Characterization of the electromagnetic side channel in frequency domain

  • Authors:
  • Olivier Meynard;Denis Réal;Sylvain Guilley;Florent Flament;Jean-Luc Danger;Frédéric Valette

  • Affiliations:
  • Institut TELECOM, TELECOM ParisTech, CNRS, LTCI, UMR, Département COMELEC, Paris Cedex 13, France and DGA, MI, CELAR, Bruz, France;DGA, MI, CELAR, Bruz, France and INSA, IETR, Rennes, France;Institut TELECOM, TELECOM ParisTech, CNRS, LTCI, UMR, Département COMELEC, Paris Cedex 13, France and Secure-IC S.A.S., Paris, France;Institut TELECOM, TELECOM ParisTech, CNRS, LTCI, UMR, Département COMELEC, Paris Cedex 13, France;Institut TELECOM, TELECOM ParisTech, CNRS, LTCI, UMR, Département COMELEC, Paris Cedex 13, France and Secure-IC S.A.S., Paris, France;DGA, MI, CELAR, Bruz, France

  • Venue:
  • Inscrypt'10 Proceedings of the 6th international conference on Information security and cryptology
  • Year:
  • 2010

Quantified Score

Hi-index 0.01

Visualization

Abstract

In this article, we propose a new approach to characterize the EM leakage of electronic devices by identifying and focusing on the signals' frequencies leaking the most information. We introduce a set of tests based on cryptanalysis methods that will help vendors and users of sensitive devices to estimate the security risks due to leakage through electromagnetic emanations. We propose two approaches: an empirical one and another based on information theory. Both provide a characterization of the leakage i.e. the frequencies and the bandwidths where information is contained. These techniques are low cost, automatic, and fast as they can be performed with an oscilloscope and some softwares for the characterization. Such evaluation could also be carried out with TEMPEST. But TEMPEST evaluations require dedicated apparatus and time consuming step work that consists in scanning all the spectrum frequencies. Our approach does not substitute to regulatory TEMPEST evaluation, but nonetheless can identify the leakage with high confidence. To illustrate the relevance of our approach, we show that an online software filtering at some identified frequencies allows us to recover a key stroked in one measurement at the distance of 5 meters from the keyboard.