Scholarlink

<p>Cloud data storage plays a significant role in protecting sensitive data from unnecessary access. The demand for security management of cloud data has necessitated the growth of advanced encryption technologies that guarantee data secrecy without compromising the efficiency of computation. This paper presents an innovative hybrid encryption framework called Improved Salp-Swarm Optimization based Paillier Federated Multi-Layer Perceptron encryption method (ISSO-based PF-MLP encryption), integrating Improved Salp-Swarm Optimization with the Paillier Federated Multi-Layer Perceptron model for data security enhancement. The ISSO algorithm optimizes key generation by utilizing chaotic maps, guaranteeing robustness against cryptographic attacks and high entropy. Then, the generated keys are deployed within the Paillier cryptosystem to enable additive homomorphic encryption. Subsequently, this combination eases secure federated learning, permitting several data users collaboratively to train a global model without revealing sensitive data. During the process of encryption and decryption, access control policy verification is performed and the data integrity is checked at both server and user levels. Evaluation of the proposed method through theoretical analysis determines the proposed method’s security and correctness under the discrete logarithm assumption and the decisional composite residuosity assumption. The performance is validated across different datasets such as synthetic, MNIST, and real-world medical data. Experimental outcomes determine that the proposed method outperforms existing encryption methods with less computation time of 5.3&#xa0;s, less encryption overhead of 3.3&#xa0;s, less decryption overhead of 2.7, high accuracy of 98.7%, less encryption time of 2200&#xa0;ms, less decryption time of 2000&#xa0;ms, less processing time of data blocks of 15%, and less storage requirements. Moreover, the proposed method determines its scalability and robustness, making it applicable for shared cloud systems, high-security environments, and large-scale cloud implements. On the whole, the proposed method provides an efficient, scalable, and secure solution for cloud data privacy management, thus granting significant benefits for paramount data privacy applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An intelligent hybrid encryption framework for cloud systems in cybernetics using ISSO and Paillier cryptosystem

  • 期刊论文

  • 发表日期：

    2025年9月11日

查看全文
  • R. S. Kanakasabapathi,
  • J. E. Judith

摘要

Cloud data storage plays a significant role in protecting sensitive data from unnecessary access. The demand for security management of cloud data has necessitated the growth of advanced encryption technologies that guarantee data secrecy without compromising the efficiency of computation. This paper presents an innovative hybrid encryption framework called Improved Salp-Swarm Optimization based Paillier Federated Multi-Layer Perceptron encryption method (ISSO-based PF-MLP encryption), integrating Improved Salp-Swarm Optimization with the Paillier Federated Multi-Layer Perceptron model for data security enhancement. The ISSO algorithm optimizes key generation by utilizing chaotic maps, guaranteeing robustness against cryptographic attacks and high entropy. Then, the generated keys are deployed within the Paillier cryptosystem to enable additive homomorphic encryption. Subsequently, this combination eases secure federated learning, permitting several data users collaboratively to train a global model without revealing sensitive data. During the process of encryption and decryption, access control policy verification is performed and the data integrity is checked at both server and user levels. Evaluation of the proposed method through theoretical analysis determines the proposed method’s security and correctness under the discrete logarithm assumption and the decisional composite residuosity assumption. The performance is validated across different datasets such as synthetic, MNIST, and real-world medical data. Experimental outcomes determine that the proposed method outperforms existing encryption methods with less computation time of 5.3 s, less encryption overhead of 3.3 s, less decryption overhead of 2.7, high accuracy of 98.7%, less encryption time of 2200 ms, less decryption time of 2000 ms, less processing time of data blocks of 15%, and less storage requirements. Moreover, the proposed method determines its scalability and robustness, making it applicable for shared cloud systems, high-security environments, and large-scale cloud implements. On the whole, the proposed method provides an efficient, scalable, and secure solution for cloud data privacy management, thus granting significant benefits for paramount data privacy applications.