Practical private information retrieval supporting keyword search in the cloud

In cloud computing environment, just protecting the contents of the queries from users to a large database server is far away from enough. Because it does not protect the leak of access patterns from careful observations. It is thus important to make sure the server learning nothing about the queries including access patterns. However, this implies an expensive computation or communication cost of all the data on the server. Existing solutions are not efficient due to their impractical communication and computation cost. Besides, most of them do not support keyword search. In this paper, we introduce the mechanism of pricing to solve the problem of impractical cost. Using our scheme called KSPIR, we achieve the minimum communication and computation cost according to the flexible privacy and budget specified by users. It is indeed a kind of tradeoff between the cost of retrieval and the degree of privacy. It is worth noting that it also supports keyword search. It allows users to retrieve the data items containing the keywords they are interested in. The experimental results confirm the correctness and efficiency of KSPIR.