Design And Implementation Of Beam Alignment And Tracking Algorithm Based On Software Defined Radio

Millimeter wave,with its advantages of ultra-wide bandwidth and ultra-high data rates,has become the central technology for research on Beyond 5G wireless communication systems.Millimeter wave combined with massive MIMO provides high beamforming gain through highly directional spatial processing,but beam misalignment is a key factor leading to decreased performance in millimeter wave wireless communication.This is reflected in two aspects: long delays and high cost during beam alignment,and high tracking complexity and lag in beam tracking.Therefore,firstly,the key focus of low-latency beam alignment is to reduce search costs without relying on channel state information as prior knowledge.Secondly,the key to robust beam tracking is to achieve real-time updating of beam direction and provide stable wireless communication service quality.This thesis focuses on low-complexity millimeter-wave beam alignment and tracking.The thesis proposes a low-latency and high-precision multi-directional hybrid beam alignment algorithm and a robust and efficient beam tracking algorithm for generating a candidate beam set from channel chart.The beam alignment is based on multi-directional search.For the hybrid beam alignment,this thesis proposes a two-stage search algorithm from the receiver’s perspective,which has higher accuracy than the wide plus narrow beam search.From the transmitter’s perspective,the thesis proposes a Bayesian inference method to resolve sparse propagation paths from multiple directional information without the need for feedback.Additionally,this thesis builds a 26 GHz mm Wave wireless communication platform based on software-defined radio and broadband transmission systems to verify the performance of the proposed algorithm.The thesis’ s innovative points are summarized around low-complexity beam management:(1)A fast hybrid beam alignment algorithm based on two-stage scanning andBayesian inference is proposed.The algorithm takes the framework of multi-directional beam search for beam alignment,and discusses separately two feasible methods of multibeam and multi-arm beamforming based on different characteristics of the system’s RF link.In addition,for the initial access process,considering the characteristics of the millimeter-wave communication that the receiving end has available measurement feedback and the transmitting end has no prior knowledge or feedback of the blind channel,the beam alignment problem that requires joint transmission and reception is decoupled into the receiving algorithm with dynamic reception conditions and the Bayesian inference problem based on measurement information for the transmitting end,achieving fast beam alignment under the mixed angle determination strategy.Simulation results show that the algorithm has the advantage of search complexity,and it is significantly better than other algorithms with only 20% of search complexity.(2)A low-complexity beam tracking algorithm based on generating a candidate beam set from the channel chart is proposed.The algorithm fully utilizes the environmental characteristics of the channel and compresses high-dimensional channel state information into low-dimensional space as the core idea.It is based on deep learning training using auto-encoders to construct a channel chart with neighborhood range constraints and angle information in the low-dimensional space.In the channel chart,the path correlation is reflected in the distribution of the low-dimensional space,and a small number of candidate beam sets are generated based on the classification features of the channel map,greatly reducing the number of searches.Simulation results show that the low-complexity channel chart beam tracking has weak tracking lag and high accuracy.(3)Beam alignment and beam tracking algorithms were tested in actual deployment,based on a real measurement platform built with software-defined radios and a 26 GHz antenna.Field tests showed that the proposed fast mixed beam alignment algorithm has a significant advantage in search complexity,with only 20% of the search complexity compared to other algorithms,and an accuracy of over 80%,consistent with simulation results,which verifies the performance of the beam alignment algorithm under a 15 dB signal-to-noise ratio.The channel chart,trained from actual field measurement data to achieve low positioning error and low search error,is used to generate a candidate beam set.The beam tracking performance is then tested in NLOS scenarios.This method can maintain a 3 dB performance advantage and achieve high-performance and stable beam tracking with minimal scanning overhead.In summary,the multi-directional fast mixed beam alignment algorithm and beam tracking algorithm based on channel chart generation of candidate beam sets proposed in this article have low search complexity and high stability performance.