Low Cost Secure Computation for the General Client-Server Computation Model

Due to the large number of attacks on open networks, information theft becomes a more and more severe problem. Secure computation can offer highly assured confidentiality protection to critical information and data against external and insider attacks. However, existing secure computation methods are not widely used in practice due to their excessive performance overheads and limited applicability. In this paper, we consider secure computation under a general client-server model which fits the computation of most modern application systems. A novel secure computation protocol has been developed to support highly efficient arithmetic operations. Since the algorithms are based on matrix computations, they are highly efficient. Due to its efficiency and applicability, our secure computation approach can benefit many application systems, including medical databases, secure financial systems, defense information systems, E-commerce systems, agent-based computing, etc.