SliceNetVSwitch: definition, design and implementation of 5G multi-tenant network slicing in software data paths

Antonio Matencio-Escolar, Qi Wang, Jose M. Alcaraz Calero*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

17 Citations (Scopus)
93 Downloads (Pure)

Abstract

Network slicing is a primary Fifth-Generation (5G)mobile networking technology to create virtualised and soft-warised logical networks for various vertical businesses withdiverging Quality of Service (QoS) requirements. Meanwhile,there is a clear gap in providing network slicing capabilitiesin 5G multi-tenant networks to enable guaranteed QoS in termsof well-defined network metrics for the multiple tenants sharingthe same physical infrastructure. This paper designs and imple-ments novel software data path architecture that enables suchnetwork slicing with assured QoS in 5G multi-tenant networks.Highly flexible and customisable definition of network slicingis also allowed to be aligned with different existing definitionson demand and at run time. The proposed architecture hasbeen prototyped based on the popular Open Virtual Switch(OVS), and empirically validated to demonstrate the deploymentand management of network slices with the above capabilities.Intensive scalability results are provided where more than 8,192network slices are achieved simultaneously with warranted QoSthrough performance isolation in terms of bandwidth and delayin a real softwarised 5G multi-tenant infrastructure at speeds ofup to 10 Gbps
Original languageEnglish
Pages (from-to)2212-2225
Number of pages14
JournalIEEE Transactions on Network and Service Management
Volume17
Issue number4
Early online date8 Oct 2020
DOIs
Publication statusPublished - 9 Dec 2020

Keywords

  • 5G
  • network slicing
  • network management
  • network control

Fingerprint

Dive into the research topics of 'SliceNetVSwitch: definition, design and implementation of 5G multi-tenant network slicing in software data paths'. Together they form a unique fingerprint.

Cite this