A Secure Multi-Factor Authentication System Integrated with Biometrics and Behavioral Analytics Using Reinforcement Learning

Abstract

As standard verification procedures comprising passwords and PINs remain at high risk of attacks from phishing as well as brute-force and social engineering procedures. This work presents a new authentication solution that builds an innovative protection system by joining biometric features with behavioral biometrics to boost user authentication procedures. The new system employs fingerprints together with keystroke dynamic analysis and mouse pattern identification for a secure authentication system that is non-duplicable. Real-time user behavior adaptation occurs through the Double Deep Q-Network (DDQN) model in the Reinforcement Learning (RL) process, which enables continuous authentication. Multiple-layered architecture decreases errors while simultaneously spotting anomalies and delivers peak security results. The main features of this system deliver real-time information processing while adaptive learning and ongoing session tracking functions ensure secure access features, high usability levels, and strict security systems across multiple platforms.