Paper 2024/1716

Rate-1 Statistical Non-Interactive Zero-Knowledge

Abstract

We give the first construction of a rate-1 statistical non-interactive zero-knowledge argument of knowledge. For the $\mathsf{circuitSAT}$ language, our construction achieves a proof length of $|w| + |w|^\epsilon \cdot \mathsf{poly}(\lambda)$ where $w$ denotes the witness, $\lambda$ is the security parameter, $\epsilon$ is a small constant less than 1, and $\mathsf{poly}(\cdot)$ is a fixed polynomial that is independent of the instance or the witness size. The soundness of our construction follows from either the LWE assumption, or the $O(1)$-$\mathsf{LIN}$ assumption on prime-order groups with efficiently computable bilinear maps, or the sub-exponential DDH assumption. Previously, Gentry et al. (Journal of Cryptology, 2015) achieved NIZKs with statistical soundness and computational zero-knowledge with the aforementioned proof length by relying on the Learning with Errors (LWE) assumption.