Research On Optical Signal Modulation Format Identification And OSNR Joint Monitoring Technology Based On CNN

With the increasing number of users and the diversification of user services,the spectrum bandwidth resources of optical networks are drying up.High-order modulation formats(MF)with high-speed and high-spectral efficiency become the inevitable trend of the next-generation optical communication networks.High-speed signal transmission performance is greatly affected by various channel impairments.Flexible and effective optical performance monitoring(OPM)technology plays an increasingly important role in ensuring the correct and effective transmission of signals.Therefore,real-time and effective monitoring of signal quality must be carried out in the process of signal transmission,especially at intermediate nodes.Optical signal-to-noise ratio(OSNR)is a measure of the degree of noise interference from a signal and is directly related to the bit error rate(BER),making it an important monitoring target in OPM.As a prior knowledge of OPM,MF must be known in advance or obtained from the upper protocol.In the next generation of dynamic optical networks,the MF may change,and the processing capacity of intermediate nodes is limited.It is difficult to obtain prior information such as modulation format and transmission rate from the upper protocol.Therefore,the modulation format identification(MFI)at the intermediate node is also an indispensable part of OPM.In this paper,OPM technologies for high-order modulated signals are studied.The main work and innovation are as follows.:(1)The advantages and disadvantages of existing OPM technologies are investigated,and OSNR monitoring and MFI technologies are emphatically studied.Considering the flexibility,cost-effectiveness,implementation complexity and other factors,it is pointed out that the asynchronous sampling based monitoring scheme has the characteristics of low complexity and high flexibility,so it is more suitable for application in intermediate nodes.(2)OSNR monitoring for intermediate nodes requires simple,flexible and efficient implementation.In view of the above characteristics,an OSNR monitoring scheme based on asynchronous amplitude sampling(AAH)and deep neural network(DNN)is proposed from the data point of view.Simple data distribution statistics and human brain-like DNN model reduce the computational complexity and improve the accuracy of data analysis.The simulation results show that in back-to-back transmission,the scheme can achieve 100%correct discrimination of OSNR of three quadrature amplitude modulation(QAM)signals,but the scheme has some deficiencies in medium and long distance transmission.(3)Aiming at the problem of poor performance of AAH-based OSNR monitoring scheme in medium and long distance transmission,an OSNR monitoring scheme based on asynchronous delay tap sampling(ADTP)and convolutional neural network(CNN)is proposed from the perspective of image.The simulation results show that the scheme can correctly distinguish the OSNR of three QAM signals when the chromatic dispersion(CD)is 1920 ps/nm,and its strong anti-dispersion ability is more suitable for medium and long distance transmission.(4)Aiming at the difficulty of obtaining a prior information such as MF at intermediate nodes,a joint monitoring scheme of OSNR and MFI based on ADTP and CNN is proposed.The feasibility of the joint monitoring scheme is verified by ADTP data sets of three QAM signals with CDs of 1280 ps/nm,and OSNR monitoring with transparent modulation format is realized.