A novel deep learning technique with cryptographic transformation for enhancing data security in cloud environments

Cloud computing is becoming more and more popular, which is a sign of its importance in the information technology industry because it allows for creative ways to manage various types of data and information systems. In recent years, the amount of data that is available has increased significantly due to the exponential growth of new software applications. Due to high storage, a lot of companies and industries are storing their information on the cloud. However, many customers are reluctant to use the cloud due to security and privacy concerns. To tackle this problem, in this paper, a deep learning algorithm with a new lightweight cryptographic transformation algorithm for enhancing data security on the cloud is proposed. The proposed approach consists of two main stages namely, sensitive data selection and data security. Initially, we separated the sensitive data from the collected data using a deep learning technique called SqueezeNet. To improve the performance of SqueezeNet, the hyper-parameters present in the SqueezeNet are optimally selected using the Rat optimization algorithm (ROA). After that, the sensitive data are encrypted using a lightweight transformation model (LWTM). Finally, the encrypted data are stored on the cloud. As can be seen from the experimental results, the suggested algorithm provided improved security when compared to current cloud computing standards in terms of both cipher size and execution time. The proposed LWTM model yielded an Average Processing Period of 1.53 and an Average Throughput (kb/s) of 190.08, respectively, for evaluating the results.