Quality of transmission estimator retraining for dynamic optimization in optical networks

Abstract

Optical network optimization involves an algorithm and a Physical Layer Model (PLM) to estimate the Quality of Transmission (QoT) of connections while examining candidate optimization operations. In particular, the algorithm typically calculates intermediate solutions until it reaches the optimum which is then configured to the network. If it uses a PLM that was aligned once to reflect the starting network configuration, then the algorithm within its intermediate calculations can project the network into states where the PLM suffers from low accuracy, resulting in a suboptimal optimization. In this paper, we propose to solve dynamic multivariable optimization problems with an iterative closed control loop process, where after certain algorithm steps we configure the intermediate solution so that we monitor and realign/retrain the PLM to follow the projected network states. The PLM is used as a Digital Twin (DT), a digital representation of the real system which is realigned during the dynamic optimization process. Specifically, we study the dynamic launch power optimization problem, where we have a set of established connections and we optimize their launch powers while the network operates. We observed substantial improvements in the sum and the lowest margin when optimizing the launch powers with the proposed approach over optimization using a one-time trained PLM. The proposed approach achieved near to optimum solutions as found by optimizing and continuously probing and monitoring the network, but with a substantial lower optimization time.

Adaptive and Iterative QoT Estimator Retraining for Launch Power Optimization

“…This problem is known to be convex and polynomial. The convex optimization algorithms, such as (sub)gradient methods, interior point, etc., are iterative; at each iteration they calculate partial derivatives for the objective and constraint functions [9]. If a PLM is used without closed form equations for partial derivatives, a way to calculate them is to use a finite differences method [10].…”

Impact of Multi-Vendor Transponders Performance on Design Margin in Optical Networks

“…Consequently, the NLI can be modeled as an additive Gaussian noise that is statistically independent of signal and ASE noise [21]. The GN model is also a well-accepted PLM for single-and multi-vendor networks [6], [16]- [18], [28].…”

“…To make a multi-vendor PLM, we first identify the parameters of the PLM model, and in particular the , that need to be fitted/trained. In this work, we select the following two sets of parameters to be trained based on the monitored information: i)fiber coefficients-attenuation, non-linear coefficients, dispersion coefficients, and a bias (TP independent) [17], [28]. ii)…”

“…The parameter vector plays a crucial role in attaining good estimation accuracy. In our previous work [17], [28], we trained the main parameters of GN model based PLM, such as fiber attenuation, nonlinear and dispersion coefficient ( ) along with a bias (TP independent). In the following it is shown and discussed how much accuracy can be attained by proper selection of these parameters ( and/or ) in case of a partial disaggregated network scenario where additional uncertainties are present.…”

“…This case is used as a reference case in our work and is inspired from [17] and [28]. In this case, only the fiber coefficients and an additional bias (TP independent) are trained with the LM algorithm to minimize the nonlinear least square fitting error between ( ) and the ground truth .…”