Predicting loss in optical transport segments: A GNN-GRU approach for a nationwide optical network

Optical networks play a crucial role in providing high-speed Internet and communication services. The performance and reliability of these networks are critical to ensuring the quality of the services provided to customers. Hence, predicting and preventing potential downtime in optical networks is of utmost importance. In this paper, we present a novel approach for predicting loss in optical networks.Our strategy combines Graph Neural Networks (GNNs) and Gated Recurrent Units (GRUs) to capture both spatial and temporal features of the optical network. By utilizing the GNN to capture spatial data and the GRU to detect temporal relationships, this methodology can predict the location of future problematic links and simulate the effects of traffic redistribution. We gathered and analyzed real data from a nationwide Internet Service Provider optical network to train and test the proposed algorithm.By exploiting the satisfactory results obtained from our methodology, network operators can proactively take measures to improve the network's performance and avoid potential downtime.