Subsampling and Knowledge Distillation on Adversarial Examples: New Techniques for Deep Learning Based Side Channel Evaluations

Gohr, Aron; Jacob, Sven; Schindler, Werner

doi:10.1007/978-3-030-81652-0_22

Cited by 5 publications

(12 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Section: Main Contributionsmentioning

confidence: 99%

“…The neural network architecture used in our deep learning approach was first introduced in [11] to break a protected AES implementation. In [11], the side-channel attack extracted the Hamming weights of all AES subkeys. Afterwards, a equation solving stage was used to perform a limited amount of error correction on the extracted Hamming weights while simultaneously deriving the full key values from the Hamming weight guesses.…”

Section: Main Contributionsmentioning

confidence: 99%

“…Afterwards, a equation solving stage was used to perform a limited amount of error correction on the extracted Hamming weights while simultaneously deriving the full key values from the Hamming weight guesses. This work uses the same network architecture, but cannot use any of the subsequent post-processing ideas used in [11]. There are multiple reasons for this: the Clyde-128 key schedule does not impose significant constraints on the values we might recover the side-channel data made available in the contest does not cover the full round number of the cipher anyway and we never get to see any unmasked internal state in the Clyde attack.…”

Section: Main Contributionsmentioning

confidence: 99%

“…• We basically predict Hamming weights of some sensitive variables derived from the cipher state shares. To do so, we use a deep residual neural network following the architecture first introduced in [11].…”

Section: Overviewmentioning

confidence: 99%

“…• Subsampling factor q 1 and subsample size q 2 : number of subsampled versions of the input trace that are analyzed in parallel by the network as in [11] and the size of each subsampled trace. By definition, we have…”

Section: Leakage Modelmentioning

confidence: 99%

See 4 more Smart Citations

Breaking Masked Implementations of the Clyde-Cipher by Means of Side-Channel Analysis

Gohr¹,

Laus

Schindler

2022

TCHES

Self Cite

View full text Add to dashboard Cite

In this paper we present our solution to the CHES Challenge 2020, the task of which it was to break masked hardware respective software implementations of the lightweight cipher Clyde by means of side-channel analysis. We target the secret cipher state after processing of the first S-box layer. Using the provided trace data we obtain a strongly biased posterior distribution for the secret-shared cipher state at the targeted point; this enables us to see exploitable biases even before the secret sharing based masking. These biases on the unshared state can be evaluated one S-box at a time and combined across traces, which enables us to recover likely key hypotheses S-box by S-box.In order to see the shared cipher state, we employ a deep neural network similar to the one used by Gohr, Jacob and Schindler to solve the CHES 2018 AES challenge. We modify their architecture to predict the exact bit sequence of the secret-shared cipher state. We find that convergence of training on this task is unsatisfying with the standard encoding of the shared cipher state and therefore introduce a different encoding of the prediction target, which we call the scattershot encoding. In order to further investigate how exactly the scattershot encoding helps to solve the task at hand, we construct a simple synthetic task where convergence problems very similar to those we observed in our side-channel task appear with the naive target data encoding but disappear with the scattershot encoding.We complete our analysis by showing results that we obtained with a “classical” method (as opposed to an AI-based method), namely the stochastic approach, thatwe generalize for this purpose first to the setting of shared keys. We show that the neural network draws on a much broader set of features, which may partially explain why the neural-network based approach massively outperforms the stochastic approach. On the other hand, the stochastic approach provides insights into properties of the implementation, in particular the observation that the S-boxes behave very different regarding the easiness respective hardness of their prediction.

show abstract

Section: Main Contributionsmentioning

confidence: 99%

Section: Main Contributionsmentioning

confidence: 99%

Section: Main Contributionsmentioning

confidence: 99%

Section: Overviewmentioning

confidence: 99%

Section: Leakage Modelmentioning

confidence: 99%

See 3 more Smart Citations