Two party secure computation of a secret Boolean function
Secure computation is the computation of a function over private inputs. In the general setting, party 1 (Alice) holds a private input x and party 2 (Bob) holds a private input y. After applying the protocol, Alice learns f(x, y) and Bob learns g(x, y). Alice and Bob never learn each other's input other than what they can derive from f( x, y) and g(x, y) respectively. We consider a slightly different problem, which we call “two-party secure computation of a secret boolean function”. Alice knows input x and Bob knows a function f. After the protocol, Alice learns f(x) but no information about f. Bob doesn't learn anything about Alice's input x. We are interested in an application of this protocol in a tax preparation software, and possibly other softwares available through web services. The main concern is preserving the privacy of the customer using the service. In this thesis, we describe our protocol, which is a modification of the protocol of privacy preserving auction by Naor, Pinkas and Sumner, in which they apply these ideas to electronic auction. We also describe an implementation of the main ingredient of our protocol. While our protocol introduces significant overhead, it can be applied to any web service where privacy preservation is more important than efficiency.
Chen, Wei, "Two party secure computation of a secret Boolean function" (2004). ETD Collection for University of Texas, El Paso. AAI1423698.