Protocol Research Of Sharing For Attribute Name And Equi-join For Encrypted Data Across Relational Cloud Databases

Cloud database is a database that is deployed and virtualized on the cloud computing platform,and typically is provided by Service Provider(SP)to different users.Because SP may leak user’s privacy information when it manages and maintains those databases,users have to encrypt their sensitive data to protect privacy information.Data encryption can protect privacy information effectively,but cause problems in information sharing between different users.The lack of sharing protocol attribute name and the low efficiency of existing equi-join protocol are the main problems in existing encrypted information sharing schemes in Cloud database.In semi-honest model,this thesis researches encrypted information sharing between two tenants across relational cloud database(CRDB).The detailed research contents include:(1)According cloud database’s characteristics,this thesis analyzes the security threats in encrypted information sharing,and designs an encryption information sharing scheme for users in CRDB.This scheme’s participants,architecture and execution flow are described clearly in this part.The execution flow includes user connection,database connection and sharing protocol implement.(2)A sharing method for attribute names based on locality-sensitive Hashing(LSH)is proposed.Then,the AS-sim protocol implemented by simhash algorithm is proposed.The experiment results showed that AS-sim protocol has the best performance when the length of the attribute name was longer than nine Chinese characters,and the threshold was between60% and 70%.While the best performance means the accuracy and recall rate of searching sharable attribute are all over 90% there.(3)This thesis also presents an efficient equi-join sharing protocol called EAVS protocol.EAVS used commutative encryption function to find their all intersection for one attribute,and share those attribute values which two users both have and the corresponding Cartesian product of the tuples to create the equi-join results.The security analysis proves that it can complete equi-join sharing fairly between two users in semi-honest model;and efficiency analysis indicates that it can remarkably improve the sharing efficiency.