First Demonstration of Imbalanced Data Learning-Based Failure Prediction in Self-Optimizing Optical Networks with Large Scale Field Topology

Machine learning is a promising solution to address some issues in large scale optical networks. Failure prediction is a typical application which can be completed by the aid of machine learning. However, the constraint of massive imbalanced data distribution limits the performance of machine learning in large scale optical networks. The paper first demonstrate the imbalanced data learning based failure prediction in self-optimizing optical networks (SOON). The large scale field topology with 274 nodes and 487 links is considered. Experimental results show that the proposed imbalanced data learning based failure prediction solution can get high accuracy effectively.