The Martini Synch: joint fuzzy hashing via error correction

  • Authors:
  • Darko Kirovski;Michael Sinclair;David Wilson

  • Affiliations:
  • Microsoft Research;Microsoft Research;Microsoft Research

  • Venue:
  • ESAS'07 Proceedings of the 4th European conference on Security and privacy in ad-hoc and sensor networks
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Device pairing is a significant problem for a large class of increasingly popular resource-constrained wireless protocols such as Bluetooth. The objective of pairing is to establish a secure wireless communication channel between two specific devices without a public-key infrastructure, a secure near-field communication channel, or electrical contact. We use a surprising user-device interaction as a solution to this problem. By adding an accelerometer, a device can sense its motion in a Cartesian space relative to the inertial space. The idea is to have two devices in a fixed, relative position to each other. Then, the joint object is moved randomly in 3D for several seconds. The unique motion generates approximately the same distinct signal at the accelerometers. The difference between the signals in the two inertially conjoined sensors should be relatively small under normal motion induced manually. The objective is to derive a deterministic key at both sides with maximized entropy that will be used as a private key for symmetric encryption. Currently, our prototype produces between 10-15 bits of entropy per second of usual manual motion using off-the-shelf components.