Edoardo Longo
Physics-Guided Machine Learning for Non-Linear Interference Estimation in Dispersion-Managed Scenarios.
Rel. Vittorio Curri, Emanuele Virgillito. Politecnico di Torino, Master of science program in Communications Engineering, 2026
|
Preview |
PDF (Tesi_di_laurea)
- Thesis
Licence: Creative Commons Attribution Non-commercial No Derivatives. Download (2MB) | Preview |
Abstract
Optical fibers are the backbone of high-capacity communications, with submarine links carrying most of the international traffic. Modern systems rely on coherent technology, where chromatic dispersion is digitally compensated at the receiver (uncompensated transmission, UT). In contrast, legacy IMDD systems relied on dispersion compensation along the link (dispersion-managed, DM). The continued use of DM links motivates the development of rapid quality-of-transmission evaluation tools. Although optical transmission is an aggregated process, our simulations show that it can be accurately approximated with a spectrally and spatially disaggregated approach. The proposed model quantifies the non-linear interference generated by each pump using the GN model, the correlation coefficient accounting for coherency accumulation between spans, and the signal Gaussianization coefficient.
Since estimating these coefficients requires a prohibitive number of simulations and no analytical solution is available, we adopt a hybrid physics-guided machine learning approach
Relators
Academic year
Publication type
Number of Pages
Course of studies
Classe di laurea
URI
![]() |
Modify record (reserved for operators) |
