White Rose University Consortium logo
University of Leeds logo University of Sheffield logo York University logo

Service Level Agreement-based adaptation management for Internet Service Provider (ISP) using Fuzzy Q-learning

Bin Ramli, Ahmad Kamal (2018) Service Level Agreement-based adaptation management for Internet Service Provider (ISP) using Fuzzy Q-learning. PhD thesis, University of Leeds.

BinRamli_AhmadKamal_Computing_PhD_2018.pdf - Final eThesis - complete (pdf)
Available under License Creative Commons Attribution-Noncommercial-Share Alike 2.0 UK: England & Wales.

Download (5Mb) | Preview


Internet access is the vital catalyst for online users, and the number of mobile subscribers is predicted to grow from dramatically in the next few years. This huge demand is the main issue facing the Internet Service Providers (ISPs) who need to handle users’ expectations along with their current resources. An adaptive mechanism within the ISPs architecture is a promising solution to handle such situation. A Service Level Agreement (SLA)is the legal catalyst to monitor any contract violation between end users and ISPs and is embedded within a Quality of Service (QoS) framework. It strengthens and advances the quality of control over the user’s application and network resources and can be further stretched to fulfill the QoS terms through negotiation and re-negotiation. Moreover, the present literature does not focus on the combination of rule-based approaches and adaptation together to update the established learning repository. Therefore, this mainstream of this research in the context of SLAs is to fill in this gap by addressing the combination of rule-base uncertainties and iteration of the learning ability. The key to the proposed architecture is the utilization of self - * capabilities designed to have self-management over uncertainties and the provision of self-adaptive interactions. Thus, the Monitor, Analyse, Plan, Execute and Knowledge Base (MAPE-K) approach is able to deal with this problem together with the integration of Fuzzy and Q-Learning algorithms. The proposed architecture is in the context of autonomic computing. An adaptation manager is the main proposed component to update admission control on the ISP current resources and the ability to manage SLAs. A general methodology type-2 fuzzy logic is applied to ensure the uncertainties and precise decision-making are well addressed in this research. The proposed solution, demonstrating Q-Learning works adaptive with QoS parameters, e.g. Latency, Availability and Packet Loss. With the combination of fuzzy and Q-Learning, we demonstrate that the proposed adaptation manager is able to handle the uncertainties and learning abilities. Q-Learning is able to identify the initial state from various ISPs iterations and update them with appropriate actions, reflecting the reward configurations. The higher the iterations process the higher is the increase the learning ability,rewards and exploration probability. The research outcomes benefit the SLA framework by incorporating the information for SLA policies and Service Level Objectives (SLOs). Lastly, an important contribution is the ability to demonstrate that the MAPE-K approach is a contender for ISP SLA-based frameworks for QoS provision.

Item Type: Thesis (PhD)
Keywords: Autonomic Computing , Adaptive Environments , Fuzzy , Reinforcement Learning , Internet Service Provider (ISP), Service Level Agreement (SLA) , Quality of Service (QoS)
Academic Units: The University of Leeds > Faculty of Engineering (Leeds)
The University of Leeds > Faculty of Engineering (Leeds) > School of Computing (Leeds)
Identification Number/EthosID: uk.bl.ethos.773981
Depositing User: Mr Ahmad Kamal Bin Ramli
Date Deposited: 15 May 2019 11:39
Last Modified: 18 Feb 2020 12:50
URI: http://etheses.whiterose.ac.uk/id/eprint/23913

You do not need to contact us to get a copy of this thesis. Please use the 'Download' link(s) above to get a copy.
You can contact us about this thesis. If you need to make a general enquiry, please see the Contact us page.

Actions (repository staff only: login required)