CSED: Client-Side encrypted deduplication scheme based on proofs of ownership for cloud storage

Abstract

As digital data are explosively generated nowadays, data management becomes a critical problem, which makes cloud storage services important and popular. In reality, the storage overhead can be reduced significantly by performing date deduplication. Among the outsourced data, some of them are very personal and sensitive, and should be prevented for any leakage. Generally, if clients conventionally encrypt the data, deduplication is lost. Message-locked encryption (MLE) is a cryptographic primitive supporting encrypted data deduplication. A secure client-side deduplication scheme can be built upon MLE to reduce both communication and computation overhead for cloud storage systems, where a client interacts with the cloud server to check the duplicate data and only the data which has not been outsourced by other clients before is required to be uploaded. However, existing client-side encrypted data deduplication schemes are confronted with brute-force attacks that can recover files falling into a known set. Furthermore, existing schemes are vulnerable to illegal content distribution attacks, where the adversary can distribute data to other users via the cloud server without detecting. In this paper, we propose a secure and efficient client-side encrypted data deduplication scheme (CSED). In CSED, a dedicated key server is introduced in generating MLE keys to resist brute-force attacks. We propose a Bloom filter-based proofs of ownership (PoW) mechanism and integrate it into CSED to resist illegal content distribution attacks. Moreover, a hierarchical storage architecture is employed to improve the I/O efficiency on the cloud server. Security analysis and performance evaluation demonstrate that CSED is secure and efficient.