Mining API Calls and Permissions for Android Malware Detection

The popularity of Android platform is increasing very sharply due to the large market share of Android and openness in nature. The increased popularity is making Android an enticing target for malwares. A worrying trend that is alarming is the increasing sophistication of Android malware to evade detection by traditional signature based scanners. Several approaches have been proposed in literature for Android malware detection. However, most of them are less effective in terms of true positive rate and involves computational overheads. In this paper, we propose an effective approach to attenuate the problem of Android malware detection using static code analysis based models. The proposed models, in this paper, are built to capture features relevant to malware behaviour based on API calls as well as permissions present in various Android applications. Thereafter, models are evaluated using Naive Bayesian as well as K-Nearest Neighbour classifiers. Proposed models are able to detect real malwares in the wild and achieve an accuracy of 95.1% and true positive rate with highest value one.