Toward secure sensitive data in the cloud

Abstract

Cloud storage and computation have many exciting features such as low cost, scalability, and unbounded resources. These features attract many individuals and organizations to migrate their sensitive data, including biometric data, to the cloud. However, security and privacy are major concerns in the cloud, especially within a single cloud, due to issues such as availability, key management, insider attacks, brute-force attacks, and privacy concerns. These issues become more challenging in the cloud, especially when dealing with sensitive data such as biometric data. This dissertation addresses the security and privacy issues in the cloud. The first contribution of this dissertation addresses the challenges of availability and insider attack in a single cloud storage. To overcome these obstacles, we propose N-Cloud scheme. N-Cloud scheme splits a file into smaller chunks and replicates these chunks into multiple clouds in a non-overlapping manner. The second contribution aims to address brute-force attacks and key management issues in the cloud storage. We overcome these obstacles with the proposed CloudStash scheme, which applies the secret-sharing scheme on the file itself instead of the keys to store multiple shares of a file into multiple clouds. We conclude this dissertation by addressing the security and privacy issues of the biometric data in the cloud. We propose a Cloud-ID-Screen scheme that is designed specifically for storing and processing biometric fingerprint data in the cloud while preserving the privacy and security of the biometric fingerprint data. The Cloud-ID-Screen scheme splits the biometric fingerprint data into small subsets and distributes these small subsets of fingerprint features over multiple clouds. This distribution ensures that no single cloud stores all of the fingerprint features, which provides security and privacy while also improving the matching speed up and getting comparable accuracy.