Keeping the SZK-verifier honest unconditionally

Abstract

This paper shows that using direct properties of a zeroknowledge protocol itself, one may impose a honest behavior on the verifier (without additional cryptographic tools). The main technical contribution is showing that if a language L has an Arthur-Merlin (i.e. public coins) honest-verifier statistical SZK proof system then L has an (anyverifier) SZK proof system when we use a non-uniform simulation model of SZK (where the simulation view and protocol view can be made statistically closer than any given polynomial given as a parameter). Three basic questions regarding statistical zero-knowledge (SZK) are solved in this model: – If L has a honest-verifier SZK proof then L has an any-verifier nonuniform simulation SZK proof. – If L has an SZK proof then Ḹ has an non-uniform simulation SZK proof. – If L has a private-coin SZK proof then L has a public-coin nonuniform simulation SZK proof.