi-user communication systems are researched,i.e.,the deployment optimization of RISs and joint optimization of RIS-assisted cell-free communication net-works.Because of the passive reflecting property of RISs and the "double fading" effect which signals suffer when they are transmitted from transmitters to receivers via RISs,the coverage of RISs is constrained compared with active devices,and the location of RISs will greatly affect their performance.Thus in the work of the deployment optimization of RISs,a RIS-assisted single-cell multi-users communication system is considered,and the optimization objective is to maximize the system's ergodic capacity,which is based on the statistical information of the channels.An adaptive cyclic coordinate descent algo-rithm is proposed to tackle the optimization problem,and simulation results show that the optimized deployment of RISs can effectively improve the system's ergodic capacity.As a kind of massive MIMO technology,cell-free network is considered valuable due to its ability to manage interference and provide uniformly high data rates everywhere.However,the hardware cost will significantly increase as the network becomes large.So,in this thesis,low-cost RISs are used to replace some of the base stations in the cell-free network and assist the others to reduce the hardware cost.Besides this,an effective method is needed to solve the optimization problems related to the beamforming vectors of base stations and the reflecting coefficients of RISs.In the work of the joint optimization of RIS-assisted cell-free communication networks,the objective is to maximize the sum rate of users through jointly design the beamforming of base stations and the reflecting coefficient matrices of RISs.According to this problem,a deep reinforcement learning method called twin delayed deep deterministic policy gradient algorithm(TD3)is adopted to design the beamforming vectors and reflecting coefficient matrices,whose input state is defined as the global channel state information.Simulation results show that,compared with heuristic algorithm,TD3 algorithm has a huge advantage to achieve higher sum rates and reveals good robustness.