Detect Resource Related Events in a Cloud-Edge Infrastructure using Knowledge Graph Embeddings and Machine Learning

Abstract

Edge and cloud computing infrastructures consist of multiple resources that may belong to different providers and are utilized in a shared manner by distributed applications for computing and storage purposes. Detecting events that affect the efficient operation of such infrastructures is a challenge and absolutely necessary for providing high quality cloud-edge services. In this work, we model cloud-edge infrastructures using knowledge graphs and use graph embeddings to transform the graphs into vectors. Then, traditional data-driven machine learning algorithms are used in order to detect anomaly events that relate to the infrastructure usage.