Convolutional Neural Networks with Data Augmentation Against Jitter-Based Countermeasures

Abstract

In the context of the security evaluation of cryptographic implementations, profiling attacks (aka Template Attacks) play a fundamental role. Nowadays the most popular Template Attack strategy consists in approximating the information leakages by Gaussian distributions. Nevertheless this approach suffers from the difficulty to deal with both the traces misalignment and the high dimensionality of the data. This forces the attacker to perform critical preprocessing phases, such as the selection of the points of interest and the realignment of measurements. Some software and hardware countermeasures have been conceived exactly to create such a misalignment. In this paper we propose an end-to-end profiling attack strategy based on the Convolutional Neural Networks: this strategy greatly facilitates the attack roadmap, since it does not require a previous trace realignment nor a precise selection of points of interest. To significantly increase the performances of the CNN, we moreover propose to equip it with the data augmentation technique that is classical in other applications of Machine Learning. As a validation, we present several experiments against traces misaligned by different kinds of countermeasures, including the augmentation of the clock jitter effect in a secure hardware implementation over a modern chip. The excellent results achieved in these experiments prove that Convolutional Neural Networks approach combined with data augmentation gives a very efficient alternative to the state-of-the-art profiling attacks.

Recent advances in deep learning‐based side‐channel analysis

“…As mentioned earlier, Maghrebi and others applied the CNN architecture to SCA . Cagli and others then proposed a CNN‐based SCA on a protected AES with jitter‐based hiding methods . They were the first to show that a CNN could be used to neutralize jitter‐based hiding countermeasures without any other pre‐processing.…”

“…They were the first to show that a CNN could be used to neutralize jitter‐based hiding countermeasures without any other pre‐processing. In , through experimentation with learning the Sbox output of AES protected by random delay insertion and clock jitter, the robustness of a CNN to data distortions was demonstrated. Although CNN has fewer weights to train than MLP, it requires ample learning data to learn the general invariant features of the traces from a device protected by a hiding method.…”

Neural Cryptanalysis: Metrics, Methodology, and Applications in CPS Ciphers

“…One major direction is to the learning-aided side channel attacks. [8], [19]- [21]. Researchers apply learning approaches to replace the traditional statistical profiling phase in side-channel attacks.…”

A New Type of Countermeasure against DPA in Multi‐Sbox of Block Cipher

“…Chose ( +1)mod8 (19) go to Line (8) (20) end if (21) end if there was one obvious peak in original DPA of DES algorithm for each Sbox. On the contrary, several peaks in our scheme with 5000 traces we found in Figure 5 were "ghost" peaks, which leads to wrong key corresponding to the target Sbox.…”