Network Anomaly Event Prediction and Optimal Resource Control in Cloud Native Network Functions

Explore network anomaly event prediction and optimal resource control in Cloud Native Network Functions (CNFs) through this conference talk. Gain insights into leveraging CNFs for telco services while meeting high availability and resource allocation requirements. Learn about eBPF observability in 5G core networks deployed with Kubernetes, and discover how detailed container metrics enable future anomaly detection using Deep Learning. Delve into autonomous computational resource control systems compliant with ETSI Open Source MANO and ITU-T Y.3177 standards. Examine the deployment of control systems in the Japanese public network testbed JGN. Understand the framework for resource operations, verification infrastructure, and benefits of direct interfaces in XNF operation systems.

Syllabus

Intro
Example of network anomaly events
Overview of our approach
Per-container detailed system metrics collected by eBPF
Summary - Optimal Resource Control for CNFS
Framework of Resource Operations for XNES
Verification infrastructure on a public network testbed (GN)
ETSI OSM based XNF operation system
Benefits of Direct Interface and components
General Conclusion