leakiEst

leakiEst — Tool for "Leaking information Estimation"

leakiEst is a command-line tool and Java API for estimating information leakage from previous records of a system's execution.

  • leakiEst performs statistical tests on either discrete or continuous data to distinguish an insecure system with a very small information leak from a secure one with no leakage, and calculates confidence intervals for its estimations. It can estimate the most popular measures of information leakage: mutual information and min-entropy leakage.
  • leakiEst can also calculate exact values of several popular leakage measures (including g-leakage for some gain functions) from given channels and priors.

leakiEst is freely available, and can be downloaded from the download page.

News

  • 20 January 2015: leakiEst Version 1.4.5 is released.
    In this version some bugs in the statistical estimation of mutual informations are fixed.
  • 12 August 2014: leakiEst Version 1.4.2 is released.
    In this version some bugs in the statistical estimation of min-entropy leakage using Chi-square tests are fixed.
  • 25 April 2014: leakiEst Version 1.4.0 is released.
    In this version the statistical estimation of min-entropy leakage using Chi-square tests is supported.
    For observations files and ARFF files, input distributions are also estimated from the trial runs.
  • See History for old releases.

People

The authors of the underlying theory and implementation of leakiEst are:

Bug reports, feature requests and suggestions should be sent to the maintainer, Yusuke Kawamoto.

Publications

For a list of related publications and other documents by the authors, see the information leakage publications page.

Acknowledgement

Some of the work was supported by EPSRC Research Grant EP/J009075/1. Work by Yusuke Kawamoto was partially supported by a postdoc grant funded by the IDEX Digital Society project in France.