Network Function Virtualization for 5G Network Slicing

Xiao, Da

Network Function Virtualization for 5G Network Slicing

Xiao, Da

Permalink

Publication Type:: Thesis
Issue Date:: 2023

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download thesisAdobe PDF (3.64 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Xiao, Da
dc.date.accessioned	2023-09-27T00:20:22Z
dc.date.available	2023-09-27T00:20:22Z
dc.date.issued	2023
dc.identifier.uri	http://hdl.handle.net/10453/172301
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	Network slicing is a core technique of the 5G system and beyond. To meet the varied industrial demands, the International Telecommunication Union (ITU) has classified the current 5G networks into three main categories: Ultra-Reliable Low Latency Communications (URLLC), enhanced Mobile broadband (eMBB), and massive Machine-Type Communications (mMTC). Packets belonging to the same category will be aggregated and then travel through the corresponding network slice, which can be symbolized as a service function chain (SFC). An SFC consists of a sequence of Virtual Network Functions (VNFs) and Virtual Links (VLs) connecting them. To accommodate as many SFCs as possible with limited hardware resources, service providers need to refrain from over-provisioning resources, which may hinder them from cutting capital expenditures (CAPEX)/operating expenses (OPEX) for 5G infrastructure. Therefore, efficient and automatic placement of SFCs becomes one of the most critical technologies for meeting such requirements. This thesis aims to develop efficient SFC embedding algorithms in three scenarios: single data center (DC), multiple DCs (MDC), and fog networks. We aim to maximize the acceptance ratio of SFC requests while minimizing the cost in all scenarios. In the first scenario, we developed two algorithms for SFC embedding. In the first proposed algorithm, under the classical virtual network embedding model, we designed a binary search-assisted transfer learning algorithm to embed SFCs in an ever-changing environment. In the second proposed algorithm, under a model where we can flexibly allocate resources to virtual elements, we designed a sub-action-aided reinforcement learning (RL) algorithm to embed SFCs and allocate resources to VNFs and VLs in SFCs. In the second scenario, to resolve the unbalanced load issue in MDC networks, we proposed a two-stage Graph Convolutional Networks (GCN)-based RL framework for the placement of SFCs in an MDC scenario, where the requests load may vary from DC to DC. In the third scenario, we designed a two-agent RL algorithm for SFC embedding in a fog environment, where the SFC requests load may vary from location to location and some requests have physical isolation requirements. The proposed algorithm features a two-agent RL model evolved from the typical multi-agent RL model. This proposed model can be applied to similar scenarios where the action includes two steps and the second sub-action depends on the first. Numerical evaluations show that, compared to state-of-the-art methods, our proposed schemes can improve the overall acceptance ratio of SFC requests and cost-effectiveness in the aforementioned scenarios.	en_US.UTF-8
dc.format	Thesis (PhD)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/172301/1/thesis.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	© 2023 Da Xiao
dc.rights	au.edu.uts.lib/cph
dc.title	Network Function Virtualization for 5G Network Slicing	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

Network slicing is a core technique of the 5G system and beyond. To meet the varied industrial demands, the International Telecommunication Union (ITU) has classified the current 5G networks into three main categories: Ultra-Reliable Low Latency Communications (URLLC), enhanced Mobile broadband (eMBB), and massive Machine-Type Communications (mMTC). Packets belonging to the same category will be aggregated and then travel through the corresponding network slice, which can be symbolized as a service function chain (SFC). An SFC consists of a sequence of Virtual Network Functions (VNFs) and Virtual Links (VLs) connecting them. To accommodate as many SFCs as possible with limited hardware resources, service providers need to refrain from over-provisioning resources, which may hinder them from cutting capital expenditures (CAPEX)/operating expenses (OPEX) for 5G infrastructure. Therefore, efficient and automatic placement of SFCs becomes one of the most critical technologies for meeting such requirements. This thesis aims to develop efficient SFC embedding algorithms in three scenarios: single data center (DC), multiple DCs (MDC), and fog networks. We aim to maximize the acceptance ratio of SFC requests while minimizing the cost in all scenarios. In the first scenario, we developed two algorithms for SFC embedding. In the first proposed algorithm, under the classical virtual network embedding model, we designed a binary search-assisted transfer learning algorithm to embed SFCs in an ever-changing environment. In the second proposed algorithm, under a model where we can flexibly allocate resources to virtual elements, we designed a sub-action-aided reinforcement learning (RL) algorithm to embed SFCs and allocate resources to VNFs and VLs in SFCs. In the second scenario, to resolve the unbalanced load issue in MDC networks, we proposed a two-stage Graph Convolutional Networks (GCN)-based RL framework for the placement of SFCs in an MDC scenario, where the requests load may vary from DC to DC. In the third scenario, we designed a two-agent RL algorithm for SFC embedding in a fog environment, where the SFC requests load may vary from location to location and some requests have physical isolation requirements. The proposed algorithm features a two-agent RL model evolved from the typical multi-agent RL model. This proposed model can be applied to similar scenarios where the action includes two steps and the second sub-action depends on the first. Numerical evaluations show that, compared to state-of-the-art methods, our proposed schemes can improve the overall acceptance ratio of SFC requests and cost-effectiveness in the aforementioned scenarios.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/172301