Privacy-Preserved Mobile Sensing through Hybrid Cloud Trust Framework

Mobile sensors embedded in smart phones and smart buildings enable mobile sensing and users' behavior modeling and thus open the doors for edge-cutting applications such as personalized intelligent computing, activity prediction, health/wellbeing monitoring and behavior intervention. One critical obstacle in mobile sensing and behavior modeling is privacy. Users do not always trust the public cloud to store and process their detailed personal data. In this paper, we propose a novel approach of using hybrid cloud to distribute the computing among mobile devices, personal cloud and public cloud. Raw sensor data is stored within the personal cloud where users have full, physical control. User can authorize analytic widgets (e.g., health monitor or marketing survey) to only collect user-approved data. We demonstrate that with this approach, users' privacy anxiety is significantly reduced and the acceptance rate of the mobile sensing technology increases from 23% to 60%.