IZPR: Instant Zero Knowledge Proof of Reserve

Description

This is the source code of the IZPR Project, presented at CoDecFin'24 ( T. Conley, N. Diaz, D. Espada, A. Kuruvilla, S. Mayne, and X. Fu. "IZPR: Instant Zero Knowledge Proof of Reserve". Pre-print at https://eprint.iacr.org/2023/1156.pdf).

Based on the recent progress in pre-processed lookup argument, we present a system that can prove the asset of a financial organization, instantly and incrementally, in zero knowledge. The prover complexity is O(nlog(n)), where n is the platform throughput (e.g., 7 for BTC and 2000 for Visa network). The proof size and verifier cost are O(1). Concretely, running with 32-threads on a 5 dollar/hr Google Cloud C2D instance, it supports throughput of 1024 transactions per second. The proof size is 3.4kb.

Getting Started

Dependencies

• Cargo/Rustc 1.7.3
• arkswork library (https://github.com/arkworks-rs/).

Installing

• Run ./scripts/compile.sh

Executing program

• Run python3 scripts/paperdata.py (or cd data; ./run.sh)

Help

One specific version of the arkswork library is placed in the /dependency with some slight modifications for retrieving some of its private data members in structs and resolving rust compiler incompatibility. If backporting, search "Modified by" and check the changes made.

TODOs

• The zero knowledge cq protocl in this version uses a combination of zk-VPD scheme (Y. Zhang, D. Genkin, J. Katz, D. Papadopoulos, and C. Papamanthou. "A Zero-Knowlede Version of vSQL", available at: https://eprint.iacr.org/2017/1146.pdf), and $k$-bounded leaking KZG poly commits. The proof size and prover speed can be improved about 1x if the cq+ protocol (M. Campanelli, A. Fanoi, D. Fiore, T. Li and H. Lipmaa, "Lookup Arguments: Improvements, Extensions, and Applications to Zero-Knowledge Decision Trees". available at: https://hal.science/hal-04234948/document) is used.

• We have another zk-CQ protocol using QA-NIZK with similar cost to cq+. Source code to be released later.

Authors

Trevor Conley, Nilsso Diaz, Diego Espada, Alvin Kuruvilla, Xiang Fu ([email protected]).

License

This project is licensed under the Apache and Open MIT License.

Acknowledgments

• We implemented a zk-derivative of the CQ protocol by L. Eagen D. Fiore, and A. Gabizon (https://eprint.iacr.org/2022/1763.pdf).
• The group FFT algorithm is adapted from arkworks' FFT algorithm for field elements.