Soft Failure Localization Using Machine Learning with SDN-based Network-wide Telemetry
Kayol S. Mayer, Jonathan Aguiar Soares, Rossano P. Pinto, Christian Esteve Rothenberg, Dalton S. Arantes, Darli A. A. Mello
Abstract
Targeting soft failure localization, we present a machine learning approach using SDN streaming telemetry of network-wide parameters. The artificial neural network only requires a few training scenarios to achieve adequate interpolation performance. The framework implementation is validated using gNMI telemetry in an emulated NSFNet topology.
Topics & Concepts
TelemetryComputer scienceNetwork topologyInterpolation (computer graphics)Artificial neural networkComputer networkArtificial intelligenceDistributed computingTopology (electrical circuits)Real-time computingEngineeringTelecommunicationsElectrical engineeringFrame (networking)Advanced Photonic Communication SystemsIntegrated Circuits and Semiconductor Failure AnalysisSoftware-Defined Networks and 5G