Multiparty proximity testing with dishonest majority from equality testing

  • Authors:
  • Ran Gelles;Rafail Ostrovsky;Kina Winoto

  • Affiliations:
  • University of California, Los Angeles;University of California, Los Angeles;University of California, Los Angeles

  • Venue:
  • ICALP'12 Proceedings of the 39th international colloquium conference on Automata, Languages, and Programming - Volume Part II
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Motivated by the recent widespread emergence of location-based services (LBS) over mobile devices, we explore efficient protocols for proximity-testing. Such protocols allow a group of friends to discover if they are all close to each other in some physical location, without revealing their individual locations to each other. We focus on hand-held devices and aim at protocols with very small communication complexity and a small constant number of rounds. The proximity-testing problem can be reduced to the private equality testing (PET) problem, in which parties find out whether or not they hold the same input (drawn from a low-entropy distribution) without revealing any other information about their inputs to each other. While previous works analyze the 2-party PET special case (and its LBS application), in this work we consider highly-efficient schemes for the multiparty case with no honest majority. We provide schemes for both a direct-communication setting and a setting with a honest-but-curious mediating server that does not learn the users' inputs. Our most efficient scheme takes 2 rounds, where in each round each user sends only a couple of ElGamal ciphertexts.