Machine Learning-Based Network Status Detection and Fault Localization

Although the autonomous detection of network status and localization faults can be a valuable tool for service operators, very few works have investigated this subject. As result in today's networks, fault remains mostly manual process. In article, we propose machine learning (ML) method that automatically detect localize faults. Our uses decision tree, gradient boosting (GB), extreme GB ML algorithms to as normal, congestion, fault. comparison, existing related work at best classify faulty or nonfaulty. Experimental results show our yields accuracies up 99% on dataset collected through an emulated network.