Physical-Layer Impairment Estimation for Arbitrary Spectral-Shaped Signals in C+L-Band Optical Networks

Fiber-optic networks are the data transmission foundation of modern communications. Traditional C-band transmission can no longer meet the escalating communication demands, prompting the commercialization of C+L-band fiber-optic networks as a solution. Accurate modeling of physical-layer impairments (PLIs) is crucial for optimizing the performance of backbone fiber-optic networks. The widely-applied inter-channel stimulated Raman scattering-Gaussian noise (ISRS-GN) model has been developed to analytically estimate the accumulation of noise considering the ISRS effects during C+L-band signal transmission in a state-dependent manner. However, this model assumes that signals have rectangular spectra, leading to significant estimation errors when this assumption is violated. In practice, signal spectra are not rectangular; they are typically root-raised-cosine-shaped at transmission and are affected by various channel impacts, such as filtering effects. To address this issue, we propose the C+L-band component-wise Gaussian noise (C+L-CWGN) PLI model that accounts for arbitrarily spectral-shaped demands. The C+L-CWGN model is computationally efficient and suitable for most network management approaches. Our results indicate that the C+L-CWGN model can reduce PLI overestimation by more than 27.1% in typical scenarios.