基于行为的Android恶意软件检测系统

With the number of Android malware growing year by year, many researches on malware detection techniques have been conducted. In those researches, behavior analysis techniques are highly effective on malware detection. In this paper, a behavior-based Android malware detection system is proposed. It uses Android APIs and libc function calls combined with their arguments to describe sensitive behaviors and conducts behavior analysis and malware detection via machine learning techniques. The experiment is conducted with 750 real-world applications which are composed of various types of malware and benign softwares, and three standard metrics are used to evaluate the effect of software classification and malware detection of the system. The experimental result shows that the system can effectively detect Android malware.