Permission-Based Anomalous Application Detection on Android Smart Phone

Abstract

Information applications are widely used by millions of users to perform many different activities. Android-based smart phone users can get free applications from Android Application Market. But, these applications were not certified by legitimate organizations and they may contain malware applications that can steal private information from users. The proposed system develops a permission-based malware detection to protect the privacy of android smart phone users. This system monitors various permissions obtained from android applications and analyses them by using a statistical technique called Singular Value Decomposition (SVD) to estimate the correlations of permissions. The dataset including approximately 4000 malware JSON files are downloaded from https://www.kaggle.com/goorax/static-analysis-of-androidmalware-of-2017. The training phase emphasizes on the malware samples (approximately 300) which includes the most significant patterns of the current malware environment according to the analysis results. The testing phase is conducted on 120 malware and goodware apps. The proposed system evaluates the risk level (High, Medium, and Low) of Android applications based on the correlation patterns of permissions. The overall accuracy of the system is 85% for malware applications and goodware applications as the test results.