Secure and Efficient Cloud Data Retrieval Using Privacy-Preserved Hybrid CRNN with Swallow Swarm Optimization

Abstract

The application of Cloud Computing (CC) has increased popularity in recent years. This technology allows for resource sharing and extensive capabilities, making it feasible to store and analyze data remotely on the cloud. However, it is not secured, some parties can access to a network like the internet and read or alter data, making this cloud untrustworthy. Consequently, one of the issues that must be resolved while utilizing CC involves preserving data security and privacy. Several strategies based on various Encryption (Enc) systems have been explored to address data privacy and integrity. Cloud-related risks include data loss and leakage, malware attacks, and exploited vulnerabilities. In order to prevent attacks and to preserve privacy when keeping data in the cloud, it is crucial to make sure that a foolproof protecting system is in operation. In order to entirely understand the value of Medical Data (MD) and realize data collaborative sharing, a Deep Learning (DL) architecture that uses Homomorphic Encryption (HE) technology was established in this work to protect training parameters and created a MD security sharing scheme based on the communication mode.
In this case, the training parameters are protected by using the Paillier HE (PHE) to achieve additive homomorphism. This article offers a Privacy-Preserving (PP) hybrid Convolutional Recurrent Neural Network (CRNN) based on Swallow swarm optimization technique (SSO) to address the issue of privacy leaking. By combining DL with HE, a knowledge transfer strategy with PP is created. Researchers may infer from the simulation results that Learning Rate (LR), batch size, and other factors are connected to the model prediction accuracy. The outcomes demonstrate that this method has good performance, completes the accurate disease prediction, and accomplishes Data Sharing (DS) while maintaining data privacy.