Detecting Malicious Behaviors in Repackaged Android Apps with Loosely-Coupled Payloads Filtering Scheme.

Lulu Zhang,Yongzheng Zhang,Tianning Zang
DOI: https://doi.org/10.1007/978-3-319-23829-6_31
2015-01-01
Abstract:Recently, the security problem of Android applications has been increasingly prominent. In this paper, we propose a novel approach to detect malicious behaviors in loosely-coupled repackaged Android apps. We extract and modify the FCG of an app based on its loosely-coupled property, and divide it into several sub-graphs to identify primary module and its related modules. In each remaining sub-graph, API calls are added and sensitive API paths are extracted for dynamic instrumentation on top of APIMonitor. The experiments are conducted with 438 malwares and 1529 apps from two third-party Android markets. Through manual verification, we confirm 5 kinds of malwares in 16 apps detected by our approach. And the detection rate of collected malwares reaches 99.77 %. The reduction rate of monitored functions reaches 42.95 % with 98.79 % of malicious functions being successfully saved. The time spent on static and dynamic analysis is 74.9 s and 16.0 s on average.
What problem does this paper attempt to address?