Proactive Network Slices Management Algorithm Based on Fuzzy Logic System and Support Vector Regression Model

Abstract

Software Defined Networks (SDN), Network Function Virtualization (NFV) and Network Slicing are the key technologies for future network implementation. Their aggregation allows more flexibility for the networks by provisioning network slices according to specific use cases requirements. However, in order to ensure these requirements during all the slice execution time, a management module has to be implemented. In this paper, we present our considered architecture for the management of network slices. We detail especially the network controller components. Moreover, we propose a proactive dynamic approach which forecasts the future workload behavior of network slices. Based on the actual and predicted load state, the management algorithm, which is based on a fuzzy logic system (FLS), will determine the adequate management decision for the deployed slices. Based on real network traces, an evaluation of the efficiency of our algorithm is presented.