Securing Sensitive Data in Multi-Cloud Storage using ML and Homomorphic Encryption
Mohammed El Moudni, Ziyati Elhoussaine · 2024
The fast-growing adoption of cloud and digital technologies has made it indispensable to store and process personal data in cloud environments. However, increasing reliance on cloud services has raised concerns regarding the security of sensitive personal information. To address these challenges, this study introduced a design that leverages Machine Learning and homomorphic encryption concepts by utilizing serverless and secret vault services provided by cloud services. Our model employs data encryption to ensure robust protection while handling data. We explored and evaluated the effectiveness of sensitivity classification using logistic regression, support vector machines, and convolutional neural network (CNN) algorithms. The results highlight the good performance of CNN, achieving remarkable accuracy and recall. Additionally, an analysis of the time costs for encryption reveals that while using our model, there is a minimal impact on the time complexity. Comprehensive experiments and analysis of the suggested datasets demonstrated the effectiveness of our approach for enhancing data storage protection in a multi-cloud environment.