MULTIPLE ASPECTS OF DEX FILE FOR ANDROID MALWARE FAMILY CLASSIFICATION USING FCM, FPT, AND DECISION TREES

Abstract

Android malware classification and assigning the appropriate an droid malware family is challenging. Traditional static analysis meth ods can easily be misguided by malware, and dynamic analysis con sumes more space and time. This research proposed a fuzzy-based android malware family classification using multiple aspects of the DEX file. The considered aspects are Permissions of Android applica tion, Image obtained from DEX file sectional features, Dalvik Opcode, and Bytecode of corresponding DEX file. The feature vectors acquired from these multiple aspects are fuzzified using a triangular fuzzifier. The obtained fuzzy sets are classified using an FPT classifier and clus tered using Fuzzy C-means. FPT and FCM are combined according to the views, and a Decision Tree model is obtained for classifying the Android malware family. The final model produces an accuracy up to 95.75%.