Wide collisions in practice

  • Authors:
  • Xin Ye;Thomas Eisenbarth

  • Affiliations:
  • Department of Mathematical Sciences, Florida Atlantic University, Boca Raton, FL;Department of Mathematical Sciences, Florida Atlantic University, Boca Raton, FL

  • Venue:
  • ACNS'12 Proceedings of the 10th international conference on Applied Cryptography and Network Security
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Detecting collisions in the power consumption of a modern cryptographic engine is usually difficult due to small leakages and possible countermeasures. Wide collisions offer a much stronger leakage that significantly facilitates their detection. This is the first time wide collisions are exploited in a power analysis attack. In this work we introduce a collision detection method based on the detection of characteristic outliers. Detection results are compared to optimized subspace-based templates. We show that the outlier detection method, while not requiring a template building phase, is almost as effective in detecting collisions as the template-based approach.