Machine Learning based Multi-Parameter Light Path Quality of Transmission Estimation and Prediction in Optical Networks

In this paper an overview on telemetry of randomly constructed Indian network and existing Japan network, a regression model is powered by machine learning to evaluate and predict the quality of transmission (QoT) of a light path (LP) before it is ever deployed. In addition to enabling the option of automatically deploying the lightpaths with a little margin and in a reliable way, this advance assessment and forecast of the QoT of light Path in networks has a promising metrics like ASE and NLI. Using the well-respected and tried-and-true network simulation programme GNPy, the real field data to be measured is reproduced in the response of the current simulation scenario. Our proposed work uses machine-learning based approaches with modulation schemes such as DP-QPSK, DP-QAM and demonstrated the GNSR prediction using the obtained data set in which machine learning based GRU technique outperforms with a RMSE of 0.104 for Indian topology and 0.13 for Japan topology.