Client-side-based Storage Deduplication In Cloud And A Research On Graphical Password

Due to the rapid development of cloud storage,deduplication techniques draw extensive attention.Simultaneously,with the popularity of graphical devices,Android Unnlock Pattern has been one of the most widely used graphical authentication schemes.In this paper,we provide a solution respectively for these issues.Deduplication techniques are widely deployed in cloud storage service,which save the bandwidth and make a better use of the storage utilization.Message-Locked Encryption(MLE),proposed by Bellare et.al.,provides a solution of identical plaintext encrypted into identical ciphertext even by different clients to perform secure deduplication.However,there are many files similar but not identical in practical cloud storage,and it is necessary to achieve a fine-grained deduplication on those files in order to further increase the utilization rate of storage.To address these issues,we present MLE-based Chunk-Level Encrypted Deduplication with Key Management in Cloud Storage(MCL-KM),which supports both chunk-level and file-level deduplication and achieves chunk key management through Ciphertext Policy Attribute-Based Encryption(CP-ABE).Security analysis and performance experiments show that MCL-KM keeps a negligible difference with MLE,but holds stronger security and reliability via fine-grained Po W and file access control via key management based on CP-ABE.As the most widely used graphical password scheme on mobile terminals,Android Unlock Patterns(AUP)in practice are not uniformly distributed on its theoretical password space,which exposes a tremendous hazard easily exploited by the attacker to expedite dictionary attack or violence crack.To this point,this paper proposed a new scheme,Android-Unlock-Pattern Based on Random Points Exclusion(AUP-RPE),by which the user is forced to change his habitual choice due to the new arrangement.And by performing a large-scale user study with over 1,100 people to collect patterns in real use.Modeling based on those patterns,we have found that the entropy of AUP-RPE increases over 3 orders of magnitude than the entropy of AUP,which means that AUP-RPE has a much stronger security.