Master's Thesis

Awareness of optical network performance with machine learning methods

Alípio Fernandes Arruda — 2022

Key information

Authors:

Alípio Fernandes Arruda (Alípio Fernandes Arruda)

Supervisors:

Pétia Georgieva; Paulo Sérgio De Brito André (Paulo Sérgio De Brito André)

Published in

June 30, 2022

Abstract

Machine Learning (ML) has recently received a lot of attention in optical networks as a method for engineers to automate various management decisions and optimize network resource allocation. Data required to train ML-based models must be large and rich enough in order to build reliable models that output good predictions and avoid bias. Therefore, the first goal of this dissertation was to generate relevant data, using the GNPy transmission simulation techniques. GNPy is a vendor-independent software program that calculates the optical signal-to-noise ratio of a light channel on a particular route given the resource allocation of a fully coherent wavelength division multiplexed network. The second goal of this dissertation was to find the most efficient Machine Learning model for estimating the Optical Signal-to-Noise Ratio (OSNR) based on 11 measured quantities (features) related with typical optical transmission scenarios. Several ML regression methods were comparatively studied, namely Polynomial Regression, Random Forest, and Artificial Neural Networks (ANN). The ANN outperformed the other models and found to be the best compromise between two conflicting objectives, i.e. model performance (in terms of Mean Squared Error) and model complexity (in terms of number of trainable parameters). The final goal of this work is study the feature importance and build repression models based on a reduced feature space.

Publication details

Authors in the community:

Supervisors of this institution:

Fields of Science and Technology (FOS)

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

Publication language (ISO code)

eng - English

Rights type:

Embargo lifted

Date available:

May 2, 2023

Institution name

Instituto Superior Técnico