Context-centric security

摘要: Users today are unable to use the rich collection of third-party untrusted applications without risking significant privacy leaks. In this paper, we argue that current and proposed data-centric security policies do not map well users' expectations privacy. eyes a user, peripheral devices exist merely provide functionality should have no place in controlling Moreover, most users cannot handle intricate dealing with system concepts such as labeling data, application permissions virtual machines. Not only impenetrable users, they also lead problems privilege-escalation attacks implicit information leaks. Our key insight is naturally associate data real-world events, want control access at level human contacts. We introduce Bubbles, context-centric explicitly captures user's desires by allowing contact lists clustered events. Bubbles infers information-flow rules from these simple access-control enable secure on data. We new programming model for allows them be functional while still upholding policies. evaluate model's usability porting an existing medical writing calendar app scratch. Finally, show design our prototype running Android uses bubbles automatically infer all dangerous any user intervention. prevents Android-style permission escalation requiring specify complex flow rules.