Permissionshield: A Hybrid Static And Dynamic Analysis Approach For Android Permission Misuse Detection

Abstract

The Rapid Growth Of Android Applications Has Increased Concerns Regarding Excessive And Unjustified Permission Usage, Which Can Lead To Privacy Breaches, Data Leakage, And Unauthorized Access To Sensitive Resources. Existing Android Security Solutions Often Rely Solely On Static Or Dynamic Analysis, Which Limits Their Accuracy And Fails To Provide Comprehensive Insights Into Real-world Permission Misuse. To Address These Limitations, This Work Proposes PermissionShield, A Hybrid Static–dynamic And Forensic Analysis Framework Designed To Detect, Predict, And Visualize Suspicious Permission Behaviors In Android Applications. The System Integrates Multi-stage Analysis, Beginning With Static Extraction Of Declared Permissions, Followed By Dynamic Evaluation Of Runtime Behavior To Identify Inconsistencies Between Requested And Actual Usage. A Machine Learning–based Prediction Model Further Enhances Detection Accuracy By Classifying Potentially Malicious Permission Patterns Using Historical Datasets And Feature Encoding. The Framework Also Incorporates A Feature-extraction Engine To Quantify Risk Levels And Generates Detailed Forensic Reports Along With Severity Visualizations To Assist Developers, Analysts, And End Users In Understanding The Threat Landscape. Experimental Results Demonstrate That PermissionShield Effectively Identifies High-risk Permissions, Reduces False Positives, And Provides A Scalable And Interpretable Solution For Android Permission Misuse Detection. This Research Contributes Toward Strengthening Mobile Security, Improving Transparency In Permission Handling, And Enabling Proactive Protection Of User Privacy.

Keywords