Storage Deduplication

Data outsourcing to a cloud storage brings forth one of new challenges for the efficient resource utilization as well as keeping security for the outsourced data simultaneously. Recently, Zheng and Xu proposed a Proof of Storage with Deduplication (POSD) scheme for a secure and efficient cloud storage service [1]. Exploiting the public verifiability [2], the POSD scheme couples two notions of Proof of Data Possession (PDP) [2][3] and Proof of Data Ownership (POW) [4] and provides a solution to achieve both of security and efficiency. Using the POSD scheme, a client can be assured the integrity of its outsourced data. In addition, a storage server can take advantage of deduplication techniques in a secure manner. That is, the storage server can efficiently utilize resources such as storage space and network bandwidth while preventing information leakage [5][6]. In the POSD scheme, the verification of auditing and deduplication protocol entirely depends on public keys, which are created and provided by clients [1]. Hence, the validity of the scheme is implicitly based on an assumption, which we call random key assumption, that all clients are honest in terms of generating their keys. In the cross-multiple users and the cross-domain environment of the cloud computing, however, such an assumption is unrealistic. Eliminating random key assumption may cause storage systems that utilize the POSD scheme to face a new security threat not considered before. Unfortunately, the scheme 2 has a serious security breach under new attack model allowing malicious clients to make dishonestly manipulated keys. In this paper, we present the security weakness of the POSD scheme. More specifically, we show that the scheme fails to satisfy two security requirements, server unforgeability and (κ, θ)-uncheatability, under new attack model that is very reasonable and effective. A countermeasure against this attack is provided by modifying the scheme such that the clients-created keys are blended with the random values contributed by the storage server. The proposed solution actually weakens the client’s capability to control their keys. The modification is minimized so that our scheme preserves the efficiency while providing more robust security. This paper is organized as follows: In Section 2, we briefly review the POSD scheme. New attack model and some attack scenarios are presented in Section 3, and countermeasure against the attack is described in Section 4. Finally, we conclude this paper in Section 5.

Reference Link:

https://eprint.iacr.org/2012/554.pdf