Verifiable Single-Server Private Information Retrieval

摘要

Single-server Private Information Retrieval (SPIR) allows a client to privately retrieve some data from a database stored on a server. While many SPIR schemes exist, these previous SPIR schemes are generally under the honest-but-curious server model. This model however is not suitable for many real world scenarios such as involving the untrusted cloud server. In this paper, we first propose an SPIR scheme that is based on the learning with (binary) errors assumption under the honest-but-curious server model. Specifically, compared with some previous SPIR schemes, our proposal provides a low communication complexity. Then, according to the above warm-up scheme, we introduce a Verifiable SPIR (VSPIR) scheme under the malicious server model where the server may provide some fraudulent answers. To the best of our knowledge, our scheme is the first practical VSPIR scheme that employs the probabilistic verification process. Finally, for our proposal, we present the theoretical analyses of the properties (i.e., correctness, privacy and security), and give the detailed implementation results.