INTELLIGENT VIDEO AUTHENTICATION PROTOCOL FOR ATM USERS TO ENHANCE CYBERSECURITY

Abstract

This paper proposes an intelligent protocol for enhancing the cybersecurity of user authentication in automated teller machines (ATMs) through the integration of a video channel that performs both biometric identification and behavioral analysis. The proposed approach enables the system to recognize a user’s identity and verify the authenticity of their actions in real time, thereby reducing the probability of spoofing, replay, and substitution attacks. The architecture of the video authentication protocol unifies visual feature extraction, liveness detection, and encrypted data exchange within a formalized interaction model between the user, terminal, and decision-making server. The protocol’s decision logic is modeled as a Markov Decision Process (MDP) with a reward function that balances security confidence and response time. In the event of network degradation, the system performs a graceful fallback, shifting critical verification tasks to the terminal and executing asynchronous server-side validation with deferred confirmation. Data confidentiality is ensured through on-device template storage, pseudonymization, and differential privacy mechanisms, while integrity is maintained using digital signatures and nonce-based freshness control. Simulation results demonstrate the feasibility of integrating the proposed protocol into existing banking infrastructures without significant latency overhead. The developed model provides a foundation for further research in intelligent multimodal authentication systems that combine video analytics, behavioral biometrics, and adaptive cybersecurity frameworks for next-generation financial technologies. The research contributes to the development of resilient, user-centric authentication mechanisms that enhance trust and operational security in digital financial ecosystems.