On AI Based Beamforming And AP Management In Mmwave Networks

With the rapid development of wireless communication technology and the emergence of a large number of intelligent terminal devices,people's requirements for wireless high-speed communication are also getting higher and higher.However,the conventional microwave spectral resources have been crowded and they cannot meet the requirement of high-speed wireless transmission.To address this issue,the communication industry turns its attention to mmWave in high frequency band,therefore,mmWave will be used for wireless communication in the next-generation wireless local area network standard IEEE 802.11 ay.Furthermore,in order to increase the transmission range of the wireless signal and improve the robustness of signal transmission,Multi-AP system can be used for signal transmission,in the Multi-AP system,APs are connected to the controller through wired or wireless.The controller is used to centrally control the entire system.When Muli-AP system uses mmWave for communication,many problems will be solved more complicated and difficult.For example,beamforming in mmWave Multi-AP system will become more complicated,especially when STAs(Stations)have strong mobility.In addition,the wireless resource management problem of the system is urgently needed to be solved.For convenience of description,we refer to the pairing between APs and STAs and the AP power allocation in the mmWave Multi-AP system as an AP management problem.Since AI(Artificial Intelligence)technology has many advantages compared with traditional methods,this thesis uses AI technology to study beamforming and AP management in mmWave Multi-AP system.We firstly investigate beamforming problem in mmWave Multi-AP system,an improved mmWave Multi-AP beamforming scheme is proposed based on the IEEE 802.11 ay beamforming mechanism,the simulation results show that the improved scheme can save more training overhead.For mmWave Multi-AP mobile system,we use coordinated transmission to improve the robustness of system transmission,and then we design a coordinated beamforming scheme for coordinated transmission.In order to further reduce the training overhead of the coordinated beamforming scheme and improve the performance of the system,combining deep learning with coordinated beamforming scheme,we propose a mmWave coordinated beamforming scheme based on deep learning.The deep learning model is designed to predict the optimal beamforming vector of each AP.The simulation results show that the mmWave coordinated beamforming scheme based on deep learning can obtain higher effective rate of system,and the deep learning model has good performance in both the LOS(Line-of-sight)environment and the NLOS(Non Line-of-sight)environment,it can quickly adapt to the changing wireless communication environment.Then,we investigate the AP management problem in mmWave Multi-AP system to maximize the sum-rate of entire system.The AP management problem includes the pairing between APs and STAs and the AP power allocation.We first build a mathematical model,and then describe the AP management problem as an optimization problem.Since the optimization problem is an NP-hard problem and non-convex,the direct solution cannot be used to obtain the optimal solution,Therefore,we design an iterative algorithm to decompose the optimization problem into two sub-problems and solve them separately to obtain a suboptimal solution for the AP management.Finally,an AP management scheme based on deep learning is proposed.For the its solution,we use an iterative algorithm of the design to generate training and testing data,which are then used for training and testing of the deep learning model.The simulation results show that the performance of the AP management scheme based on deep learning is close to the iterative algorithm,but its computation time is much lower than that required by the iterative algorithm.