Joint Beamforming For Reconfigurable Intelligence Surface Assisted MISO Systems

With the formal commercial application of the fifth generation of mobile communication technology,it is expected in the next few years to people’s demand for wireless communication network traffic will times the speed of growth,and the world will realize ubiquitous wireless connections,but complex network,high hardware cost and high energy consumption are the important problems to be solved in future wireless communication.Therefore,it is necessary to seek a high-speed wireless transmission scheme with low cost and low energy consumption.Among the current candidate technologies,Reconfigurable Intelligent Surface stands out for it’s unique characteristics of low cost,low energy consumption,programmable and easy deployment,and is a feasible solution to improve the transmission performance of communication systems.With the further expansion of artificial intelligence applications,artificial intelligence based on deep learning has been widely used in communication fields such as power distribution,beamforming and channel estimation.This paper studies the joint beamforming algorithm in RIS-assisted MISO system based on deep learning,so as to improve the overall communication quality between base station and user.In this regard,this paper carries out the following analysis and research:Firstly,the RIS-assisted wireless communication system is modeled and the optimization problem is determined.Considering the RIS-assisted multi-user MISO system scenario,the communication quality may be greatly affected due to the existence of obstacles,so an auxiliary reflection link with high communication quality can be established between the base station and the user by reasonably deploying RIS and adjusting its reflection characteristics.In this paper,the design optimization of base station and RIS combined beamforming in downlink communication system is studied based on the user’s sum rate.Secondly,the principle of channel estimation based on LMMSE and beamforming based on BCD algorithm,channel estimation based on deep learning and beamforming based on BCD algorithm is analyzed,and the process of solving RIS assisted MISO system based on BCD algorithm is described.Then,the network structure of the deep learning algorithm is designed.Python environment simulation is used to build the neural network using Tensor Flow framework.After training,testing and optimization,the joint beamforming algorithm based on deep learning is finally determined.Finally,the joint beamforming algorithm of RIS-assisted MISO system based on deep learning is compared with the comparison algorithm,and the simulation is carried out in the same scene to explore the relationship between different pilot length,downlink transmission power,number of users and speed,and the simulation results are compared and analyzed.The simulation results show that the joint beamforming algorithm in RIS assisted MISO system based on deep learning is superior to the traditional comparison algorithm,which improves the system performance,reduces the complexity,and has certain generalization ability.