Research Article
ReHAna: An Efficient Program Analysis Framework to Uncover Reflective Code in Android
@INPROCEEDINGS{10.1007/978-3-030-94822-1_19, author={Shakthi Bachala and Yutaka Tsutano and Witawas Srisa-an and Gregg Rothermel and Jackson Dinh and Yuanjiu Hu}, title={ReHAna: An Efficient Program Analysis Framework to Uncover Reflective Code in Android}, proceedings={Mobile and Ubiquitous Systems: Computing, Networking and Services. 18th EAI International Conference, MobiQuitous 2021, Virtual Event, November 8-11, 2021, Proceedings}, proceedings_a={MOBIQUITOUS}, year={2022}, month={2}, keywords={Program analysis Android Java reflection}, doi={10.1007/978-3-030-94822-1_19} }- Shakthi Bachala
Yutaka Tsutano
Witawas Srisa-an
Gregg Rothermel
Jackson Dinh
Yuanjiu Hu
Year: 2022
ReHAna: An Efficient Program Analysis Framework to Uncover Reflective Code in Android
MOBIQUITOUS
Springer
DOI: 10.1007/978-3-030-94822-1_19
Abstract
The recent adoption of dynamic features such as Java reflection and Android dynamic code downloading (RDCL) coupled with recent security attacks that can be detected only at runtime have led to higher usage of hybrid analysis to address dependability and security concerns. While effective, however, hybrid analysis can be inefficient due to a multi-step process involving static analysis, code instrumentation, and runtime information logging. As such, existing hybrid analysis techniques can work during code development and testing, but are too slow for production and security vetting.
In this paper, we introduceReHAna, a hybrid analysis framework for Android apps. We designed our framework to perform hybrid analysis efficiently through the use of a Virtual Class-Loader (VCL), which enables incremental program analysis. We then conducted a study to assess the program analysis performance of using VCL and found that it yields several benefits over the existing compiler-based program analysis approach. We also illustrated the hybrid analysis capability ofReHAnaby implementing a technique to detect and analyze dynamically loaded components based on reflection and dynamic code loading mechanisms in Android apps. We compared the performance ofReHAnaagainst that ofStaDynA, a hybrid analysis approach that performs the same task. Our empirical evaluation shows thatReHAnais as effective asStaDynAbut also significantly more efficient and scalable.
