CORRDroid - Android Malware Detection using Association amongst Permissions

The introduction of Apple’s iPhone in 2007 marked the beginning of a new era for mobile devices and applications. In 2012, more Android smartphones were sold than iPhones, and since then, Android smartphones have become increasingly popular. Android’s ubiquity has attracted the attention of attackers and attacks against the platform are on the rise. Numerous malware applications target mobile devices and compromise sensitive and private information stored on them. Hence, stronger security solutions need to be developed to detect such threats. In the literature, there is a variety of static Android malware detection techniques based on the analysis of manifest file components such as permissions. To the best of our knowledge, none of these attempts have aimed to determine the most distinguishing permission pair based on the association of permissions that could result in greater accuracy. In this study, we propose a risky permission-based malware detection system that uses association between permissions to create a set of important permissions and then uses supervised learning techniques to effectively classify malicious and benign applications. The experimental results demonstrate that the SVM gave the best accuracy of 97.2%.