Research On Beamforming Of Wireless Communcation System Based On Reconfigurable Intelligent Surface

With the growth of communication demand,the problem of spectrum resource scarcity is becoming increasingly serious.The millimeter wave frequency band has larger bandwidth and higher transmission rate,but it suffers from significant attenuation during propagation,which limits the distance and coverage range of signal transmission.Although the integration of massive antenna technology can enhance the gain of high-frequency signals,it also leads to increased hardware costs and energy consumption.Reconfigurable intelligent surface(RIS),consisting of a large array of passive reflecting units,can dynamically adjust electromagnetic field parameters to change the wireless signal transmission environment,expand communication coverage,and reduce hardware costs and energy consumption,making it a highly promising key technology in future wireless communications.Due to involving non-convex optimization,the passive beamforming design problem is recognized as a challenge in the field of RIS-assisted communications.Based on the above background,this thesis studies the beamforming design of the RIS in continuous and discrete phases,and the specific work is as follows:1.For the challenge of continuous phase beamforming design,we propose a double RIS-based hybrid beamforming architecture for mutiuser multi-input single-output(MU-MISO)millimeter wave(mmWave)communication system,where one RIS is used to achieve the analog beamforming at the transmitter,while the other is used to aid MU mm Wave communication system near the user cluster.Firstly,in order to overcome the complexity of joint optimization of three variables,the RIS at the transmitter is pre-designed based on angle of arrival(AOA)and angle of departure(DOA)information.Then,by jointly designing active beamforming and passive beamforming matrices,the weighted sum-rate(WSR)is maximized.This non-convex optimization problem is solved through an alternating optimization framework.The simulation results show that compared with the existing single-intelligent reflector millimeter wave communication system,the proposed system architecture can increase the speed of communication system by about 2.8dB.2.For the challenge of discrete phase beamforming design,this thesis proposes a beam search algorithm based on codebook classification and constructs the corresponding network model.The model includes a codebook searcher and a codebook mapper.The codebook searcher is constructed based on convolutional neural network(CNN),with channel state information as input and the index range of the optimal codeword in the codebook as output.Next,the codebook mapper searches for the optimal beamforming codeword within the index range.Furthermore,we also parallel train the proposed beam search network on the DeepMIMO dataset and Saleh-Valenzuela channel model,and train the model with imperfect channel state information to improve its robustness.Finally,the test results show that under the condition of SNR=10dB,the prediction accuracy of the codebook searcher can reach 97%.Compared with the exhaustive search method,this algorithm can find the optimal code word faster,and the system rate is close to the optimal.