Yuwen Zhang

EECS Department, University of California, Berkeley

Technical Report No. UCB/

May 1, 2024

http://www2.eecs.berkeley.edu/Pubs/TechRpts/Hold/9eb82f705edcc5405a7365a91f78ed3f.pdf

Zero knowledge succinct non-interactive arguments of knowledge (zkSNARKs) allow an untrusted prover to cryptographically prove that a certain statement is true without compromising their privacy. Though powerful, many existing applications of zkSNARKs do not scale for larger systems. By tailoring protocol and system design for specific use cases, I demonstrate that systems using zkSNARKs can scale well in many dimensions. In my first chapter, I focus on privacy- preserving analytics systems. Existing deployments use a small set of non-colluding servers alongside some specialized zkSNARK constructions in order to compute aggregate statistics over client data without learning any individual’s information. Our system, Whisper, improves upon prior work by drastically reducing inter-server communication at the cost of slightly larger client proofs, resulting in large dollar cost savings. In my second chapter, I discuss techniques for delegated proof generation for complex circuits. In particular, I focus on the delegated prover environment, where a trusted delegator outsources proof generation to third party workers. Existing solutions either trust these workers with their secret inputs in plaintext, or they fail to fully take advantage of worker parallelism. Our system, DFS, achieves state of the art scaling without trusting workers with sensitive delegator secrets.

Advisors: Raluca Ada Popa

---

BibTeX citation:

@mastersthesis{Zhang:31385,
    Author= {Zhang, Yuwen},
    Title= {Scaling Zero Knowledge Proofs Through Application and Proof System Co-Design},
    School= {EECS Department, University of California, Berkeley},
    Year= {2024},
    Number= {UCB/},
    Abstract= {Zero knowledge succinct non-interactive arguments of knowledge (zkSNARKs) allow an untrusted
prover to cryptographically prove that a certain statement is true without compromising their
privacy. Though powerful, many existing applications of zkSNARKs do not scale for larger
systems. By tailoring protocol and system design for specific use cases, I demonstrate that systems
using zkSNARKs can scale well in many dimensions. In my first chapter, I focus on privacy-
preserving analytics systems. Existing deployments use a small set of non-colluding servers
alongside some specialized zkSNARK constructions in order to compute aggregate statistics over
client data without learning any individual’s information. Our system, Whisper, improves upon
prior work by drastically reducing inter-server communication at the cost of slightly larger client
proofs, resulting in large dollar cost savings. In my second chapter, I discuss techniques for delegated
proof generation for complex circuits. In particular, I focus on the delegated prover environment,
where a trusted delegator outsources proof generation to third party workers. Existing solutions
either trust these workers with their secret inputs in plaintext, or they fail to fully take advantage
of worker parallelism. Our system, DFS, achieves state of the art scaling without trusting workers
with sensitive delegator secrets.},
}

EndNote citation:

%0 Thesis
%A Zhang, Yuwen 
%T Scaling Zero Knowledge Proofs Through Application and Proof System Co-Design
%I EECS Department, University of California, Berkeley
%D 2024
%8 May 1
%@ UCB/
%F Zhang:31385