Advanced Stealth Protection for Android Devices
Govind Suryawanshi, S. C. Patil, Krishna Sharma, A. M. Gore, Amey Joshi · 2025
As mobile devices become essential tools in daily life, their vulnerability to malware and phishing attacks has surged, particularly within the Android ecosystem. Current security frameworks predominantly rely on static or dynamic detection techniques, each with inherent limitations. Static analysis involves examining an application’s code and permissions to identify potential threats. However, this method often falls short against obfuscated malware and zero-day exploits that can dynamically alter their behavior to avoid detection. Conversely, dynamic analysis captures an application’s behavior during runtime, but it is constrained by the requirement for real-world user interaction and can miss threats that do not exhibit malicious behavior in controlled environments.To address these limitations, this project introduces an innovative Android security application that integrates both static and dynamic detection mechanisms through a robust cloud-based architecture. The application employs the DL-AMDet architecture for deep learning-based malware detection, utilizing a combination of CNN-BiLSTM for static feature analysis and deep autoencoders for dynamic feature extraction. Additionally, it implements advanced phishing detection algorithms for both website URLs and SMS messages, utilizing machine learning techniques tailored for URL classification and SMS pattern recognition. By harnessing cloud resources, the system not only enhances performance and scalability but also provides comprehensive security coverage that adapts to emerging threats. This hybrid approach ultimately aims to fortify the Android user experience against increasingly sophisticated cyber threats.