A Privacy Enhancing Infrastructure for Context-Awareness

Context-awareness enables applications and services to better fulfil the needs of users by adapting to their situation and their preferences. However, the use of contextual information is complicated by privacy concerns. A subjects context is personal and needs to be regarded as sensitive. Hence, contextual information must only be used with the consensus of the subject and according to their privacy preferences. This thesis examines the development of privacy-friendly context-aware systems. In particular the focus is on (A) improving the overall level of privacy, (B) evaluating access control mechanisms, (C) providing development support, and (D) offering protection to third-party infrastructures. The hypothesis investigated is whether these objectives can be achieved through the use of a privacy enhancing infrastructure. As part of the investigation two conceptual models are presented describing the assumptions made about context and privacy. Also presented is a decentralised privacy enhancing infrastructure developed and implemented to determine the validity of the hypothesis. Along with the infrastructure mechanisms for privacy protection including authentication, access control, and anonymity are discussed. A general data format for context communication in the infrastructure is also presented. Finally the thesis presents the findings uncovered during the investigation and evaluation of the hypothesis. This includes a qualitative analysis of whether the privacy enhancing infrastructure meets the key objectives, a user survey examining the performance of two candidate access control mechanism, a performance measure of the infrastructure when run with resource constrained devices, and a comparison with the approaches taken in related work.