Abstract

Modern societies have become very dependent on information and services. Technology is adapting to the increasing demands of people and businesses. Context-Aware Systems are becoming ubiquitous. These systems comprise mechanisms to acquire knowledge about the surrounding environment and adapt its behaviour and service provision accordingly. Service oriented computing is the main stream software development methodology. In Service-oriented Applications (SOA), service providers publish the services created by them in service registries. These services are accessed by service requesters during discovery process. For large scale SOA, the registry structure and the type of quires that it can handle are central to efficient service discovery. Moreover, the role of context in determining services and affecting execution is central. This thesis investigates the structure of a context-aware service registry in which context-aware services are stored by service producers and retrieved by service requesters in different contexts. The thesis builds on an existing rich theoretical service model in which contract, functionality, and contexts are bundled together. The thesis investigates generic models and structures for context, context history, and context-aware registry. Also, it studies state of the arts database technologies to analyse its suitability for implementing a registry for rich services. Specifically, the thesis provides a thorough study of the structures, implementation, performance, limitations, and features of Key-Value, Documented Oriented, and Column Oriented databases while considering options for implementing a rich service registry. Database models of contexts and context-aware services are discussed and implemented. The relative performance of the models are discussed after evaluating the test results run on large data sets. Based upon test results a justification for the selected model is given.