Analyzing, Comparing, and Detecting Emerging Malware: A Graph-based Approach

• DOI: 10.48550/arxiv.1902.03955
• Publication Date: 2019-01-01
• Authors: Hisham Alasmary; Aminollah Khormali; Afsah Anwar; Jeman Park; Jinchun Choi; DaeHun Nyang; Aziz Mohaisen
• Keywords: Malware analysis

The growth in the number of Android and Internet Things (IoT) devices has witnessed a parallel increase malicious software (malware), calling for new analysis approaches. We represent binaries using their graph properties Control Flow Graph (CFG) structure conduct an in-depth graphs extracted from IoT malware to understand differences. Using 2,874 2,891 corresponding samples, we analyze both general characteristics algorithmic properties. CFG as abstract structure, then emphasize various interesting findings, such prevalence unreachable code malware, noted by multiple components CFGs, larger nodes compared highlighting higher order complexity. implement Machine Learning based classifiers detect benign ones, achieved accuracy 97.9% Random Forests (RF).