VerSum: Verifiable Computations over Large Public Logs
Name
Kaashoek_VerSum.pdf
Size
277.54 KB
Format
Adobe PDF
Checksum (MD5)
546bf6f770eae189d1d6512b6d3f9aaf
Author(s)
van den Hooff, Jelle
Kaashoek, M. Frans
Zeldovich, Nickolai
Date Issued
November 2014
Journal
Proceedings of the 2014 ACM SIGSAC Conference on Computer and Communications Security (CCS '14)
Publisher
Association for Computing Machinery (ACM)
Citation
Jelle van den Hooff, M. Frans Kaashoek, and Nickolai Zeldovich. 2014. VerSum: Verifiable Computations over Large Public Logs. In Proceedings of the 2014 ACM SIGSAC Conference on Computer and Communications Security (CCS '14). ACM, New York, NY, USA, 1304-1316.
Version
Author's final manuscript
Abstract
VerSum allows lightweight clients to outsource expensive computations over large and frequently changing data structures, such as the Bitcoin or Namecoin blockchains, or a Certificate Transparency log. VerSum clients ensure that the output is correct by comparing the outputs from multiple servers. VerSum assumes that at least one server is honest, and crucially, when servers disagree, VerSum uses an efficient conflict resolution protocol to determine which server(s) made a mistake and thus obtain the correct output.
VerSum's contribution lies in achieving low server-side overhead for both incremental re-computation and conflict resolution, using three key ideas: (1) representing the computation as a functional program, which allows memoization of previous results; (2) recording the evaluation trace of the functional program in a carefully designed computation history to help clients determine which server made a mistake; and (3) introducing a new authenticated data structure for sequences, called SeqHash, that makes it efficient for servers to construct summaries of computation histories in the presence of incremental re-computation. Experimental results with an implementation of VerSum show that VerSum can be used for a variety of computations, that it can support many clients, and that it can easily keep up with Bitcoin's rate of new blocks with transactions.
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1145/2660267.2660327
