Investigation Of Modeling And Mitigation Of Nonlinear Penalties In The Next-Generation Elastic Optical Networks

The next-generation elastic optical networks(EONs)are evolving towards the desired features including large capacity,intelligence and programmability.To meet the requirement of large capacity,the transmission rate of the single wavelength in optical networks has been continuously improved from 10G to 100G.Now the 200G-system has been widely used commercialy while the 400G-system has been small amount of commercial.However,the impairements induced by the" nonlinear effects are increasingly obvious with the increase of transmission rate,limiting the maximum transmission distance.Therefore,modeling,prediction and mitigation of nonlinear penalties have been the key techniques for the capacity expansion of the next-geneartion EONs.Additionaly,the limited maximum transmission distance means the increasing demand of relay nodes.It not only directly leads to the capital expenditures(CAPEX)for construction of infrastructure,but also introduces additional filtering impairments.Since the relay nodes are mainly based on the reconfigurable of optical add-drop multiplexing(ROADM),the filtering effect is strengthened with node numbers.Consequently,the filtering penalty has been one of the obstacles for upgrading the existing networks to the next-generation EONs.This dissertation focus on two aspects:the first one is modeling and compensation of the nonlinear noise induced by the cross-phase modulation(XPM),and the other is estimating the filtering penalty in elastic optical networks within cascased ROADMs.Several technical schemes are proposed.The main innovations for this dissertation are as follows:Firstly,two kinds of estimation models based on the Gaussian noise(GN)model are proposed,including the full-scenario estimation model and the agile estimation model.The full-scenario estimator removes the assumption that the transmitted signals behaves as Gaussian noise.The agile estimation model takes the add/drop at the ROADM node into consideration.Different form the GN model,the two proposed models can be applied to the dynamic optical networks.Simulations show that both models have good accuracy and can be good tools to help planning the networks.Secondly,the multi-parameter blind equalization based on the time varying ISI model is proposed to mitigate the nonlinear noise in the optical networks.The time varying ISI model shows that the nonlinear noise has time-dependent characteristics and their autocorrelation function have long temporal correlation.Based on this time-dependent feature,the adaptive tap coefficients and step size are used at the same time to update the qualization,enabled to converge faster and predict the filtering penalty with more accurary.Simulations show that the proposed equalization can achieve better mitigation gain compared with RLS filter.Thirdly,focusing on the filtering penalty induced by cascaded ROADMs,an optical filtering penalty estimation using artificial neural networks(ANN)in EONs is proposed.We first investigate the impact of ROADM location distribution and bandwidth allocation on the narrow filtering effect.Afterwords,an approach based on ANN is proposed to estimate the filtering penalty under various link conditions.Extensive simulations with 9600 links are implemented to demonstrate the superior performance of the proposed scheme.