Friday, March 18, 2016

Spring School on Symmetric Cryptography

This is part one of a two-part blog post collaboratively written by Matthias and Ralph.

The Spring School on Symmetric Cryptography takes place at Beckmanns Hof just one minute south of Ruhr University Bochum.
The Spring school, which can be seen as a supplementary event for FSE 2016 next week, lies literally on the verge of the botanical garden of Bochum!

The Spring School's program covers lectures on the theory of "Boolean Functions" and "Statistical Models" in symmetric cryptography. Additionally there are exercises to gain a firm understanding thereof which were compiled by Anne Canteaut and Celine Blondeau.
The theoretical part is supplemented by the more practically oriented talks of the invited researchers Joan Daemen and Ventzi Nikov.

For the lecture on "Boolean Functions" Anne Canteaut preferred a sympathetic, old-fashioned presentation with chalk on black board in the lecture hall over an overhead presentation in the modern seminar room.

The topic ranged from Boolean functions and their algebraic normal form to the interpretation as Reed–Muller codes. Then the Walsh transform was introduced as a tool to analyze and approximate Boolean functions by linear ones.

In the lecture on "Statistical Models" Celine Blondeau (lecture notes here)
reminded us of basic probability distributions well known to statisticians and we explored some of their uses in cryptography. The discussion lead us from Binomial-, Poisson-, Normal- over Chi-Square- and Hypergeometric distributions to the study of Kullback-Leibler divergence and the applications for i.e. Linear attacks, Differential attacks and Impossible differential cryptanalysis. She also gave directions of ongoing work in that field.

The advantages of "Permutation Based Cryptography" were presented by Keccak (SHA-3) and Rijndael (AES) (co-)designer Joan Daemen.

The talk about "Threshold Implementations" by Ventzi Nikov covered techniques to decompose (non-linear) functions to separate the inputs from intermediate results and masking.
The key-values are masked by splitting up function inputs into separate shares such that knowledge of less than all shares does not allow to recover the value.

Matthias and Ralph will stay tuned for some S-Box theory and the statistical wrap-up on Saturday morning and the slides and recorded videos of this Spring School will soon be available at the official website.

1 comment:

  1. The slides for the invited talks are now also available at: