Research On Privacy Preserving And Its Application Framework Over Outsourced Queries

With the development of cloud computing, the outsourcing of data storage and queryis becoming increasingly popular. However, data privacy preserving cannot be ignored. En-cryption is a well established technology for protecting sensitive data. Researchers haveexplored many techniques to execute queries over encrypted data, but these approaches onlysupport limited queries. Homomorphic encryption, which allows one to evaluate over en-crypted data without decrypting them, makes it possible to solve this issue in theory. In thispaper, we talk about how to query over outsourced data using homomorphic encryption, andour main research work is as follows:(1) We have analyzed existing homomorphic encryption technologies and query meth-ods over encrypted data. By comparing advantages and disadvantages of all homomorphicencryption schemes in aspect of efciency and storage space, we use the homomorphic en-cryption scheme based on the RLWE assumption in this paper. We implement using the NTL library. Further more, we implement some optimization techniques, such asre-linearization, pack, permuting/routing.(2) We construct a query application framework called enQuery. The ciphertext gener-ated by is random, so comparisons can not be conducted directly on ciphertexts. Inthis paper, we employ a third party, called agent server, besides the data owner and storageserver. When comparison operations are needed for the storage server, it processes Sub onciphertexts and sends the diference to the agent server. The agent processes the Dec opera-tion and returns results to the storage server in the form of ciphertexts. The data owner, thestorage server and the agent comprise the framework enQuery.(3) Based on the framework enQuery, we process queries on encrypted data. In thispaper, We mainly talk about equality condition and range queries, join queries, sort, skylinecomputation and top-k. In order to get better performance and reduce communication, wepreprocess the diference of adjacent data and store them in the agent. With the use of thepre-processed data, we reduce a lot of homomorphic operations on ciphertexts in the storageserver and the agent. And we presort for join queries to improve efciency. Our research work on privacy preserving and its application framework over outsourcedqueries has important significance, extends applications in homomorphic encryption, andprovides some reference for relative researches.