Diploma and Master Theses (authored and supervised):
"A Data Quality Framework for Pervasive and Ubiquitous Systems";
Supervisor: S. Dustdar, F. Li;
Institut für Informationssysteme, AB Verteilte Systeme,
The problem of data quality is inherent to various information systems. Due to abundance of electronic data, effects of data quality are becoming more and more critical. Further problems arise with emerging systems such as ubiquitous computing systems that pose additional requirements regarding the quality of data. The application field of emerging large-scale pervasive and ubiquitous computing systems ranges over multiple domains e.g. healthcare, smart grids, civil service, smart buildings and smart cities.
Due to complexity and diversity of run-time scenarios pervasive systems are deployed in, observing and understanding the environment become very difficult. Additionally, large number of diverse data sources, usually utilized opportunistically, only further complicate the matter. On top of this, diversity and geographical distribution of data sources , near real-time data processing and delivery requirements together with a lack of common schema pose additional challenges on data quality management. Therefore, it becomes crucial to be able to acquire quality information which can be translated into meaningful insights.
The most important work that addressed the problem of data quality in ubiquitous computing systems is Quality of Context (QoC). However, there has been criticism towards QoC that it lacks general and widely adopted framework, which could help context-aware service designers
to understand data quality issues such as specification, evaluation and run-time usage of data quality information. Consequently, the field of data quality management in pervasive systems is still in its infancy. Therefore, many issues remain open and even unclear to the designers and developers of those systems.
In this work we set a corner-stone for development of comprehensive and general data quality framework for large-scale ubiquitous computing systems. Most important concepts of the framework are data quality measurement and assessment mechanism, together with quality
driven data source selection and continuous quality assurance mechanism. We provide a comprehensive analysis of representation, sensing, evaluation and run-time usage of data quality information in pervasive and ubiquitous systems. Finally, the developed data quality framework is evaluated on the real-life data gathered from Cosm IoT platform. In the experiments it is shown that by utilizing proposed data quality metrics and mechanisms the system provides highly stable and responsive environment for client applications. Thus, by mploying the proposed approaches we are able to account for inherently unstable and highly dynamic nature of data sources in ubiquitous computing systems.
Created from the Publication Database of the Vienna University of Technology.