Controlling Smartphone User Privacy via Objective-driven Context Mocking

Smartphones represent the most serious threat to user privacy of any widely-deployed computing technology because these devices are always on and always connected, making them the perfect candidate to know most about the owner. Unfortunately, existing permission models provide smartphone users with limited protection, in part due to the difficulty to users in distinguishing between legitimate and illegitimate use of their data; for example, a mapping app may upload the same location information it uses to download maps (legitimate) to a marketing agency interested in delivering location-based ads (illegitimate). As a result, smartphone users find themselves forced to make burdensome and error-prone tradeoffs between app functionality and privacy. To combat this, we propose a new approach called PocketMocker. By allowing substitution of real data streams with artificial or mocked data, PocketMocker allows users to manipulate impressions of their behavior in well-defined ways, such as appearing more fit, more social, or more on-time than they actually are. Instead of focusing on privacy, we explore providing users with better management of their smartphone-derived digital identities. We discuss the design of PocketMocker, which uses user-initiated context trace recording and replay to enable objective-driven context mocking. Our evaluation shows that users want to use PocketMocker, that PocketMocker can mock popular smartphone apps, and that PocketMocker is usable.