Research And Implementation Of Beam Scanning And Tracking In Wireless Millimeter-Wave Communication

With the development of millimeter wave technology,a wealth of theoretical research and prototype validation has been carried out by many organizations.The high path loss characteristics of millimeter wave signals promote the use of beamforming techniques to increase the link budget.High quality communication links can be established by using beam tracking technology in dynamic environments.This thesis aims to build a millimeter wave application scenario for ultra-high definition video transmission and design a suitable beam control scheme for two different types of phased array antennas.The tracking performance is improved by implementing an intelligent beam tracking algorithm,and a corresponding optimization scheme is proposed for the existing problems.Firstly,the millimeter wave communication physical layer standard and multi-antenna technology are studied,and a video transmission system is designed based on the NI millimeter wave prototype.In the baseband processing scheme,parallel processing with different priorities is designed to match the baseband clock and sampling rate.In the video transmission scheme,the field programmable gate array(FPGA)is extended as a network card to enable fiber optic communication of ultra-high speed video streams.In the video streams processing scheme,the interface rate matching is achieved through a data padding scheme,a FIFO video buffer structure is designed based on highcapacity synchronous dynamic random memory,and a boundary protection mechanism is proposed to solve packet boundary corruption problem.Then,the overall beam control scheme is designed for the characteristics of phased array antennas.The data is interacted between the baseband processing module and the beam control module based on the inter-FPGA communication mechanism,including beam measurement value and beam training enable signal.The control frame data stream is written to the antenna through a custom transmission protocol.The modules for beam initialization,beam tracking,beam recovery and antenna control are designed by establishing separate state machines,which work in coordination through multiple custom enable signals.Next,the hardware implementation of beam control is completed based on the NI-PXIe platform.The interface and control frame structure of the two antennas are analyzed in detail and the codebook pre-stored in the memory module is generated according to the respective control mode.A local codebook search is used to implement the beam tracking process based on the historical optimal beam,and a threshold detection mechanism is established to implement the beam recovery process.According to the antenna communication interface characteristics,highspeed SPI protocol and parallel port communication protocol are designed respectively.Under the guideway motion scenario,it is experimentally verified that the system can switch beam direction at high speed,achieve continuous and stable beam tracking,and can significantly improve the quality of the communication link.Finally,intelligent beam tracking algorithms are introduced in combination with reinforcement learning.From the perspective of sensing dynamic changes in the environment,a beam tracking algorithm based on a random choice optimization model is introduced.A maximum confidence upper bound algorithm is used as the behavioral selection strategy,and an implementation scheme is designed on the hardware platform to verify the performance improvement of the algorithm in scenarios changing dynamically.The limitations of the algorithm are also analyzed in terms of environmental information utilization and selection mechanism establishment,after which corresponding optimization schemes are proposed,including continuous beam state definition and quadratic behavior construction.After that,an improved algorithm is proposed based on Markovian decision process.The algorithm is validated to be slightly lower than exhaustive search in performance,but much higher than several other intelligent beam tracking algorithms,and far lower than exhaustive search in training cost while similar to several other algorithms by simulation.