Research On Oblivious Searchable Encryption Technology In Cloud Environment

With the development of cloud storage and cloud computing technology,massive amounts of sensitive or private data generated by various applications,including personal health records,emails,etc.,are outsourced to the cloud platform.Although cloud storage provides convenience to outsourcing users and reduces the burden of large-scale data management in a cost-effective manner,data privacy and security have increasingly become a worrying issue.To alleviate this concern,sensitive data is usually outsourced to the server in ciphertext to prevent unauthorized access.However,storing the encrypted data on the server will restrict our legitimate access,such as search and update functions.In order to solve this problem,researchers have conducted extensive research on the problem of searchability in the context of encrypted documents.But there were a great deal of information leakage and privacy security issues in many solutions,which limit the application of searchable encryption solutions in real life.Therefore,how to design a safe,efficient and accurate searchable encryption scheme has become a hot issue of current research.The main research work and contributions of this thesis are as follows:1.In terms of exact matching,this thesis designs a privacy protection searchable encryption scheme based on oblivious random access for the problems of access pattern,search pattern,and response length leakage.A tree-based non-clustered secure index data structure is designed and constructed using oblivious random access mechanism,which realizes random access on the index structure,guarantees index security and improves retrieval efficiency.At the same time,Paillier homomorphic encryption is used to realize ciphertext domain retrieval,and the design realizes an efficient oblivious random access strategy,which improves the security of the scheme.2.In terms of similarity matching,aiming at the problems of text similarity leakage and text feature sparsity,this thesis designs a secure similarity retrieval scheme based on differential privacy protection.The feature vector space model fused with differential privacy is used to extract features of the text,that is,the Laplace noise mechanism is used to add random noise to the text features,and the similarity privacy is protected by differential privacy,and the text feature sparse feature representation is constructed,and then used locality sensitive hashing reduces the dimensionality of sparse features to obtain the signature matrix.Then the Jaccard distance is used to measure the similarity of low-dimensional text features,and a secure index structure based on hash buckets is constructed.Finally,a homomorphic encryption algorithm is used to design and implement a secure similarity search scheme.Finally,the two searchable encryption schemes proposed in this thesis are theoretically analyzed and experimentally verified on real public datasets.The security analysis and experimental results show that the schemes proposed in this thesis are safe and effective.