Privacy Protection for Perceptual Applications on Smartphones

Abstract

Today's smartphones are equipped with various embedded motion sensors, such as accelerometer, gyroscope, and orientation sensors. Perceptual applications perceive the environment of mobile users via sensors. However, malicious applications may use these sensors to steal user's privacy, and attackers can use sensors as side channel data to infer user's inputs. Existing solutions suffer from limited sensors and overhead problems. In this paper we present Perceptual Assistant (PA), a practical privacy protection system for all the sensors and untrusted perceptual applications. PA allows users to customize the sensor policy of third-party applications, and prevent malicious application from accessing sensors at runtime. We evaluate PA with several typical perceptual applications that perform diverse tasks. PA system shows both practical and lightweight: it can protect user's privacy efficiently while maintaining reasonable overhead.