MalGuard: Android Malware Detection Using Permission

The growing popularity of Android mobile devices has resulted in an increase in the likelihood that their users will fall victim to malware attacks. As is common with many forms of attack, traditional antivirus solutions tend to rely on signature patterns to try and detect malware applications; these types of solutions can struggle to identify new and evolving forms of malware as they become available. In response to this limitation, MalGuard was developed as a lightweight android malware detection system using permission analysis and machine learning. The MalGuard system utilizes the XGBoost classification algorithm to classify applications within the training set into malicious or benign application categories based on their permission pattern requests. The experimental results indicate that this model produced a detection accuracy of 85% while maintaining a run time that would allow real-time malware detection on an Android device. Therefore, it is concluded that permission-based machine learning approaches can potentially supply an efficient means of securing Android applications.