Secure Computation Using Leaky Tokens

Leakage-proof hardware tokens have been used to achieve a large number of cryptographic tasks recently. But in real life, due to various physical attacks, it is extremely difficult to construct hardware devices that are guaranteed to be leakage-proof. In this paper, we study the feasibility of general two-party computation using leaky hardware tokens. Our main result is a completeness theorem that shows that every nontrivial leaky two-party functionality can be used for general secure computation. In fact, the protocol we construct is non-interactive and unconditionally secure. There are no restrictions on the leakage functions associated with the token, except that it does not render the tokens trivial, by revealing its entire secrets to the adversary.