Privacy of Mobile Users in Contextaware Computing Environments

Abstract: This thesis provides a solution to address the difficulty of design and development of privacy-sensitive context-aware applications that can be used in peoples‟ everyday life without the concerns regarding potential abuse. Users would like to be able to control who can access their contextual information, with what granularity, and in which situations. These users‟ privacy preferences must beconsidered when distributing context information in a context-aware environment. The current privacy policy languages have not been tailored to the specific requirements of context-aware applications, because they are mostly used to create static rules for personal information that rarely changes, while the situation of mobile users in a context-aware environment frequently changes and their privacy preferences must be updated accordingly. The thesis introduces Context Privacy Policy Language (CPPL) as a context-aware language that maps different situations of the context owner to a set of privacy rules that must be applied in the corresponding situation. Self-Adapting Applications for Mobile Users in Ubiquitous Computing Environment (MUSIC) provides an open technology platform that makes it technically and commercially feasible for the wider IT industry to develop innovative mobile applications which are context-aware, self-adapting, and inherently distributed. This thesis investigates, designs, implements, and evaluates a privacy management system based on CPPL, that is integrated to the context distribution architecture of the MUSIC project. The evaluation of designed context privacy management system revealed that the MUSIC Context distribution service is highly scalable and can respond to hundreds of subscriptions per second and enabling the privacy adds a delay of 131μs (4 %) to the average subscription response time (3,263ms). Moreover, the scalability tests showed that the context-aware privacy management system is highly scalable in a sense that increasing the number of privacy rules from 1 to 1000 has a slight affect on performance of the system by increasing the context publication time from 284,55 ms to 287,34 ms (0,9%)