Multi-App Security Analysis with FUSE: Statically Detecting Android App Collusion

Abstract

Android's popularity has given rise to myriad application analysis techniques to improve the security and robustness of mobile applications, motivated by the evolving adversarial landscape. These techniques have focused on identifying undesirable behaviors in individual applications, either due to malicious intent or programmer error. We present a collection of tools that provide a static information flow analysis across a set of applications, showing a holistic view of all the applications destined for a particular device. The techniques we present include a static binary single-app analysis, a security lint tool to mitigate the limits of static binary analysis, a multi-app information flow analysis, and an evaluation engine to detect information flows that violate specified security policies. We show that our single-app analysis is comparable with the leading approaches on the DroidBench benchmark suite; we present a brief listing of lint-like heuristics used to show the limits of the single-app analysis in the context of an application; we present a multi-app analysis, and demonstrate information flows that cannot be detected by single-app analyses; and we present a policy evaluation engine to automatically detect violations in collections of Android apps.