OAM Simulation And Recognition Based On Deep Learning

Free Space Optical Communicatin(FSO)is an emerging communication technology,which uses the atmospheric environment as a transmission medium and gets rid of the limitations of physical communication links.Compared with traditional optical fiber communication and mobile communication,FSO has the advantages of easy deployment,strong resistance to electromagnetic interference,no spectrum restrictions,large communication capacity,and high flexibility.Therefore,FSO communication system is a communication method with great application scenarios.,which can meet the needs of largecapacity communication in the era of big data.In recent years,Orbital Angular Momentum(OAM)technology has been used as a basic characteristic of light beams,which can be used in the field of free space optical communication.Since the OAM modes are infinite and different OAM modes are orthogonal to each other,the modulation of the OAM beam information can improve the transmission capacity of the FSO system.The efficiency of traditional OAM pattern recognition will be affected by the atmospheric environment and physical equipment,and it is difficult to accurately detect and identify OAM beams,thus affecting OAM-FSO communication.To address this issue,this paper explores a deep learning-based OAM pattern recognition method.The deep learning algorithm can learn and identify the characteristics of OAM beams.This method can effectively detect and identify OAM beams,thereby improving the efficiency of communication.The emergence of this method has greatly promoted the development of free space optical communication.In the future research will receive extensive attention.The main research work and achievements of this paper are as follows:(1)Introduce the principle of OAM beam and its mathematical expression,explain the method of OAM beam generation,use Matlab to simulate the phase and light intensity of OAM beam,and conduct in-depth research on the generation and detection of composite OAM beam,An optimal design method is proposed,and the proposed scheme is verified by a combination of numerical simulation and experiment.(2)The principle and characteristics of atmospheric turbulence and the transmission effect of OAM beams in atmospheric turbulence were analyzed.Using MATLAB simulation technology,the transmission of composite OAM beams in a simulated turbulent environment was realized.Through experimental research,the atmospheric channel pair communication was analyzed.The influence of system transmission,which mainly includes the influence of turbulence intensity and transmission distance on transmission performance.(3)The transfer learning model and the generative adversarial neural network were used to establish the recognition model,and the multi-mode OAM beam recognition was realized in the simulated atmospheric turbulence environment,and the influence of two different optimizers on the OAM pattern recognition rate was studied.In different turbulence environments,the network parameters are optimized respectively.The experimental results show that the method based on deep learning can effectively identify different OAM modes and improve the performance of the OAM-FSO communication system.