Paper 2021/1263

Transparency Dictionaries with Succinct Proofs of Correct Operation

Ioanna Tzialla, Abhiram Kothapalli, Bryan Parno, and Srinath Setty

Abstract

This paper introduces Verdict, a transparency dictionary, where an untrusted service maintains a label-value map that clients can query and update (foundational infrastructure for end-to-end encryption and other applications). To prevent unauthorized modifications to the dictionary, for example, by a malicious or a compromised service provider, Verdict produces publicly-verifiable cryptographic proofs that it correctly executes both reads and authorized updates. A key advance over prior work is that Verdict produces efficiently-verifiable proofs while incurring modest proving overheads. Verdict accomplishes this by composing indexed Merkle trees (a new SNARK-friendly data structure) with Phalanx (a new SNARK that supports amortized constant-sized proofs and leverages particular workload characteristics to speed up the prover). Our experimental evaluation demonstrates that Verdict scales to dictionaries with millions of labels while imposing modest overheads on the service and clients.

Metadata
Available format(s)
PDF
Category
Cryptographic protocols
Publication info
Preprint. MINOR revision.
Keywords
transparencySNARKsproofskey-value stores
Contact author(s)
srinath @ microsoft com
it608 @ nyu edu
akothapa @ andrew cmu edu
parno @ cmu edu
History
2021-09-22: received
Short URL
https://ia.cr/2021/1263
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2021/1263,
      author = {Ioanna Tzialla and Abhiram Kothapalli and Bryan Parno and Srinath Setty},
      title = {Transparency Dictionaries with Succinct Proofs of Correct Operation},
      howpublished = {Cryptology ePrint Archive, Paper 2021/1263},
      year = {2021},
      note = {\url{https://eprint.iacr.org/2021/1263}},
      url = {https://eprint.iacr.org/2021/1263}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.