Pervasive Context Sharing in Magpie: Adaptive Trust-Based Privacy Protection

Today’s mobile and pervasive computing devices are embedded with increasingly powerful sensing capabilities that enable them to provide exceptional spatio-temporal context acquisition that is not possible with traditional static sensor networks alone. As a result, enabling these devices to share context information with one another has a great potential for enabling mobile users to exploit the nearby cyber and physical environments in participatory or human-centric computing. However, because these mobile devices are owned by and sense information about individuals, sharing the acquired context raises significant privacy concerns. In this paper, we define Magpie, which implements an alternative to existing all-or-nothing sharing solutions. Magpie integrates a decentralized context-dependent and adaptive trust scheme with a privacy preserving sharing mechanism to evaluate the risk of disclosing potentially private data. The proposed method uses this assessment to dynamically determine the sharing strategy and the quality of the context shared. Conceptually, Magpie allows devices to actively obfuscate context information so that sharing is still useful but does not breach user privacy. To our knowledge this is the first work to take both trust relationships and users’ individual privacy sensitivities into account to balance sharing and privacy preservation. We describe Magpie and then evaluate it in a series of application-oriented experiments running on the Opportunistic Network Environment (ONE) simulator.