Private Information Retrieval And It's Application

Private Information Retrieval (PIR) allow a client to retrieve the data of d1 from server's n-bit data string d=d1d2...dn, while keeping the value of i private. When protect the privacy of server's data as well as client's privacy in PIR scheme, this kind of PIR is called symmetrically private information retrieval (SPIR). There are many private information retrieval schemes, according to privacy requirements are divided into:information-theoretic private information retrieval protocol, computationally private information retrieval and symmetrically private information retrieval. According to participants are honest whether or not, the schemes of private information can be divided into:semi-honest model private information retrieval and malicious model private information retrieval. Private information retrieval is a strong cryptographic tool that is widely used in various security-related applications, such as secure query scheme, anonymous authentication and probabilistically checkable arguments and so on.In the semi-honest model, this thesis mainly studies the computationally secure private information retrieval and symmetric private information retrieval. The main work of this paper is as follows: At first, fully-homomorphic encryption based symmetrically private information retrieval. At present, the best communication complexity of computational SPIR protocols is O(log2 n), in this paper, we propose a fully-homomorphic encryption (FHE) based SPIR, it improve communication complexity of SPIR from O(log2 n) to O(log n), which is better than existing SPIR scheme. The privacy of client and server is based on the security of fully-homomorphic cryptosystem. namely, CPA-secure. In addition, our scheme can be directly used to implement l-out-of-n non-interactive oblivious transfer with O(1) sender-side communication complexity.Secondly, accredited private information retrieval scheme. Access control is a new research problem of private information retrieval, the existing accredited private information retrieval need to make a lot of exponentiations when update the client's authority, the complexity of computation is expensive. In this paper, we propose an accredited private information retrieval with preprocess. and our solution don't need extra computation cost compared to traditional PIR scheme, and also don't need to do exponentiations when update the client's authority, so the complexity of computation is get reduced. In an accredited private information retrieval scheme, only the accredited client could recover the correct data from the retrieval result.