Research On Updatable Block-level Ciphertext Deduplication Protocol In Cloud Storage

With the advent of the big data era,the volume of global data has exploded rapidly.The related studies have shown that there is a large amount of duplicate data stored in the cloud storage servers,which consumes lots of storage resources.In order to alleviate the storage pressure of the cloud storage servers,data deduplication technology as a special "data compression technology" is widely used in cloud storage systems to delete the redundant data.The smaller the granularity of deduplication,the higher the deduplication ratio.In addition,the client-side deduplication schemes can effectively reduce users' upload bandwidth and save the upload time.At present,the data deduplication technology in cloud storage servers should protect users' data privacy and support the data-updated operations to better meet users' requirements.Therefore,the updateable block-level ciphertext deduplication technology is becoming a research hotspot in data deduplication field.However,there are three big challenges in implementing this technology.First,how to achieve block-level ciphertext deduplication across different files.Second,how to support update operations on the cloud data safely and efficiently.Third,how to deal with file ownership management issues to prevent users' data from being illegally accessed.Although the message-locked encryption algorithm can initially solve the first challenge,there is no such a solution to solve these three challenges at the same time.To address these three issues,we present an updatable block-level ciphertext deduplication scheme with dynamic ownership management(UBLDe).Specifically,we design the dynamic balanced skip list(DBSL)as the authenticated data structure for proof of ownership to support efficient update operations.In addition,we propose a fast hybrid and separation algorithm MIX.The servers use the group key and MIX algorithm to protect the ciphertext of blocks and key file from being recoverd by any malicious users and achieve the file ownership management securely and efficiently.And then we give the security analysis of UBLDe scheme in detail.Finally,the experimental results show that the UBLDe scheme is efficient and practical with lower computation and communication overhead compared to the existing schemes,especially when the file size are large.UBLDe scheme focuses on block-level deduplication and update operations on the encrypted data as well as the file ownership management.Although the UBLDe scheme can reduce the upload bandwidth of the duplicate files,the duplicate encrypted blocks belonging to the nonduplicate files are still required to be uploaded to the servers for deduplication.Meanwhile,UBLDe scheme does not explicitly provide a metadata storage strategy for files and file blocks.Therefore,we propose a block-level client-side deduplication scheme with serverside storage management on encrypted data(BL-CDSM)that is improved on the basis of UBLDe scheme.BL-CDSM scheme achieves the both fine-grained block-level deduplication and coarse-grained file-level deduplication at client side,which greatly reduces users' upload bandwidth.In order to effectively implement the metadata storage for files and file blocks,we propose hash table tree(HTT)as the storage management structure,which can quickly search and update file/file block metadata with scalable storage ability,and build the "user-file-file block" relationship explicitly.And then we give the security analysis of BL-CDSM scheme in detail.Finally,the experimental results show that the BLCDSM scheme can greatly reduce users' upload bandwidth compared with the UBLDe scheme,but it always consumes more computational resources.However,once the block deduplication ratio(BDR)is more than 90%,the client-side computational overhead incurred by the BL-CDSM scheme will be greatly reduced,even less than that of the UBLDe scheme.