Deep learning for multi-step performance prediction in operational optical networks
Deep learning for multi-step performance prediction in operational optical networks
Mezni, Ameni and Charlton, Douglas W. and Tremblay, Christine and Desrosiers, Christian
Optics InfoBase Conference Papers 2020
Abstract : We propose a 1D Convolutional Neural Network to predict the performance of a lightpath using field bit error rate data and we evaluate the model robustness over forecast horizons up to 24 hours.