Detecting Malware in Android Applications by Using Androguard Tool and XGBoost Algorithm

Abstract

Android is the most popular operating system for smartphones and tablets. With its popularity, Android mal ware has also grown dramatically. Many conventional malware detection techniques are now not sufficient, due to sophisticated detection avoidance strategies. According to ongoing research, one harmful Android software is released every 10 seconds. To counter these significant mal ware campaigns, scalable detection approaches require that can provide quick and accurate identification of mal ware apps. To overcome the above issues, we proposed a method to detect malware in Android applications by extracting features like activities, services, requested permissions, and intent filters from the manifest file. Furthermore, the androguard tool is used to disassemble the code and extract all suspicious API calls by reading the dex code. These extracted features are serialized in feather data format for efficient retrieval. After that, the XGBoost algorithm is used to detect the malware. The result of the proposed method gives 97% accuracy.