Research On Self-Attention Nonlinear Compensation Method For OAM Mode-Division Multiplexing Optical Fiber Communication System

With the rapid development of social networks and the continuous generation of various high technologies,the transmission capacity of single-mode fiber(SMF)gradually reached the Shannon limit,and various multiplexing technologies were widely used.Based on wavelength division multiplexing and polarization multiplexing,Orbital Angular Momentum(OAM)multiplexing has been proposed to improve the communication capacity,which has gradually become a research hotspot in the communication field because it can increase the network bandwidth and improve the channel capacity.However,there are some inherent nonlinear impairments in the OAM fiber optic communication system,which reduces the transmission rate and capacity and affects the signal transmission performance.Therefore,how to achieve nonlinear compensation in OAM mode-division multiplexing fiber optic communication has become an important research hotspot at national and international.This thesis takes the research of nonlinear compensation in OAM mode-division multiplexed fiber optic communication system as the research direction,the main research contents include:self-attentive based nonlinear compensation method,shift autocorrelation based nonlinear compensation method and principal component analysis based nonlinear compensation method self-attentive,the main research work and innovation points of this paper are as follows:(1)A nonlinear compensation method based on self-attention is proposed,which introduces the self-attention mechanism into the nonlinear compensation of OAM mode-division multiplexing fiber optic communication system,improves the nonlinear compensation accuracy,and reduces the time of the algorithm.The feasibility of the method is experimentally verified for the PAM4-OAM mode-division multiplexing fiber optic communication system.The final results show that the method has a sensitivity gain of 2 dB and 1 dB compared to the Volterra-based nonlinear compensation method and the convolutional neural networkbased nonlinear compensation method,respectively.(2)A nonlinear compensation method based on shift autocorrelation self-attention is proposed,which incorporates the shift autocorrelation matrix of the signal based on the weight matrix of the self-attention algorithm,and the accuracy of nonlinear compensation is improved by adding the feature modeling of the sequence.The feasibility of the method is experimentally verified for the PAM8-OAM mode-division multiplexed fiber optic communication system,and the performance of the method is investigated.The final results show that when the BER threshold is 1e-2,the sensitivity gain of the method is 3 dB and 2 dB compared with Volterra and convolutional neural network-based nonlinear compensation-based methods,respectively.(3)A self-attention nonlinear compensation method based on principal component analysis is proposed.The method is based on the selfattentiveness algorithm and uses the method of principal component analysis to reduce the dimensionality of the three high-dimensional matrices in the algorithm,which reduces the computational complexity while improving the performance of the fiber optic communication system.The feasibility of the method is experimentally verified for the PAM8OAM mode-division multiplexed fiber optic communication system.The results show that the method can effectively compensate for nonlinearities in fiber optic communication systems and that the computational complexity is reduced and the fiber optic communication system is more capable of nonlinear compensation compared with the self-attentive algorithm.