AI and Cryptography in IoT Security: A Comprehensive Review of Authentication Approaches

Cybersecurity is crucial for safeguarding data, networks, and systems against cyber threats. However, cybersecurity solutions may not always align with existing hardware and software, potentially leading to security vulnerabilities. The advancements in artificial intelligence (AI) have significantly improved cyber-attack detection by leveraging machine learning, deep learning, and reinforcement learning algorithms. This article reviews various AI-driven approaches for cyber-attack detection, examining their efficiency and effectiveness. Additionally, various cryptographic techniques, including symmetric, asymmetric, and homomorphic encryption, are explored for security enhancement. Authentication schemes such as multifactor authentication, digital signatures, and biometric authentication are also studied to assess their reliability in securing sensitive information. The performance of these algorithms is analyzed based on key parameters, including accuracy (96.24%), positive predictive value (96.10%), hit rate (95.80%), F1 score (95.95%), error rate (21.33%), encryption time (96 sec), and decryption time (85 sec). The findings indicate that deep learning-based deep neural networks (DNN) outperform other models in cyber-attack detection. Additionally, the data encryption standard (DES) algorithm provides improved security, while biometric authentication offers enhanced security and user convenience compared to traditional authentication methods. This study highlights the effectiveness of AI and cryptographic approaches in strengthening cybersecurity frameworks.