Androknight: lightweight android malware detection for low-resource mobile devices

Abstract

Our aim is to develop a light weight learning based android malware detection solution, that can not only accurately distinguish malware from goodware but also classify / report its category, even when functioning under resource-constrained computing infrastructures. into particular category of malware such as trojan, adware or ransomware etc. The model will take static as well as dynamic features as input and perform classification. In order to extract both types of features, we will also create a tool which will take the APK file and extract static features before actually installing and running it on an android phone.