Research Of Beam Training Algorithms In Hybrid Beamforming Systems

Millimeter-wave communication and massive MIMO technology have been considered as the core technologies to realize the bright vision and requirements of future wireless communication systems.Hybrid beamforming technology takes advantages of traditional digital precoding and analogue beamforming technologies,makes a compromise between hardware complexity and system performance,and provides a feasible approach to yield millimeter-wave massive MIMO systems.However,under hybrid beamforming systems,there are many problems to be solved urgently in analog beam training part.The low SNR brought by the large bandwidth,narrower and numerous beams caused by large-scale antenna arrays,result in the addition of complexity and beam alignment difficulties.Meanwhile,the millimeter signals are highly vulnerable to link blockage,thus the topic of beam recovery handling is still to be improved.Therefore,this paper mainly focus on the research of beam training and beam recovery in hybrid beamforming systems.In this paper,a two-stage beam training algorithm is proposed based on fixed subarray formation.The division of antenna subarrays results in an exponential search complexity.The two-stage training scheme is designed by using the unilateral search pattern,the search complexity is reduced to linear complexity by gradually determining the optimal beam of each subarray.Simulation results show that the performance of the proposed scheme has extremely limited loss compared with exhaustive search.Futhermore,in this paper,an anti-blockage beam training-switching algorithm based on channel clustered characteristics is proposed to overcome link blockage during data transmission.Considering sparsity and clustered characteristics of mm Wave channels and high probability for adjacent beams to be simultaneously blocked,we propose a new criterion for the selection of backup beams based beam power and correlation,design two beam grouping algorithms and detail beam training and depict the detailed beam training and beam switching procedures.Simulation results show that the proposed beam training scheme achieves near-optimal performance at initial beam training stage.Furthermore,the new method for identifying backup beam pairs is highly effective to improve the spectral efficiencies of systems under blockage environments.