The Research On Intelligent Reflecting Surface-aided Millimeter Wave Wireless Communication Technology

The Millimeter wave(mm Wave)band has opened a new era for 5G communications.Due to the short wavelength of millimeter wave,in the process of the transmission path loss is more serious,and massive MIMO technology can provide greater transmission gain,therefore the millimeter wave communication technology combined with a massive MIMO technology can effectively alleviate the problem,but as the MIMO ever greater will inevitably lead to increased the antenna number,thus increasing the cost of base station deployment and high-power consumption.In addition,in the room,the mm Wave line-of-sight link is not reliable due to the presence of obstacles.Intelligent reflecting surface(IRS)is a cost-effective solution that not only improves system performance by using a large number of low-cost passive components to reflect incident signals,but also provides reliable transmission for non-line-of-sight link.This thesis mainly studies the IRS-aided mm Wave wireless communication technology.In this thesis,the conventional mm Wave massive MIMO system model and channel model are established,and on this basis,IRS is introduced to establish the IRS-aided MIMO wireless communication system model.And then two channel models from base station to IRS and IRS to target user are given.The optimization problem of IRS reflect phase shifter is studied,and the performance difference between continuous and discrete values of IRS reflect phase shifter is analyzed.In this thesis,IRS-aided single user mm Wave massive MIMO communication system under line-of-sight link is firstly studied.Considering IRS discrete phase shift 1-bit quantization,the optimization problem of reflect phase shifter was transformed into discrete value search problem,and genetic algorithm is used to search the value of optimal phase shifter.Secondly,the problem is extended to multi-user on the basis of single user,the IRS-aided multi-user mm Wave communication system under non-line-of-view link is studied,and the performance difference between 1-bit quantization and 2-bit quantization when IRS reflect phase shifter takes discrete value is analyzed by using machine learning method.Finally,based on the above analysis,the case of IRS phase shifter taking a continuous value is considered.Due to the existence of non-convex constraints,conventional convex optimization methods can not obtain the optimal solution of the problem,so the gradient projection algorithm is considered,and the effect of continuous phase shift and discrete phase shift on the performance is analyzed.