Research On Keyword Based Ciphertext Retrieval Technology In Cloud Storage

For convenience and cost savings,more and more enterprises and individuals are outsourcing their private data to public cloud storage server.To protect data privacy,they encrypt data before uploading them to the cloud storage server,which makes traditional keyword-based search methods difficult.Recently,the technology for keyword search on ciphertext has got extensive attentions and a lot of keyword search schemes have been proposed.However,existing keyword based ciphertext search schemes face many challenges,such as untrusted cloud storage servers,low search efficiency,poor search flexibility,and memory leak attacks.In order to resolve these problems,several keyword search schemes on ciphertext for different requirements are proposed.The main contributions of this thesis are summarized as follows:(1)In the semi-honest and curious cloud server model,a Merkle hash tree based verifiable keyword search scheme on ciphertext is proposed.In the proposed scheme,Lucene search engine toolkit and Merkle hash tree are utilized to build a secure inverted index and the authentication data structure,which achieves high search efficiency and the search results verification.The improvement on the traditional plaintext index structure realizes the high efficiency.Meanwhile,the Merkle hash tree structure achieves the verification of search results.(2)To realize the high efficiency and dynamic updating simultaneously,an efficient dynamic multi-keyword ranked search scheme on ciphertext is proposed.The proposed scheme utilizes vector space model to construct binary vector index,and uses a sparse matrix to encrypt the index and trapdoor to improve retrieval efficiency.Besides,a special inverted vector structure and update algorithm is proposed to achieve dynamic updating for the multi-keyword ranked search scheme.(3)To resist the memory leakage attack in the existing ciphertext search schemes,a memory leakage-resilient multi-keyword ranked search scheme is proposed.The proposed scheme utilizes physically unclonable functions and fuzzy extractor to generate the secret key in real time,so that the key need not to be stored in plaintext in memory,thereby resisting memory leakage attacks.In addition,in order to further improve security,the similarity scores are encrypted using an order preserving function.The proposed scheme makes the multi-keyword ranking search more secure and more practical.Finally,the related research works on keyword search schemes are summarized and the future works are outlined.