DEEP LEARNING-BASED ENHANCED CLOUD AUTHENTICATION USING COGNITIVE BIOMETRICS AND SECURE ENCRYPTION TECHNIQUES

Abstract

With the increasing reliance on digital systems, the want for tightly closed and green authentication mechanisms has emerge as paramount. conventional password-based totally authentication techniques are liable to protection vulnerabilities such as phishing, brute-pressure attacks, and credential leaks. Biometric authentication, specifically face popularity, has emerged as a extra dependable opportunity, presenting a seamless consumer experience. but, biometric structures continue to be susceptible to spoofing assaults, variations in facial features as a result of growing older or accessories, and antagonistic manipulations. To deal with those demanding situations, this study proposes a cloud-primarily based cognitive picture authentication framework that integrates deep gaining knowledge of, cognitive biometrics, and encryption techniques to beautify authentication protection. The proposed framework leverages MobileNetV2, an optimized deep mastering version for efficient face recognition in cloud environments. The authentication procedure starts off evolved with photograph segmentation and XOR encryption, which prevents unauthorized reconstruction of biometric information while keeping computational efficiency. The model is skilled on both actual-time and publicly available biometric datasets to make certain robustness throughout numerous authentication scenarios. Comparative analysis with deep studying fashions which includes CNN, InceptionV3, VGG19, and ResNet50 demonstrates that MobileNetV2 outperforms different models in phrases of accuracy, precision, and computational performance. The experimental effects suggest that the proposed model achieves 95.30% check accuracy on actual-time datasets, substantially enhancing authentication reliability. The look at evaluates the effect of encryption strategies on cloud-primarily based authentication. The consequences display that XOR encryption is computationally quicker than AES encryption, making it appropriate for real-time cloud authentication. The proposed gadget balances security and efficiency, making sure that authentication is both invulnerable and scalable. the mixing of cognitive protection responses provides an additional layer of protection, reducing the risks of spoofing and unauthorized access. This research affords a novel, impervious, and scalable biometric authentication framework that leverages deep mastering and encryption for more suitable cloud security. The findings demonstrate that the proposed technique notably improves authentication accuracy at the same time as keeping sturdy security measures. future paintings will explore the integration of federated gaining knowledge of for decentralized authentication, adversarial robustness, and multi-modal biometrics to further toughen authentication protection.