Keystroke Dynamics Based User Authentication Focusing on Fixed Text Approaches

Protecting sensitive information has become highly critical in the digital age as traditional authentication methods like passwords and PINs are susceptible to theft and brute-force attacks. Even though physical biometric systems, such as fingerprints, iris scans, and facial recognition, offer better security, they usually demand expensive hardware and remain prone to spoofing techniques, such as fake fingerprints, voice recordings, and deepfakes. Behavioral biometrics, which analyzes user interaction patterns such as keystroke dynamics, provides a more secure and user-friendly alternative by focusing on unique typing behaviors. This paper introduces a hybrid CNN-LSTM architecture designed to effectively capture both spatial patterns and temporal dependencies in keystroke dynamics data. The proposed model aims to enhance authentication accuracy while maintaining usability, offering a robust solution to modern security challenges.