Traffic Prediction in Optical Networks Using Graph Convolutional Generative Adversarial Networks
Connor Vinchoff, Nathan Chung, Tyler Gordon, Liam Lyford, Michał Aibin
Abstract
In this paper, we use a non-linear GCN-GAN model to predict burst events in the optical network. We model three distinct burst events as Plateau, Single-Burst and Double-Burst. Plateau represents the network under steady traffic, Single-Burst represents the network experiencing a rapid traffic spike followed by a steady decrease, and Double-Burst represents the network experiencing a rapid traffic spike followed by an unexpected greater traffic spike. We verify the model's effectiveness to predict these burst events in the real optical networks by comparing it to a basic LSTM, which has been shown to outperform other state-of-the-art models.
Topics & Concepts
Computer scienceSpike (software development)Plateau (mathematics)GraphOptical burst switchingArtificial intelligenceTheoretical computer scienceOptical performance monitoringPhysicsMathematicsWavelength-division multiplexingOptoelectronicsSoftware engineeringWavelengthMathematical analysisAdvanced Optical Network TechnologiesSoftware-Defined Networks and 5GOptical Network Technologies