In the context of this PhD Thesis defense, Fabiano Locatelli presents his work realized at CTTC.
One of the possible ways to address the constantly increasing amount of heterogeneous and variable internet traffic is the evolution of the current optical networks towards a more flexible, open, and disaggregated paradigm. In such scenarios, the role played by Optical Performance Monitoring (OPM) is fundamental. In fact, OPM allows to balance performance and specification mismatches resulting from the disaggregation adoption and provides the control plane with the necessary feedback to grant the optical networks an adequate automation level. Therefore, new flexible and cost-effective OPM solutions are needed, as well as novel techniques to extract the desired information from the monitored data and process and apply them. In this dissertation, we focus on three aspects related to OPM. We first study a monitoring data plane scheme to acquire the high‑resolution signal optical spectra in a nonintrusive way. Then, we identify two main placement strategies for such monitoring solutions, enhancing them using two spectral processing techniques to estimate signal- and optical filter-related parameters. Specifically, we propose a way to estimate the Amplified Spontaneous Emission (ASE) noise using optical spectra acquired at the egress ports of the network nodes and the filter features, using spectra captured at the ingress ports of the network nodes. To do so, we leverage Machine Learning (ML) algorithms and the function fitting principle, according to the considered scenario. Finally, we identify a potential application for the monitored information. Specifically, we propose a solution for the optimization of the subchannel spectral spacing in a superchannel. Leveraging convex optimization methods, we implement a closed control loop process for the dynamical reconfiguration of the subchannel central frequencies to optimize specific Quality of Transmission (QoT)-related metrics.