IG1219 AI Closed Loop Automation – Anomaly Detection and Resolution v2.0.0

The Network and Service Operations of the future cannot continue to operate with the traditional approach, especially with virtualization and the dynamic nature of the virtualized network with VNFs, CNFs and SDN. The rate at which these networks can scale up/down and the exponential growth of the data and events produced by the network both physical and virtual and the associated complexity requires a solution that can detect anomalous patterns in real time, and continue to learn and improve over the lifecycle of the solution. The CSPs need solutions that can automate majority of the operations, reduce human error, improve operational efficiency, and help deliver innovative, uninterrupted superior quality services to their consumers.

In this version of the document, we discuss the various types of Closed-loop anomaly detection methods considered and the rationale for choosing the OODA method, a description of Closed-loop Anomaly Detection and Closed-loop Automation. Sections such as Data Analytics & Modeling, Reference Architecture etc. are work in progress and will continue to be developed during the next sprint of this work group.