PRAZdroid: A Novel Approach to Risk Assessment and Zoning of Android Applications based on Permissions

Abstract

The proliferation of Android apps has increased harmful apps that aim to influence user security, privacy, and device execution. Conventional detection techniques are becoming ineffective in identifying malicious applications as malware has enhanced its cognition and ingenuity and has reached a point where they are more impervious. Novel approaches based on machine learning have been provided to detect and classify malware threats. Still, the risk assessment of Android applications is significant for enhancing user trust and needs more attention. Permissions analysis is an effective way for risk assessment and behaviour study of Android apps because apps require permissions to access device functionality. In endorsement, this study proposes an approach (PRAZdroid) for risk assessment using permissions analysis. The proposed approach analyzed the M0droid dataset and computed five risk levels (Level 0 to Level 4). Statistical analysis is performed for risk levels and achieved 98 .07% classification accuracy with the Drebin and Anrdozoo datasets.