A sophisticated Android security application implementing a Triple-Layer Authentication system. This project utilizes behavioral biometrics—specifically Keystroke Dynamics and Pattern Rhythm Analysis—to ensure the user is the genuine owner.
- Biometric Layer: Initial hardware-level fingerprint verification using the Android Biometric API.
- Pattern Dynamics: A custom-built pattern lock that analyzes the speed and consistency of the drawing motion.
- Keystroke Dynamics: A password field that utilizes a custom QIFNN (Quantum-Inspired Firefly Neural Network) logic to verify typing rhythm.
The core of the app's security lies in the QIFNN.kt engine. Inspired by the Firefly Algorithm's optimization capabilities, it features:
- Asymmetric Bias: Penalizes slow, hesitant input (intruder behavior) by 3.5x while remaining lenient toward fast, muscle-memory input (owner behavior).
- Adaptive Learning: Updates the user's profile after every successful login using an Alpha Learning Rate, mimicking the "attractiveness" and "movement" mechanics of the Firefly algorithm to stay aligned with the user.
- Jitter Detection: Uses Standard Deviation to identify inconsistent "shaky" input typical of non-genuine users.
- Zero-Knowledge Storage: All patterns and passwords are saved as SHA-256 hashes.
- Anti-Extraction:
allowBackupis disabled to prevent data theft via ADB or cloud backups. - Lifecycle Protection: Activities auto-terminate on
onStop()to prevent background bypass.