Research On Adaptive Beam Search Algorithm In Millimeter Wave Communication

Due to the abundant available spectrum in the millimeter wave frequency band,millimeter wave communication has gradually been incorporated into major mobile communication standards and has become one of the key technologies in the field.However,the high-frequency millimeter wave signals suffer from significant path loss during propagation.To compensate for the high path loss,millimeter wave communication terminals need to use beamforming to obtain directional gain when sending and receiving signals.This is achieved by aligning the beams towards the main path direction of the channel.To attain millimeter wave beam alignment,a common method is to search for the dominant path direction of the channel by beam spatial scan and determine the direction of beamforming.Classical spatial scanning methods generally assume a fixed search time,yet different communication scenarios require different search times to achieve satisfactory beam alignment accuracy.Therefore,methods that can adaptively adjust the beam search time are urgently demanded to accommodate various mobile communication scenarios.To address this problem,a series of time-adaptive beam alignment methods are designed in this paper.All these methods are capable of stopping the beam search in time with a guaranteed beam alignment accuracy by iterative beam scanning combined with real-time inference,regardless of whether the real-time Signal-to-Noise Ratio(SNR)of the channel is known or unknown.These methods proposed are suitable for application in phases where environment information is lacking such as initial access and have a good adaptability to the scenarios.The paper first proposes a generalized likelihood ratio test(GLRT)based adaptive beam search algorithm,i.e.,GLRT algorithm.The algorithm judges the relative strength between different beams in the candidate beam set in real time by GLRT,and eliminates the weaker ones from the candidate beam set according to the judgment results.Then it checks whether the iteration of beam scanning should be stopped according to the status of the candidate beam set.In addition,a beam shifting method is proposed to assist the final beam selection,which reduces the loss of beam alignment accuracy caused by using fixed spatial resolution beam codebook when the main path of the channel is between two beams.The simulation experiments of typical millimeter wave communication scenarios such as line-of-sight(LOS)and none-line-of-sight(NLOS)demonstrate that GLRT algorithm can achieve a good trade-off between search time and search accuracy.Based on the GLRT algorithm,this paper further proposes an adaptive beamforming search algorithm based on the Multi-Arms Bandit-Upper Confidence Bound(MAB-UCB)strategy,which can shorten the search time.The algorithm designs the MAB-UCB beam scanning strategy based on the UCB method in the MAB theory.In each iterative beam scanning,a beam is selected for signal acquisition based on the UCB criterion,and the next beam to be measured and the termination of beam search are determined based on the updated signal.Compared with the GLRT algorithm’s scanning strategy that requires scanning all beams in the candidate beam set each time,the MABUCB algorithm can effectively reduce the time spent using non-optimal beams to measure channels during beam scanning,thus improving the efficiency of beam search.Simulation results verify that the MAB-UCB algorithm has higher efficiency in beam search and significantly reduces search time compared to the GLRT algorithm while achieving satisfied search accuracy.