Secure data deduplication using secret sharing schemes over cloud

Data deduplication has become an integral part of managing repositories of outsourced data to the cloud data centers. However, centralized data centers face issues of data loss and accessibility if something goes faulty as deduplication maintains just a unique copy of the content. Secure data deduplication exploits convergent encryption to perform data deduplication in encrypted domain. However, managing convergent keys provides a single point of vulnerability and overhead problems. Towards this end, we propose a secure data deduplication scheme that formally addresses the problems of fault tolerance, efficient and reliable key management, data confidentiality by obfuscation of outsourced information and integrity check at the user's end prior to downloading via computation of authentication codes. Data is distributed into random looking shares based on Permutation ordered binary (POB) number system at multiple servers and is further made secure via the notion of proof of ownership (PoW) concept. Also, key overhead is minimized using Chinese Remainder Theorem (CRT) based secret sharing. The efficacy of the proposed scheme has been demonstrated by the experimental results and security analysis validates its suitability with respect to various attacks in real time scenarios.