A probabilistic approach for failure localization

Abstract

This work considers the problem of fault localization in transparent optical networks. The aim is to localize single-link failures by utilizing statistical machine learning techniques trained on data that describe the network state upon current and past failure incidents. In particular, a Gaussian Process (GP) classifier is trained on historical data extracted from the examined network, with the goal of modeling and predicting the failure probability of each link therein. To limit the set of suspect links for every failure incident, the proposed approach is complemented with the utilization of a Graph-Based Correlation heuristic. The proposed approach is tested on a dataset generated for an OFDM-based optical network, demonstrating that it achieves a high localization accuracy. The proposed scheme can be used by service providers for reducing the Mean-Time-To-Repair of the failure.