IPDroid: Android Malware Detection using Intents and Permissions
Kartik Khariwal, Jatin Singh, Anshul Arora
Abstract
With the increasing popularity of Android smart-phones over the years, the number of malware attacks on Android has also increased. Around 26 million malware samples were detected on the Android platform in the year 2018 and those samples pose serious threats such as financial loss and information leakage. Hence, stronger security solutions need to be developed to detect such threats. Several static Android malware detection techniques exist in the literature that analyze manifest file components such as permissions or intents. However, to the best of our knowledge, none of them have aimed to find the best set of permissions and intents combined that could give better accuracy. In this paper, we use Information Gain to rank the permissions and intents intending to find the best set of permissions and intents that can detect Android malware with better accuracy. We propose a novel algorithm to find the best set by applying several machine learning algorithms such as Random Forest, SVM, and Naive Bayes. The experimental results demonstrate that the best set consisted of 37 features, i.e., 20 intents and 17 permissions, and Random Forest classifier gave the best accuracy of 94.73%.