A Study On Ciphertext Deduplication Technology Compatible With Keyword Search

The explosion of data and the large number of data leaks have led researchers to devote more energy to the study of ciphertext deduplication systems.However,the existing ciphertext deduplication system still has two problems: The user’s key update needs can not be supported,so the encrypted data can not be re-encrypted with the new key after it is leaked;The ciphertext keyword search technology involved has weak retrieval performance in the scenario of massive data and does not support correlation analysis and recommendation functions.Therefore,this thesis designs an efficient and secure ciphertext deduplication system that supports ciphertext keyword search.The specific contributions are as follows:Firstly,the CAONT technology is modified and a secure and efficient ciphertext deduplication method compatible with key update is proposed,in which there are two data encryption methods: basic encryption scheme and enhanced encryption scheme.The security of the basic encryption scheme is relatively weak but the encryption efficiency is better,and the security of the enhanced encryption scheme is relatively enhanced but the encryption efficiency is slightly lower.In addition,the goal of secure and efficient key update is achieved by fusing two existing cryptographic primitives,attribute-based encryption and key regression.Secondly,this thesis designs a keyword retrieval technology that supports ciphertext association analysis and recommendation: the client extracts the correlations between keywords and keywords,keywords and files in the file and uploads them to the server,and the server establishes a weighted undirected graph between keywords and keywords,an inverted index between keywords and files according to these two correlation degrees,and the server dynamically updates the association information and recommends the most relevant items,which heuristically affects the retrieval behavior of subsequent users.Thirdly,the above ciphertext deduplication method supporting key update and the keyword retrieval technology supporting secret association analysis and recommendation are applied to the system implementation of cloud-edge collaboration: data chunking on the client side and using appeal encryption technology to encrypt data,file-level deduplication on edge servers,and block-level deduplication on cloud servers;When a client retrieves a ciphertext keyword from an edge server,the edge server forwards the request to other edge servers by the cloud server to retrieve the ciphertext keyword,and the cloud server collects responses from each edge server and returns.Finally,a large number of experiments are used to show that the storage space of this system is saved by more than 96.9%,the upload and download speed of FSL and VM datasets is close to 95MB/s under the network bandwidth of 1Gb/s,and the retrieval time of response ciphertext keywords does not exceed 1s.