Secure message transmission with small public discussion

  • Authors:
  • Juan Garay;Clint Givens;Rafail Ostrovsky

  • Affiliations:
  • AT&T Labs – Research;Department of Mathematics, UCLA;Departments of Computer Science and Mathematics, UCLA

  • Venue:
  • EUROCRYPT'10 Proceedings of the 29th Annual international conference on Theory and Applications of Cryptographic Techniques
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

In the problem of Secure Message Transmission in the public discussion model (SMT-PD), a Sender wants to send a message to a Receiver privately and reliably. Sender and Receiver are connected by n channels, up to tn of which may be maliciously controlled by a computationally unbounded adversary, as well as one public channel, which is reliable but not private. The SMT-PD abstraction has been shown instrumental in achieving secure multi-party computation on sparse networks, where a subset of the nodes are able to realize a broadcast functionality, which plays the role of the public channel. However, the implementation of such public channel in point-to-point networks is highly costly and non-trivial, which makes minimizing the use of this resource an intrinsically compelling issue. In this paper, we present the first SMT-PD protocol with sublinear (i.e., logarithmic in m, the message size) communication on the public channel. In addition, the protocol incurs a private communication complexity of $O(\frac{mn}{n-t})$, which, as we also show, is optimal. By contrast, the best known bounds in both public and private channels were linear. Furthermore, our protocol has an optimal round complexity of (3,2), meaning three rounds, two of which must invoke the public channel. Finally, we ask the question whether some of the lower bounds on resource use for a single execution of SMT-PD can be beaten on average through amortization. In other words, if Sender and Receiver must send several messages back and forth (where later messages depend on earlier ones), can they do better than the naïve solution of repeating an SMT-PD protocol each time? We show that amortization can indeed drastically reduce the use of the public channel: it is possible to limit the total number of uses of the public channel to two, no matter how many messages are ultimately sent between two nodes. (Since two uses of the public channel are required to send any reliable communication whatsoever, this is best possible.)