Enhancing digital twin transmission model to enable optical network infrastructure as a service

In the dynamic landscape of telecommunications, optical networks have become the backbone of high-speed data transmission systems, reshaping global information flow. To optimize and share this pervasive infrastructure, there's a call for the introduction of disaggregation and openness in optical networking. Openness enables the possibility to implement an optical network digital twin (DT), which is the core for an optimized software-defined management of a complex system. The virtualization and SDN, achieved thanks to the DT development, allow the exploitation of Network-as-a-Service paradigm (NaaS). The adoption of NaaS offers several advantages, enabling a more agile and cost-effective network deployment, allowing organizations to scale their infrastructure based on demand. In the framework of open and disaggregated optical networks architectures, re-configurable add/drop multiplexers, based on wavelength selective switches (WSSs) technologies, providing switching at optical level, are key elements to be inspected in the development of an optical network DT. Such devices introduce most of the Polarization Dependent Loss (PDL) in optical systems that results in a relative degradation of signal-to-noise ratio (SNR). Depending both on frequency and ports, the PDL introduced by each WSS must be treated as a random variable in the vast majority of use cases, where a full characterization of all the WSSs is not available and not feasible. Based on the previous assertions, there is the need of a more accurate model for the estimation of PDL-induced penalty. The work of this thesis focuses on PDL mathematical modelling, deriving tools to determine a penalty in term of SNR. The guideline of all the work has been the objective to provide a new tool to deal with disaggregation, in the context of the enhancement of “GNPy” DT transmission model developed in PLANET team of OptCom research group of Politecnico di Torino. In treating the PDL problem, special attention has been dedicated to how PDL acts on noise introduced along the path, since each noise contribution experiences a different effect in term of PDL, since crossing a different number of WSS. In addition, also Digital Signal Processing (DSP) effect has been taken into account, considering that the distortion applied on the dual polarization (DP) information signal by the PDL is totally recovered by DSP, while noise’s distortion is not, since noise is, in general, introduced along the path and it does not experience the entire link PDL. The key idea behind all the developed framework is to exploit the description of PDL effect on noise given by power transfer matrices, underlining the evolution of the noise from the power point of view. The proposed model has been validated through an experimental analysis carried out at LINKS Foundation laboratories, in order to assess the goodness and the effectiveness of the adopted paradigm. The laboratory analysis has been divided into three steps in order to characterize first the PDL of each device and then of the entire system. Then, a transceiver has been inserted and some noise sources distributed along the considered path. Finally, software simulator has been developed and an extensive Monte Carlo investigation has been performed, in order to validate the mathematical model and to compare it with the experimental results collected in the previous step.