A Verifiable Data Deduplication Scheme in Cloud Computing

Abstract

Deduplication is an important technique to save the storage cost at the cloud storage server. Image is an important data type stored in cloud, but rarely discussed in previous work on deduplication. This paper studies the problem of validating the deduplication of image storage in cloud. In particular, we consider the task of allowing a cloud server to verify the correctness of deduplication. Our scheme consists of several advantages over the previous work, whose framework can be described through the following algorithms. Firstly, before each user uploads an encrypted image, he calculates its hash value as the fingerprint. Secondly, the fingerprint is sent to both cloud servers for checking duplicates. If the storage and verification servers both reply to the user with 'no deduplication', the user transfers his data to the servers. Otherwise, once the fingerprint is consistently found, the user gives up uploading data for deduplication. Specially, when the fingerprint is only found in one server, it implies that the results are inconsistent and at least one of servers is invalid. The security and efficiency analysis is also presented in this paper.