Enhancing Cloud Security: A Multi-Factor Authentication and Adaptive Cryptography Approach Using Machine Learning Techniques

Abstract

The rapid expansion of cloud computing underscores the critical need for advanced security measures to protect sensitive data on remote servers. Authentication is crucial for safeguarding these data. Despite various proposed methods, vulnerabilities persist. This article introduces a novel multi-factor authentication system integrated with a hybrid cryptographic framework that dynamically changes encryption algorithms using machine learning techniques based on an intrusion-detection system. The proposed system employs passwords, conditional attributes, and fingerprint authentication to derive the encryption key from fingerprint data. It uses a dual-encryption strategy that combines five algorithm pairs: AES + HMAC (SHA-256), ECC + HMAC (SHA-512), HMAC-MD5 + PBKDF2, Twofish + Argon2, and Blowfish + HMAC SHA3-256. A Hybrid CNN-transformer model predicts and classifies attacks by dynamically adjusting an encryption algorithm to secure the data. The framework exhibited strong resilience against brute force, spoofing, phishing, guessing, and impersonation attacks. The proposed model achieved a commendable accuracy of 96.8%, outperforming other models. Implementing this framework in a cloud authentication environment significantly enhances data confidentiality and protects against unauthorized access. This study highlights the potential of combining multi-factor authentication and adaptive cryptography to obtain robust cloud security solutions.