[en] The incompleteness of 3rd-party app testing is an accepted fact in Software Engineering. This issue makes it impossible to verify the app functionality and to confirm its safety to the end-user. To solve this problem, enterprises developed strict policies. A company, willing to use modern apps, may perform an expensive security analysis, rely on trust or forbid the app. These strategies may lead companies to high direct and indirect spending with no guarantee of safety.
In this work, we present a novel approach, called Dynamic Binary Shrinking, that allows a user to review app functionality and leave only tested code. The shrunk app produces 100% instruction coverage on observed behaviors and in this way guarantees the absence of unexplored, and therefore, potentially malicious code.
On our running examples, we demonstrate that apps use less than 20% of the codebase. We developed an approach and the ACVCut tool to shrink Android apps towards the executed code.
Repository — http://github.com/pilgun/acvcut.
Name of the research project :
FNR11289380 > Aleksandr Pilgun > > Systematically Exploring Semantic App Models for Android > 15/11/2016 > 14/11/2020 > 2016