| Paper ID | D-3-1.3 |
| Paper Title |
A Design Framework of Automatic Deployment for 5G Network Slicing |
| Authors |
Wen-Ping Lai, Hong-Lun Lai, Yuan Ze University, Taiwan; Ming-Jay Lai, National Central University, Taiwan |
| Session |
D-3-1: Digital Convergence of 5G, AIoT and Security II |
| Time | Thursday, 10 December, 12:30 - 14:00 |
| Presentation Time: | Thursday, 10 December, 13:00 - 13:15 Check your Time Zone |
|
All times are in New Zealand Time (UTC +13) |
| Topic |
Wireless Communications and Networking (WCN): Special Session: Digital Convergence of 5G, AIoT and Security |
| Abstract |
Differential services driven user-end and operator-end challenges have been the main driving forces behind the 5G network, which is well perceived as an innovative platform for digital convergence of information, control and management. With both the network slicing (NS) and service slicing (SS) technologies, precious physical resources can thus be shared among multitenant mobile virtual network operators, such as over-the-top (OTT) service providers. This paper proposes a three-stage design for automatic slice deployment called LMA, namely (1) LCP: local charm provision for VNF services, (2) MSP: model-based slice planning for service chaining, and (3) ASD: automatic slice deployment for flexible and virtual resource allocation. The LMA is a model-based slice-specific platform-neutral design framework for deploying NS and SS, not only automatically deployable on both the x86-based desk-top computers and data-center bare-metal servers, but also on public or private clouds, as well as the 5G mobile edge. Our design framework adopts the Juju-as- a-Service and Eurecom Mosaic5G software technologies, where several customizable virtual network function (VNF) components can be flexibly chained together to form a desired NS or SS. This paper studies and presents two successful deployment showcases: a web-blog-database based SS and a virtual-evolved-packet-core based NS. Our preliminary results of performance benchmarking show a strong effect of the number of CPU cores on the average latency response of SS, in particular during congestions caused by con-concurrent user requests. |