A new framework based on Software Defined Networks to support Quality of Service in a sliced architecture

Al-Haddad, Ronak (2021) A new framework based on Software Defined Networks to support Quality of Service in a sliced architecture. Doctoral thesis, Anglia Ruskin University.

[img]
Preview
Text
Accepted Version
Available under the following license: Creative Commons Attribution Non-commercial No Derivatives.

Download (3MB) | Preview
[img] Text (Word version)
Accepted Version
Available under the following license: Creative Commons Attribution Non-commercial No Derivatives.

Download (7MB)

Abstract

This study focuses on the challenges regarding traffic performance issues, including bottlenecks and congestion in network traffic communication in Software Defined Networks (SDN). These challenges are caused by the increasing demand for network services and quality across a wide range of digital applications on the internet. The study is aimed at improving network throughput as well as reducing end-to-end delay and jitter for the packet transmission process. A framework design is proposed that involves applying four main technologies: Traffic engineering (TE), SDN, network hypervisors and network slicing. Network function virtualisation (NFV) pertains to the incorporation of comprehensive automation and centralised functions at SDN’s core by combining physical and virtual systems using virtual machines (VMs), which enhances network productivity, reliability, scalability, and integrity. In this work, a new methodology of QoSVisor and a Packet Tagging Prioritisation (PTP) Agent extension algorithm for video, audio and data over TCP SDN-slicing networks has been developed and tested. QoSVisor works via a local weighting function, which sorts packets by the port and flow policy setup, depending on each packet (flow, packet ID and weighted tag), with strict priority policy added to this algorithm to guarantee more precise quality of service (QoS) and low latency queueing (LLQ) for video and audio traffic types. Furthermore, a new traffic shaping (TS) algorithm is proposed as a new implementation of (QoS) to work as a bandwidth management technique for optimising performance in an SDN-sliced network, Two algorithms, namely “packet tagging, queueing, forwarding to queues” and “allocating bandwidth”, are proposed and developed for implementing a weighted fair queuing (WFQ) technique, which works mainly as a part-function of TS the queueing mechanism, to reduce congestion and smooth traffic flow. Additionally, an ordinary queueing algorithm, First In First Out (FIFO), has been developed and implemented in an SDN-sliced framework as a baseline condition for quantitative performance comparisons to evaluate the network characteristic behaviour of SDN and QoS. The novel comparative approach for evaluating the implemented algorithms shows significant improvement on traffic performance issues. The result show that the more advanced algorithms TS and QoSVisor deliver more effective allocation of bandwidth than FIFO, and that they significantly reduce critical delays. The results show throughput for TS and QoSVisor is similarly high for the two most demanding flows, namely video and audio and so, traffic types were prioritised. Giving lower priority for audio followed by data, it is demonstrated that TS lets through just over twice the amount of information compared to QoSVisor. Comparing audio with data flow there was an approximately five-fold advantage for the former compared to the latter data flow type.

Item Type: Thesis (Doctoral)
Additional Information: Accessibility note: If you require a more accessible version of this thesis, please contact us at arro@aru.ac.uk
Keywords: Network congestion, network performance, SDN, slicing, QoS, FIFO, weighted fair queuing (WFQ), QoSVisor algorithm, Packet Tagging Prioritisation (PTP) Agent, OpenFlow, SPSS analysis of variance (ANOVA)., network performance, SDN, slicing, QoS, FIFO, weighted fair queuing (WFQ), QoSVisor algorithm, Packet Tagging Prioritisation (PTP) Agent, OpenFlow, SPSS analysis of variance (ANOVA)
Faculty: Theses from Anglia Ruskin University
Depositing User: Lisa Blanshard
Date Deposited: 03 Mar 2022 16:17
Last Modified: 31 May 2022 16:18
URI: https://arro.anglia.ac.uk/id/eprint/707368

Actions (login required)

Edit Item Edit Item