Research On Nonlinear Damage And Equalization Algorithm In Coherent Optical Fiber Communication System

The explosion of massive data services represented by mobile Internet and cloud computing,and the emergence of new applications such as the Internet of Things(IoT),enhanced/virtual reality,storage and other software instant services(SaaS)and infrastructure instant services(IaaS),leads to the rapid growth of data traffic generated at an exponential rate every year.With the constant expansion of network transmission capacity,the availability of wireless spectrum resources is becoming increasingly limited.For the backbone network,a 100 Gbit/s fiber optic communication system has been deployed and a comprehensive layout of 400 Gbit/s commercial systems is now underway.With the further upgrading of the transmission capacity,the single-channel 1 Tbit/s transmission has been experimentally verified,and is expected to develop into a highspeed commercial system shortly.In terms of the development of optical transceiver technology,the most important development in the past decade may be digital coherent systems.Intensity modulated direct detection systems are relatively simple in structure,transmitting information through the intensity of the optical signal,but the ability to analyze and compensate for system and link damage is insufficient.The utilization of coherent detection and Digital Signal Processing(DSP)in the system solution is better suited for long-distance networks.Coherent optical communication systems are beneficial due to their high sensitivity and capability for long transmission distances,and improve spectral efficiency through higher-order modulation-merge multiplexing.In the current coherent optical communication system,fiber nonlinear effect is still one of the most limiting factors,and it is important to study how to mitigate the effect of nonlinear effect in communication systems and break through the nonlinear Shannon limit.On the other hand,the complexity of existing methods to compensate for nonlinear effects is high and requires a trade-off between performance and complexity.Due to the stability of coherent communication system and the great potential it possesses,the research revolves around the nonlinear impairment and DSP algorithm of coherent optical communication system with multiple span segments.The following outlines the primary focus of the dissertation and the breaking results that are attained.1.Research on nonlinear damage of transceiver in coherent optical transmission systemFirstly,the transceiver model of the signal in the optical coherent transmission system is introduced,and the MZM is analyzed in depth.According to the modulation characteristics of the electro-optical signal of the MZM or IQ modulator,the performance loss and insertion loss of the sending signal under Gaussian distribution are analyzed in relation to the modulation coefficient,as well as the system performance research under the condition of pre-compensation transmission curve,are analyzed respectively,and the specific relations are given respectively.This is also the main work and innovation point of this section.Finally,by building a simulation platform,the transmission of 50 Gb/s QPSK signal on standard single-mode fiber(SSMF)is realized,and the system performance and loss in different scenarios are measured.The work in the second part is helpful in comprehensively evaluating the performance loss caused by the nonlinearity of the coherent transmitter.2.Theoretical study on nonlinear effects of multi-span channel in coherent optical transmission systemThe transmission process of optical signals are comprehended through the utilization of the nonlinear Schr?dinger equation.The solution method of the equation and the existing theoretical model of optical fiber nonlinear effect based on linear compensation algorithm are introduced.On this basis,the system theoretical model based on the Kerr nonlinear effect compensation algorithm is proposed innovatively to analyze the accumulation of noise during each stage of the back propagation process,the performance and complexity of the second-order Nonlinear Signal-Noise Interaction(NSNI)model are also evaluated,and give the closed theoretical SNR expression.Combined with the multi-span link 100 Gbit/s QPSK coherent transmission system,the validity of the fullorder NSNI theoretical model is verified in the low nonlinear and high nonlinear regions.The work in the third part is helpful in deeply understanding the performance loss caused by fiber nonlinearity in multi-span transmission systems.3.Theoretical study of channel nonlinear effects based on link power excursionThis dissertation studies the link performance damage and compensation method of coherent optical transmission system based on link power excursion.The single variable method is utilized to examine the buildup of signal,noise,and crosstalk during the backpropagation of the multi-span transmission system with a single excursion in the optical fiber link.At the same time,the linear and nonlinear compensation are theoretically deduced in detail,and the optimal signal power tracking scheme based on genetic algorithm is proposed.Combined with the simulation and experiment of a multi-span 200Gbit/s 16 QAM coherent transmission system,the different characteristics of linear compensation and nonlinear compensation on link power variation are verified,which is consistent with the theoretical curve.The novelty of the fourth part is the first evaluation of the link variation system performance and the proposed power tracking algorithm,which helps in deeply analyzing the performance loss of linear and nonlinear compensation structures in the multi-span transmission system with link power excursion.4.Research on utilizing neural network for equalization in coherent optical transmission systemIn the last chapter of the dissertation,the research of coherent optical transmission system equalization based on neural network is carried out.Firstly,the commonly used deep neural network equalization schemes are introduced.These schemes have the problems of poor interpretability and generalization.Then the traditional DSP process of digital coherent demodulation is introduced.The physical meaning of each step is very clear.The innovation of this section lies in proposing an interpretable neural network equalization scheme that combines depth neural network and digital coherent demodulation DSP.It not only uses the optimization ability of the neural network,but also optimizes the DSP configuration through the results,and optimizes the linear compensation and nonlinear compensation.Combined with the long-distance transmission experiment of a 128 Gbit/s 16 QAM coherent transmission system,the validation of the interpretable neural network equalization scheme is verified.The work in the fifth part is helpful to globally optimize the demodulation DSP process of coherent optical transmission system.