Research On The Privacy-preserving Retrieval Over Ciphertext On Cloud

With the rapid development of the Internet and the increasing demand of mass data processing,cloud computing is gaining more and more attention from enterprises and users.More and more users choose to store their information and data on the cloud and use cloud service provider to manage their data without to consider the details that the cloud processing data.In order to ensure the security of the data in the cloud,users will encrypt their data before store it in the cloud.Although the encryption ensures the security of the data,it brings some problems to users when searching their data at the same time.In this paper,we analyze the existing search algorithm over ciphertext,for the problem that most algorithms will disclose user's access pattern and search pattern,we need to study for a retrieval technology over ciphertext which supports privacy protection.In this paper,we mainly study and solve the problem of privacy-preserving retrieval over ciphertext on the cloud.First,in order to prevent the privacy disclosure when retrieving the ciphertext,we introduce the technology of private information retrieval and propose a new method of private information retrieval supporting keyword search which solve the problem that general private information retrieval schemes only support retrieving the data using the physical addresses.Then we describe the details of the algorithm in the detailed design of the scheme which improve the retrieval efficiency and ensure the retrieval security at the same time.The specific works as follow:(1)Design a private information retrieval scheme which supporting keyword search.For the problem that the current scheme of private information retrieval need to know the physical address of documents to be retrieved,we use bloom fliter to construct a mapping between keywords and document storage address which can achieve the query operation and ensure space efficiency without disclosing the privacy of users.(2)Detailed design of the private information retrieval scheme which supporting keyword search.The scheme is mainly divided into three modules:ciphertext storage module,keywords search module,document retrieval module.In the ciphertext storage module,we use two buffers to store document addresses.In the keywords search module,we get the document storage addresses by sending different query request to two buffers and XOR the returned result to get document storage addresses,which greatly improves the search efficiency.In the document retrieval module,we use paillier homomorphic encryption to encrypt the search request and get the retrieval result by decrypting the result that the cloud returns to users without disclosing access pattern.(3)Realization of ciphertext retrieval system on the cloud.This system implements in Hadoop environment and combines with MapReduce programming.The file in the cloud stores in each node of the distributed file system,by carrying out private information retrieval on each node,we can get the final result by merging the calculation value of each node.By using the above scheme,we can meet both the efficiency and privacy requirement.At last,we test the retrieval efficiency and accuracy of the algorithm and verify the performance of the algorithm.