Date of this Version
With the ever-increasing popularity of mobile devices over the last decade, mobile apps and the frameworks upon which they are built frequently change. This rapid evolution leads to a confusing jumble of devices and applications utilizing differing features even within the same framework. For Android apps and devices, representing over 80% of the market share, mismatches between the version of the Android operating system installed on a device and the version of the app installed, can lead to several run-time crashes, providing a poor user experience.
This thesis presents GAINDroid, an analysis approach, backed with a classloader based program analyzer, that automatically detects three types of mismatches to which an app may be vulnerable across versions of the Android API it supports. Unlike all prior techniques that focus on identifying a particular problem, such as callback APIs issues, GAINDroid has the potential to greatly increase the scope of the analysis by automatically and effectively analyzing various sources of incompatibilities that may lead an app to crash at run-time. We applied GAINDroid to 3,590 real-world apps and compared the results of our analysis against state-of-the-art tools. The experimental results demonstrate its ability to outperform the existing analysis techniques in terms of both the number and type of mismatches correctly identified as well as run-time performance of the analysis.
Adviser: Hamid Bagheri