Al-assisted workflow management framework for automated closed-loop operation

One of the key issues today's confronting telecom operator is how to realize automated operation for cost saving. Network Function Virtualization (NFV) gives us two powerful enablers for the closed-loop automation such as softwarization of network function and simplification of operation tasks thanks to virtualization technology. However, maintenance of workflows still requires tremendous efforts to sustain the automation function. In reality, huge number of workflow or script for automated deployment and fault recovery process has to be created and maintained per type of service and failure. It is hard to manually keep not only the workflow but also the automation function itself, and result in an increase in the operational cost. To tackle this issue, a sustainable workflow management framework by applying deep neural network (DNN) is proposed. In this framework, the demonstration result shows that proposed system enable to automate creation and selection of fault recovery operation workflow. In addition, we evaluate the accuracy of creation workflow, and show effectiveness of adapting DNN for sustainable workflow management.