Research On Simulation Of Orbital Angular Momentum Shift Keying Based On Machine Learning

Since the 20 th century,the capacity of optical communication system has reached the upper limit by using optical physical characteristics such as optical complex amplitude(amplitude,phase),time,wavelength and polarization.In order to further improve the capacity of optical communication system,the application of Orbital Angular Momentum(OAM)in optical communication is studied in this thesis.OAM is a new physical dimension,which is independent of the above optical physical features.The beam carrying OAM is called vortex beam.Orbital Angular Momentum Shift Keying(OAM-SK)is an optical communication method that uses vortex beam to realize information transmission in free space.However,the intensity distribution of vortex beams is greatly affected by the external environment.How to accurately detect the intensity distribution of different vortex beams is a difficult problem.Considering the advantages of Convolutional Neural Network(CNN)in optical image detection and classification,such as high speed,high accuracy and strong anti-interference ability,this thesis combines OAM-SK with CNN,which not only improves the demodulation speed of the communication system,but also improves its anti-interference ability.The main contents of this thesis are as follows:Firstly,the basic theory of vortex beam in OAM-SK is reviewed.By comparing the characteristics of different vortex beams and the generation methods of vortex beams,Laguerre Gaussian beams are selected,and holograms are set on the spatial light modulator to generate vortex beams.The detection method of vortex beam and the transmission theory of vortex beam in free space are introduced,and the detection characteristics of vortex beam and the influencing factors of vortex beam intensity distribution in OAM optical communication system are obtained,which provides the basis for the selection of detection characteristics and influencing factors for the later system design.Then,a noncoherent superimposed OAM-SK communication system is proposed to reduce the number of OAM mode values required in Mixed OAM Amplitude Shift Keying(OAM-ASK)coding.The vortex beam based on OAM-ASK encoding mode is generated by spatial light modulator,and the convolution neural network is used to recognize and process its intensity information,demodulate the digital information on the corresponding n-channel vortex beam,and integrate and classify the digital information to restore the transmitted image.The simulation results show that the proposed system performs well in weak to medium turbulence and long wavelength.When transmitting 4-bit information,the number of OAM mode values required in oam-ask coding is reduced by 4 times,and the refresh rate of spatial light modulator is also reduced.Finally,another OAM-SK communication system combined with Gaussian beam is proposed,which effectively solves the problems of difficult implementation and high system cost in the noncoherent superposition OAM-ASK scheme.Gaussian beam is introduced into OAM-ASK coding technology,and the system structure is simplified by incoherent superposition of vortex beam and Gaussian beam.The composite beam is processed by convolution neural network,and finally the transmitted digital information is recovered.The simulation results show that the system has good performance in low turbulence and long wavelength.When transmitting the same bit information,compared with the noncoherent superimposed OAM-SK communication system,it only needs half of the number of spatial light modulators,which greatly reduces the cost of the system and simplifies the structure of the system.