The Research On Coverage Enhancement Technology For Mobile Millimeter Wave Communications In Dynamic Scenarios

Millimeter-wave(mm Wave)communications have become one of the key technologies of the fifth generations mobile communication system(5G)with its sufficient spectrum resources,and will keep the trends of continuous development and evolution.However,high-frequency-bands(3 0GHz-300Ghz)mm Wave transmissions suffer from serious path loss and poor penetrability,which bring great challenges to its practical deployment and application.Firstly,aiming to compensate path loss of mm Wave,massive multiple-input-multiple-output technology is widely adopted to achieve directional beamforming at transceivers.However,on the one hand,the transmission performance of narrow beams is highly dependent on accurate alignment of beam pairs at transceivers.Moreover,in the dynamic scenarios,the mobility of user terminals not only aggravates the challenge of real-time beam alignment,but also makes it difficult to obtain channel state information and optimize network performance in real time.On the other hand,due to the poor penetrability,blockage is one of the main bottlenecks that limit the coverage of mm Wave networks.Therefore,intelligent reflecting surface(IRS)technology has been introduced into the design and optimization of mm Wave systems to further enhance the blind-spot coverage capability.The research on mobile mm Wave communications mainly focuses on beam management,beamforming,channel estimation and so on,while there are limited discussions the coverage enhancement in the dynamic scenarios,which is one of the practical problems for the deployment of mm Wave networks.Therefore,this thesis conducts the following research on mobile mm Wave coverage enhancement technology in the dynamic scenarios:(1)The current narrow-beam based orthogonal multiple access used in mm Wave systems seriously limits the user coverage.Therefore,this thesis proposes a machine learning(ML)based mm Wave non-orthogonal multiple access(NOMA)technology,which can significantly improve the user-coverage capability of mobile mm Wave systems.Several ML methods are utilized to achieve the performance optimization of mm Wave-NOMA.In this thesis,a Gaussian process ML based Angular Domain Information(ADI)prediction method is proposed to assist in efficient beam tracking.Then,a user grouping method based on unsupervised learning is proposed by using ADI.Finally,this thesis discusses the three-dimensional beamwidth control problem in the mobile mmWave-NOMA scenario,and achieves real-time beamwidth optimization through deep learning method.(2)IRS is one of the key technologies to realize the future mobile mm Wave blind-spot coverage enhancement,but the introduction of IRS seriously complicates network architecture of mm Wave networks.In this regard,in order to unlock full potentials of IRS in mobile mm Wave communication systems,we comprehensively study dynamic beam management,cascaded channel estimation and beamforming problems.Specifically,this thesis discusses the beam management of mmWave-IRS networks from the perspectives of traditional beam-space searching and ML aided schemes,respectively.Then,this thesis proposes an efficient user-position-information assisted beam forming method.Finally,an adaptive grid matching pursuit algorithm is proposed to achieve high-resolution cascading channel estimation.Finally,research contents of this thesis are summarized,and follow-up research directions of mmWave coverage enhancement technology in the dynamic scenarios are prospected.