Explained Intelligent Management Decisions in Virtual Networks and Network Slices

Abstract

Automated management of virtual networks heavily relies on Artificial Intelligence (AI) to take decisions in response to changes in their requirements. However, since most management solutions are based on Machine Learning (ML) algorithms, they lack the ability to explain the decisions they take. In this paper we propose a solution that exploits Case Based Reasoning (CBR) to take and also explain the required management decisions that are required to optimize the amount of resources assigned to virtual networks and network slices (MVN) in response to the changes in their operational environments. The relations between causes and effects, which are the base for the cases analyzed by the CBR, are formed considering both performance measurements and events occurring outside the boundaries of the MVN, as we are lately encouraging with our research work. We demonstrate by simulation results how the knowledge base is built, we show the explanations that are reasoned and provided to the tenants of the MVN, and validate the feasibility of the CBR by demonstrating that it highly improves the performance of the widely used threshold-based model for resource management.