EXPLORE PUBLICATIONS BY COUNTRIES

Privacy-preserving, Database, Hidden Queries, Hash, Efficiency.

As the increase of adopting database systems as the key data management technology by organizations for day-to-day operations and decision making, the security and privacy issues of these systems becomes crucial. Achieving privacy-preserving range query efficiently is a difficult challenge in practice. Many privacy-preserving protocols use secure multi-party computation (MPC) as building block, which present elegant privacy and security, but brings too much computation and communication overheads at the same time.

In this paper, we consider the three-party database system model: A client performing privacy-preserving range queries through a Proxy (trusted third party, TTP), to a Server’s database. The User does not learn how the query applied on the database, nor any other quarriable attributes that the database may contain. The Proxy does not learn any information about the Server’s private data, though he interacts with the Server directly. The Server on the other hand, learns nothing about the User’s query.

We propose two practical privacy-preserving query schemes for database system. The basic idea of our schemes is to first convert each data entry into a set of concrete numbers, which is called attribute value numericalization. Then combines in a novel way several efficient cryptographic techniques, such as secure hash function, pseudo-random function, XOR, etc. to check whether the records match a query. The experimental evaluation (using the data sets collected by UCI KDD) of our prototype implementation show that our protocols incur reasonable computation and communicating overhead for added privacy-preserving benefit and perform better than those MPC-based solutions.