Malicious Adware Detection on Android Platform using Dynamic Random Forest

As smartphones have evolved into more sophisticated devices that can be used anytime and anywhere and most of the smartphone users save and use all their information on their smartphones, they have become major attack targets by hackers. Hackers can steal personal information through malicious code installed on smartphones and gain financial benefits through micropayment systems and advertising services. The number of mobile malware has been exponentially increasing, along with the benefits that can be gained from smartphones infected with malware. In particular, the openness of the Android market and the high market share of Android devices allow malicious code to easily spread, therefore most of the mobile malicious code is targeting Android devices for attacks. In this paper, we investigate the characteristics of adware that is the fast-growing Android-based mobile malicious code, propose a learning algorithm for detecting malicious adware attacks, and present attack detection rates.