Permission-based approach for Android malware analysis through ensemble-based voting model

Smart devices have become an integral part of our lives as mobile and wireless technology has advanced. Although Android has the largest market share, it is also the platform most frequently attacked by hackers. More Android malware is being created as the use of Android cell phones grows. Malware detection on Android has become a critical mission. We present a permission-ensemble-based malware detection mechanism for Android in this paper. Permission combinations declared in the device manifest file is used to detect malware on Android. We collected the permission combinations that malware and benign apps often request. We developed an ensemble model that uses permission combinations to distinguish between malicious and benign apps. We showed that the combination of classifiers into an ensemble model provided better accuracy than an individual classifier. We experimentally proved that our ensemble model is robust to the changing nature of data. The malware detection rate is up to 99.3% in our experimental evaluation. According to our experiments with real malware, our proposed Android malware detection scheme is compelling and reliable.