Secure Encrypted Deduplication Based on Frequency Distribution

Data deduplication technology enhances data storage efficiency when managing large data volumes. Many encrypted deduplication schemes have been proposed to maintain privacy while supporting deduplication as people become more concerned about data privacy. Because of their deterministic nature, traditional schemes based on convergent encryption(CE) or message-locked encryption (MLE) have a weakness against brute force attacks or frequency analysis attacks. Much research has proposed solutions to balance deduplication efficiency and confidentiality while resisting frequency analysis attacks. However, they failed to notice that most of the data distributed among multiple users are shared data, while private data are mainly distributed among users themselves. In this paper, we propose a deduplication scheme based on the frequency distribution, which improves the deduplication efficiency and reduces information leakage. First, we measure the privacy needs of data based on their frequency distribution among users. Second, we design a multiple-level encryption scheme, which increases deduplication efficiency for chunks evenly distributed among users while improving data confidentiality for chunks concentrated on a small number of users. Last, We analyzed the security of our scheme and evaluated it to demonstrate the efficiency of our design.