MMCNN-ZO: Modified multi-scale convolutional neural network-based zebra optimization for enhancing data encryption system

Cloud computing (CC) is a rapidly growing technology that is deployed by various organizations due to the high level of scalability, elasticity, and virtualization it offers. However, security-related challenges arise due to insecure access control points. Security models in the cloud such as authentication, data recovery, confidentiality, accessibility, and data integrity. This includes the deployment model, barriers in cloud computing, cloud services, and security issues. Data security can be improved via the usage of encryption and decryption keys during transmission. However, the existing techniques often suffered from improved time consumption and complexity in handling more users. This paper mainly focuses on the data security issue with the outsourced sensitive data in the cloud server by designing a novel Modified Multi-scale Convolutional Neural Network-based Zebra Optimization (MMCNN-ZO) approach to improve the cloud data sharing system. The proposed MMCNN-ZO approach safeguards the crucial data effectively and outputs in the form of cipher text/image. Here, the weight functions of the MMCNN are tuned mechanically using the ZO algorithm for the enhanced encryption process. The decryption process is performed to obtain the data in user understandable format. These procedures make the proposed system effectively secure the data content by protecting against intrusions. The efficacy of the proposed MMCNN-ZO method is investigated using diverse evaluation indicators namely throughput, decryption, and encryption time. The proposed MMCNN-ZO algorithm achieved improvements in terms of throughput, encryption time, and decryption time of 7450 bits/sec, 4.82 ms, and 4.86 ms respectively compared to other existing methods of SVM, ANN, BPPSVC, and GBCRP. The experimental results inherit the better performances of the suggested MMCNN-ZO approach over other methods in terms of all metrics.