Network function virtualization deployment scenario dimensioning

Abstract

In this work we analyse the energy impact of commoditization and virtualization of network functions with the focus on Cloud-Radio Access Network (C-RAN). Two RAN functional split options are chosen to study energy variation of different deployment architectures. Furthermore, we demonstrate energy consumption when executing RAN processes on commodity hardware based on mathematical models. As such, we investigate possible energy efficiency with energy optimization techniques like DVFS when executing n number of concurrent tasks on multicore system. As a proof of concept, we perform set of RAN performance evaluation experiments on fully virtualized end-to-end mobile network, implemented using OpenAirInterface (OAI) code. Observations show important results from the energy viewpoint when deploying different functional spit options of RAN VNFs as well possible energy gain without performance degradation.