DroidHook: a novel API-hook based Android malware dynamic analysis sandbox
Yuning Cui,Yi Sun,Zhaowen Lin
DOI: https://doi.org/10.1007/s10515-023-00378-w
IF: 1.677
2023-02-26
Automated Software Engineering
Abstract:With the popularity of Android devices, mobile apps are prevalent in our daily life, making them a target for attackers to steal private data and push advertisements. Dynamic analysis is an effective approach to detect runtime behavior of Android malware and can reduce the impact of code obfuscation. However, some dynamic sandboxes commonly used by researchers are usually based on emulators with older versions of Android, for example, the state-of-the-art sandbox, DroidBox. These sandboxes are vulnerable to evasion attacks and may not work with the latest apps. In this paper, we propose a prototype framework, DroidHook, as a novel automated sandbox for Android malware dynamic analysis. Unlike most existing tools, DroidHook has two obvious advantages. Firstly, the set of APIs to be monitored by DroidHook can be easily modified, so that DroidHook is ideally suitable for diverse situations, including the detection of a specific family of malware and unknown malware. Secondly, DroidHook does not depend on a specific Android OS but only on Xposed, so it can work with multiple Android versions and can perform normally on both emulators and real devices. Experiments show that DroidHook can provide more fine-grained and precise results than DroidBox. Moreover, with the support for real devices and new versions of Android, DroidHook can run most samples properly and acquire stronger detection results, compared to emulator-based tools.
computer science, software engineering