Hybrid Precoding Method For Millimeter-Wave Massive MIMO System

With the development of society,the existing spectrum resources are far from satisfying the needs of the masses of users.People are eager to find new spectrum resources.Therefore,researchers of the fifth generation mobile communication technology(5G)in the future turned their attention to Millimeter-wave band currently not authorized.5G communication system can utilize the huge bandwidth of millimeter wave band and cooperate with the high array gain brought by large-scale MIMO technology to realize the data transmission rate of Gbps.Large-scale MIMO technology can effectively increase the system capacity,improve the spectrum efficiency and make up the millimeter wave's own existence deficiencies,such as serious path loss and others.Although large-scale MIMO technology has many advantages,its huge number of antenna arrays in millimeter-wave systems also bring high hardware costs,high algorithmic complexity and high energy consumption,while hybrid precoding technologies can effectively solve these problems.Therefore,this thesis studies and analyzes the hybrid precoding technology of large-scale MIMO system.By analyzing the limitations of the traditional full-digital precoding technology in the millimeter-wave system,a digital analog hybrid precoding technique suitable for the millimeter-wave system is proposed.This thesis first introduces the technical background of hybrid precoding using millimeter-wave massive MIMO system.Secondly,the structure model of the traditional MIMO system and the commonly used millimeter-wave large-scale MIMO system are analyzed in detail,and the corresponding signal transmission model is built up.Then the channel model is analyzed,and a complete channel matrix model suitable for millimeter wave is built up.Then,by introducing some commonly used matrix factorization methods,such as singular value decomposition(SVD)and geometric mean value decomposition,the method of precoding based on matrix factorization is introduced.Several traditional full-digital precoding method is introduced first,then introduce the hybrid precoding method based on singular value decomposition and orthogonal matching pursuit(OMP)algorithm.Because SVD-based precoding technique decomposes the channel into several subchannels with greatly different gains,a hybrid precoding method based on geometric mean decomposition is proposed,which effectively solves this problem.Simulation result shows that compared with the precoding method based on SVD and OMP,this method has better performance both in spectrum efficiency and bit error rate.Finally,the thesis introduces the system precoding design method in the presence of interference,including the traditional full-digital andseveral commonly used hybrid precoding matrix design methods,and presents a new hybrid precoding design method.Finally they are simulated and compared.