A Next-Generation Multi-Layer Adaptive Authentication Framework for Secure Digital Payments

Digital payment ecosystems are increasingly exposed to sophisticated cyberattacks including deepfake‑based biometric spoofing, automated credential‑stuffing pipelines, adversary‑in‑the‑middle interception, and high‑frequency fraud orchestration. Conventional authentication methods relying on passwords, SMS OTP, or single‑modality biometrics fail to provide durable resistance against these dynamic threats. This paper introduces a scientifically grounded, multi‑layer adaptive authentication architecture combining FIDO۲/WebAuthn passkeys, multimodal biometrics with neural‑network‑based liveness detection, continuous behavioral profiling, and an AI‑driven contextual risk engine capable of real‑time decision optimization. Experimental evaluations conducted on public anti‑spoofing datasets, behavioral datasets, and synthetic financial logs demonstrate a significant reduction in false acceptance rates, enhanced detection of coordinated fraud, and measurable improvements in user experience. The proposed framework aligns with international standards including ISO ۲۳۲۲۰, NIST SP ۸۰۰‑۶۳‑۴, and EMV specifications, ensuring regulatory compliance. This research highlights the necessity of intelligence‑centric authentication for next‑generation payment infrastructures and provides a deployable architectural model for financial institutions seeking to modernize their security layers.

کلیدواژه ها:

Adaptive Authentication ، Biometrics ، Behavioral Modeling ، Risk Engine ، FIDO۲ ، Cybersecurity ، Digital Payments