A Learning Life-Cycle to Speed-up Autonomic Optical Transmission and Networking Adoption

Autonomic Optical Transmission and Networking requires from Machine Learning (ML) models to be trained with large datasets. However, the availability of enough real data to produce accurate ML models is rarely ensured since new optical equipment, techniques, etc., are continuously being deployed in the network. Although an option is to generate data from simulation and lab experiments, such data could not cover the whole features space, which would translate into ML models inaccuracies. In this paper, we propose a ML-based algorithm life-cycle to facilitate ML deployment in real networks. Dataset for ML training can be initially populated based on the results from simulation and lab experiments and once ML models are generated, ML re-training can be performed after inaccuracies are detected to improve their precision. Illustrative numerical results show the benefits from the proposed learning cycle for general use cases. In addition, two specific use cases are proposed and demonstrated implementing different learning strategies: i) a two-phase strategy performing out-of-field training using data from simulation and lab experiments with generic equipment is followed by an in-field adaptation to support heterogeneous equipment. The accuracy of this strategy is shown for a use case of failure detection and identification; and ii) in-field re-training, where ML models are re-trained after detecting model inaccuracies. Different approaches are analyzed and evaluated for a use case of autonomic transmission, where results show large benefits of collective learning.