QoS and trust prediction framework for composed distributed systems

<p>The objective of this dissertation is to propose a comprehensive framework to predict the QoS and trust (i.e, the degree of compliance of a service to its specification) values of composed distributed systems created out of existing quality-aware services. We improve the accuracy of the predictions by building context-aware models and validating them with real-life case studies. The context is the set of environmental factors that affect QoS attributes (such as response time and availability), and trust of a service or a composed system. The proposed framework uses available context-QoS dependency information of individual services and information about the interaction patterns among the services to make predictions for the QoS and trust values of the composed system at the design phase of the development lifecycle. Such predictions made in the early phases of the system development lifecycle will reduce cost, time, and effort. We demonstrate the use of these predictions in selecting the optimum set of services to create composed systems using heuristic optimization algorithms. Additionally, the prediction model is used at runtime with fast heuristic techniques to build adaptable composed systems. The empirical results show the proposed context-dependent framework performs well in providing more accurate predictions than the prevalent approaches.</p>

Name thesis:degree_name

Doctor of Philosophy (PhD)

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Year

2016

Author dc:creator

• Gamage, Dimuthu Undupitiya

Contributors dc:contributor

• Rajeev R. Raje
• Antony Hosking
• Murat Dundar
• Zhiyuan Li
• Mihran Tuceryan

• Applied sciences
• Composition
• Context
• Predictions
• Quality of service
• Trust
• Computer Sciences

Repository record dc:identifier

https://docs.lib.purdue.edu/open_access_dissertations/1021

OAI identifier oai:identifier

oai:docs.lib.purdue.edu:open_access_dissertations-2236