Research On Provable Data Possession Models And Schemes

With the development of Internet and data explosion, cloud storage has becomean important part in cloud computing areas. Storage security in the cloud is one offundamental issues in cloud storage. Storage servers in the cloud, which suffer fromsystem failure or malicious attacks, are not necessarily trusted. Hence it is important toensure users the integrity of their data. The Provable Data Possession (PDP) model hasbeen defined for ensuring the integrity of files on untrusted storages. The proof of dataintegrity is done by a challenge-response protocol between the user and the remoteserver. Public verifiability and data dynamics are two appealing features to extend thePDP model. However, when applying public verifiability and data dynamics, thingsbecome complicated. Not only has efficiency been reduced, i.e. batch tag verificationand batch update verification are not supported, but also introduce security issue intoit, i.e user privacy may be compromised when a curious third party auditor exists. Inthis paper, the above problems are addressed in details and some improvements areput forward:Based on the tree-like hash authentication structure, a method to verify blocktags in a batch way is put forward, which improves the efficient of static tagverification;A new model which enables batch update is introduced. In our new model, datablock update and authentication can be done in batch rather than one by oneas in the traditional model. To achieve this, a variant authentication2-3treeand related algorithms are developed. The efficiency of tag authentication canbe achieved by twice when the target blocks are uniformly distributed and evenmore when the target blocks are continuously located.To remedy the information leak issue under public verification and complete the security model, a method is used to blind the data but still keep the capability ofthe third party auditor to verify the data integrity.