Improving Context Recognition In Ubiquitous Computing Environments

Author/s :Zimmer, Tobias, H.
First examiner :Prof. Dr. Beigl, M.
Institute / Publisher :Institut für Betriebssysteme und Rechnerverbund
Faculty :01 - Mathematics and Computer Science
Date :2007-06-19
Documents :
Ediss.pdf ( 5579  kB)    generate ZIP   Details >>
Size :307 S.
Language :german
Abstract :The goal of this work was to gain a significant improvement of context recognition rates of applications in ubiquitous computing environments by systematically restructuring and enhancing them. To reach this goal the concept of "Systematic Context Management" was introduced that forms a new processing layer in context sensitive systems. Systematic and formal definitions of context representation, context processing and context communication are required to build efficient quality management mechanisms. Such a formal model of a ubiquitous computing environment is introduced in this thesis. It represents the basis of a quality management system for context data. The central contribution of this work is a quality management system for contexts on basis of Context Attributes. Context Attributes are general attributes of context data that are independent from the semantically meaning of the context for an application in a ubiquitous computing environment. The presented model defines four Context Attributes: context age, spatial origin of contexts, context reliability and informational interdependence of contexts. For each of these attributes concrete algorithms and methods have been developed that allow for the quantitative assessment of context quality. These algorithms and methods are implemented in a quality-filter that is part of the newly introduced context management layer. With this filter component artefacts gain the ability to dynamically filter contexts on basis of their quality allowing for an improvement of the service quality of those artefacts. As a result of this work, it was possible to prove that by the introduction of the context management system developed in this thesis, the context recognition rate of artefacts in a ubiquitous computing environment can be improved by up to 48%.
Subject :000 Computer science, information and general works
Type :dissertation
Format :text/document
URN:NBN :urn:nbn:de:gbv:084-13297
Permalink :http://www.digibib.tu-bs.de/?docid=00020943