SSG: Sensor security guard for android smartphones

Abstract

The smartphone sensors provide extraordinary user experience in various Android apps, e.g. sport apps, gravity sensing games. Recent works have been proposed to launch powerful sensor-based attacks such as location tracing and sound eavesdropping. The use of sensors does not require any permission in Android apps, so these attacks are very difficult to be noticed by the app users. Furthermore, the combination of various kinds of sensors generates numerous types of attacks which are hard to be systematically studied. To better address the attacks, we have developed a taxonomy on sensor-based attacks from five aspects. In this work, we propose a sensor API hooking and information filtering framework, Sensor Security Guard (SSG). Unlike any rough hooking framework, this system provides finegrained processing for different security levels set by the users, or by default. The sensor data is blocked, forged or processed under different mode strategies and then returned to the apps. In addition, according to the taxonomy, SSG develops fine-grained corresponding countermeasures. We evaluate the usability of SSG on 30 popular apps chosen from Google Market. SSG does not cause any crash of either the Android system or the apps while working. The result indicated that SSG could significantly preserve the users’ privacy with acceptable energy lost.