Scaling Optical Network Fault Management with Decentralized Graph Learning

We propose a decentralized graph learning framework for scaling cognitive fault management in optical networks. Results show the proposed design achieves > 96% fault identification and localization accuracy.