CryptoDB
On Black-Box Knowledge-Sound Commit-And-Prove SNARKs
| Authors: |
|
|---|---|
| Download: | |
| Presentation: | Slides |
| Conference: | ASIACRYPT 2023 |
| Abstract: | Gentry and Wichs proved that adaptively sound SNARGs for hard languages need non-falsifiable assumptions. Lipmaa and Pavlyk claimed Gentry-Wichs is tight by constructing a non-adaptively sound zk-SNARG FANA for NP from falsifiable assumptions. We show that FANA is flawed. We define and construct a fully algebraic $F$-position-binding vector commitment scheme VCF. We construct a concretely efficient commit-and-prove zk-SNARK Punic, a version of FANA with an additional VCF commitment to the witness. Punic satisfies semi-adaptive black-box $G$-knowledge-soundness, a new natural knowledge-soundness notion for commit-and-prove SNARKs. We use a new proof technique to achieve global consistency using a functional somewhere-extractable commitment scheme to extract vector commitment's local proofs. |
BibTeX
@inproceedings{asiacrypt-2023-33492,
title={On Black-Box Knowledge-Sound Commit-And-Prove SNARKs},
publisher={Springer-Verlag},
author={Helger Lipmaa},
year=2023
}