Léo Weissbart

Leo Weissbart

[jpg]Lose everything and eat dog food. --InspiroBot

NAME

Léo Weissbart - PhD Student at TU Delft and Radboud University

SYNOPSIS

Léo Weissbart [--cv] [--gh] [--sc] [--dblp] [--mail] [-l]

DESCRIPTION

Léo Weissbart is a PhD student at EEMCS INSY of TU Delft in the AISY Lab and the Digital Security group of the Radboud University in the Cesca Lab, working in the area of physical attacks and machine learning under the supervision of Stepjan Picek and Lejla Batina. Before that, he was an engineering student at the Grenoble INP-Esisar (National Institute of advanced systems and networks, France).

OPTIONS

--cv: Curriculum Vitae
--gh: GitHub repository / GitLab repository
--sc: Google Scholar
--dblp: DBLP Computer Science Bibliography
--mail: hover @ me leo weissbart

RESEARCH TOPICS

Hardware Security, Deep Learning, Implementation of Cryptography, Physical Attacks and Countermeasures

PUBLICATIONS

PUB-2312.07783 - BarraCUDA: Bringing Electromagnetic Side Channel Into Play to Steal the Weights of Neural Networks from NVIDIA GPUs

Péter Horváth, Łukasz Chmielewski, Léo Weissbart, Lejla Batina, Yuval Yarom

Arxiv
Over the last decade, applications of neural networks have spread to cover all aspects of life. A large number of companies base their businesses on building products that use neural networks for tasks such as face recognition, machine translation, and autonomous cars. They are being used in safety and security-critical applications like high definition maps and medical wristbands, or in globally used products like Google Translate and ChatGPT. Much of the intellectual property underpinning these products is encoded in the exact configuration of the neural networks. Consequently, protecting these is of utmost priority to businesses. At the same time, many of these products need to operate under a strong threat model, in which the adversary has unfettered physical control of the product. Past work has demonstrated that with physical access, attackers can reverse engineer neural networks that run on scalar microcontrollers, like ARM Cortex M3. However, for performance reasons, neural networks are often implemented on highly-parallel general purpose graphics processing units (GPGPUs), and so far, attacks on these have only recovered course-grained information on the structure of the neural network, but failed to retrieve the weights and biases. In this work, we present BarraCUDA, a novel attack on GPGPUs that can completely extract the parameters of neural networks. BarraCUDA uses correlation electromagnetic analysis to recover the weights and biases in the convolutional layers of neural networks. We use BarraCUDA to attack the popular NVIDIA Jetson Nano device, demonstrating successful parameter extraction of neural networks in a highly parallel and noisy environment.

PUB-2023-1922 - One for All, All for Ascon: Ensemble-based Deep Learning Side-channel Analysis

Azade Rezaeezade, Abraham Basurto-Becerra, Léo Weissbart, Guilherme Perin

Cryptology ePrint Archive
In recent years, deep learning-based side-channel analysis (DLSCA) has become an active research topic within the side-channel analysis community. The well-known challenge of hyperparameter tuning in DLSCA encouraged the community to use methods that reduce the effort required to identify an optimal model. One of the successful methods is ensemble learning. While ensemble methods have demonstrated their effectiveness in DLSCA, particularly with AES-based datasets, their efficacy in analyzing symmetric-key cryptographic primitives with different operational mechanics remains unexplored. Ascon was recently announced as the winner of the NIST lightweight cryptography competition. This will lead to broader use of Ascon and a crucial requirement for thorough side-channel analysis of its implementations. With these two considerations in view, we utilize an ensemble of deep neural networks to attack two implementations of Ascon. Using an ensemble of five multilayer perceptrons or convolutional neural networks, we could find the secret key for the Ascon-protected implementation with less than 3 000 traces. To the best of our knowledge, this is the best currently known result. We can also identify the correct key with less than 100 traces for the unprotected implementation of Ascon, which is on par with the state-of-the-art results.

PUB-2023-1598 - Lightweight but Not Easy: Side-channel Analysis of the Ascon Authenticated Cipher on a 32-bit Microcontroller [PDF]

Léo Weissbart and Stjepan Picek

Cryptology ePrint Archive
Ascon is a recently standardized suite of symmetric cryptography for authenticated encryption and hashing algorithms designed to be lightweight. The Ascon scheme has been studied since it was introduced in 2015 for the CAESAR competition, and many efforts have been made to transform this hardware-oriented scheme to work with any embedded device architecture. Ascon is designed with side-channel resistance in mind and can also be protected with countermeasures against side-channel analysis. Up to now, the effort of side-channel analysis is mainly put on hardware implementations, with only a few studies being published on the real-world side-channel security of software implementations. In this paper, we give a comprehensive view of the side-channel security of Ascon implemented on a 32-bit microcontroller for both the reference and a protected implementation. We show different potential leakage functions that can lead to real-world leakages and demonstrate the most potent attacks that can be obtained with the corresponding leakage functions. We present our results using correlation power analysis (CPA) and deep learning-based side-channel analysis and provide a practical estimation of the efforts needed for an attacker to recover the complete key used for authenticated encryption. Our results show that the reference implementation is not side-channel secure since an attacker can recover the full key with 8,000 traces using CPA and around 1,000 traces with deep learning analysis. While second-order CPA cannot recover any part of the secret, deep learning side-channel analysis can recover partial keys with 800 traces on the protected implementation. Unfortunately, the model used for multi-task key recovery lacks the generalization to correctly recover all partial keys for the full key attack.

JOU-2023-101-55470 - Label Correlation in Deep Learning-Based Side-Channel Analysis

Lichao Wu, Léo Weissbart, Marina Krcek, Huimin Li, Guilherme Perin, Lejla Batina, Stjepan Picek

IEEE Transactions on Information Forensics and Security - IEEE TIFS 2023
The efficiency of the profiling side-channel analysis can be significantly improved with machine learning techniques. Although powerful, a fundamental machine learning limitation of being data-hungry received little attention in the side-channel community. In practice, the maximum number of leakage traces that evaluators/attackers can obtain is constrained by the scheme requirements or the limited accessibility of the target. Even worse, various countermeasures in modern devices increase the conditions on the profiling size to break the target. This work demonstrates a practical approach to dealing with the lack of profiling traces. Instead of learning from a one-hot encoded label, transferring the labels to their distribution can significantly speed up the convergence of guessing entropy. By studying the relationship between all possible key candidates, we propose a new metric, denoted Label Correlation (LC), to evaluate the generalization ability of the profiling model. We validate LC with two common use cases: early stopping and network architecture search, and the results indicate its superior performance.

PUB-2021-720 - On Reverse Engineering Neural Network Implementation on GPU [PDF]

Łukasz Chmielewski and Léo Weissbart

Artificial Intelligence in Hardware Security - AIHWS 2021
In recent years machine learning has become increasingly mainstream across industries. Additionally, Graphical Processing Unit (GPU) accelerators are widely deployed in various neural network (NN) applications, including image recognition for autonomous vehicles and natural language processing, among others. Since training a powerful network requires expensive data collection and computing power, its design and parameters are often considered a secret intellectual property of their manufacturers. However, hardware accelerators can leak crucial information about the secret neural network designs through side-channels, like Electro-Magnetic (EM) emanations, power consumption, or timing. We propose and evaluate non-invasive and passive reverse engineering methods to recover NN designs deployed on GPUs through EM side-channel analysis. We employ a well-known technique of simple EM analysis and timing analysis of NN layers execution. We consider commonly used NN architectures, namely Multilayer Perceptron and Convolutional Neural Networks. We show how to recover the number of layers and neurons as well as the types of activation functions. Our experimental results are obtained on a setup that is as close as possible to a real-world device in order to properly assess the applicability and extendability of our methods. We analyze the NN execution of a PyTorch python framework implementation running on Nvidia Jetson Nano, a module computer embedding a Tegra X1 SoC that combines an ARM Cortex-A57 CPU and a 128-core GPU within a Maxwell architecture. Our results show the importance of side-channel protections for NN accelerators in real-world applications.

PUB-2020-09877 - Screen Gleaning: A Screen Reading TEMPEST Attack on Mobile Devices Exploiting an Electromagnetic Side Channel [PDF]

Zhuoran Liu, Niels Samwel, Léo Weissbart, Zhengyu Zhao, Dirk Lauret, Lejla Batina and Martha Larson

Network and Distributed System Security Symposium - NDSS 2021
We introduce screen gleaning, a TEMPEST attack in which the screen of a mobile device is read without a visual line of sight, revealing sensitive information displayed on the phone screen. The screen gleaning attack uses an antenna and a software-defined radio (SDR) to pick up the electromagnetic signal that the device sends to the screen to display, e.g., a message with a security code. This special equipment makes it possible to recreate the signal as a gray-scale image, which we refer to as an emage. Here, we show that it can be used to read a security code. The screen gleaning attack is challenging because it is often impossible for a human viewer to interpret the emage directly. We show that this challenge can be addressed with machine learning, specifically, a deep learning classifier. Screen gleaning will become increasingly serious as SDRs and deep learning continue to rapidly advance. In this paper, we demonstrate the security code attack and we propose a testbed that provides a standard setup in which screen gleaning could be tested with different attacker models. Finally, we analyze the dimensions of screen gleaning attacker models and discuss possible countermeasures with the potential to address them.

JOU-2020-314-328 - Systematic Side-Channel Analysis of Curve25519 with Machine Learning [PDF]

Léo Weissbart, Łukasz Chmielewski, Stjepan Picek and Lejla Batina

Journal of Hardware and Systems Security - HASS 2020
Profiling attacks, especially those based on machine learning, proved to be very successful techniques in recent years when considering the side-channel analysis of symmetric-key crypto implementations. At the same time, the results for implementations of asymmetric-key cryptosystems are very sparse. This paper considers several machine learning techniques to mount side-channel attacks on two implementations of scalar multiplication on the elliptic curve Curve25519. The first implementation follows the baseline implementation with complete formulae as used for EdDSA in WolfSSl, where we exploit power consumption as a side-channel. The second implementation features several countermeasures, and in this case, we analyze electromagnetic emanations to find side-channel leakage. Most techniques considered in this work result in potent attacks, and especially the method of choice appears to be convolutional neural networks (CNNs), which can break the first implementation with only a single measurement in the attack phase. The same convolutional neural network demonstrated excellent performance for attacking AES cipher implementations. Our results show that some common grounds can be established when using deep learning for profiling attacks on very different cryptographic algorithms and their corresponding implementations.

PUB-2019-1476 - On the Performance of Multilayer Perceptron in Profiling Side-channel Analysis [PDF]

Léo Weissbart

Artificial Intelligence in Hardware Security - AIHWS 2020
In profiling side-channel analysis, machine learning-based attacks nowadays offer the most powerful performance. This holds especially for techniques stemming from the neural network family: multilayer perceptron and convolutional neural networks. Convolutional neural networks are often favored as state-of-the-art results suggest better performance, especially in scenarios where targets are protected with countermeasures. Multilayer perceptron receives much less attention and researchers seem less interested in this technique, narrowing the results in the literature to comparisons with convolutional neural networks. Yet, a multilayer perceptron has a much simpler structure, which enables easier hyperparameter tuning, and hopefully, could contribute to the explainability of this neural network inner working. In this paper, we investigate the behavior of a multilayer perceptron in detail in the context of the side-channel analysis of AES. By exploring the sensitivity of multilayer perceptron hyperparameters over the performance of the attack, we aim at providing a better understanding of successful hyperparameters tuning, and ultimately, the performance of this algorithm. Our results show that MLP (with a proper hyperparameter tuning) can easily break implementations having a random delay or masking countermeasures.

PUB-2019 - A Side-Channel Attack Using Order 8 Rational Points against Curve25519 on an 8-Bit Microcontroller [PDF]

Yoshinori Uetake, Keiji Yoshimoto, Yuta Kodera, Léo Weissbart, Takuya Kusaka, Yasuyuki Nogami

Seventh International Symposium on Computing and Networking - CANDAR 2019
Among the increasing evolution of IoT devices, practical applications need reliable secure protocols to communicate with each other. A major issue for modern cryptosystems is an implementation of secure and trustworthy mechanisms to rely on. A side-channel attack against these cryptosystems may overturn the guarantee of security against conventional cyber-attacks. Elliptic curve cryptography is public-key cryptography based on elliptic curves, and one of the well-known curves is Curve25519 which is used for TLS protocols as a recommended curve. This curve is mainly implemented on limited resource devices such as microcontrollers. However, this curve poses a weakness for low-order points during a Diffie-Hellman key exchange is employed. This research demonstrates possible exploitation of a threat of order 8 rational points of Curve25519 and shows results of the side-channel attacks using order 8 rational points on an embedded system. The results indicate the order 8 rational points might be applied to key extraction as attacker sides.

PUB-2019-358 - One trace is all it takes: Machine Learning-based Side-channel Attack on EdDSA [PDF]

Léo Weissbart, Stjepan Picek and Lejla Batina

Security, Privacy and Applied Cryptography Engineering - SPACE 2019
Profiling attacks, especially those based on machine learning proved as very successful techniques in recent years when considering side-channel analysis of block ciphers implementations. At the same time, the results for implementations public-key cryptosystems are very sparse. In this paper, we consider several machine learning techniques in order to mount a power analysis attack on EdDSA using the curve Curve25519 as implemented in WolfSSL. The results show all considered techniques to be viable and powerful options. The results with convolutional neural networks (CNNs) are especially impressive as we are able to break the implementation with only a single measurement in the attack phase while requiring less than 500 measurements in the training phase. Interestingly, that same convolutional neural network was recently shown to perform extremely well for attacking the AES cipher. Our results show that some common grounds can be established when using deep learning for profiling attacks on distinct cryptographic algorithms and their corresponding implementations.

PUB-2018 - Side-Channel Attack using Order 4 Element against Curve25519 on ATMega328P [PDF]

Yoshinori Uetake, Akihiro Sanada, Takuya Kusaka, Yasuyuki Nogami, Léo Weissbart and Sylvain Duquesne

The International Symposium on Information Theory and Its Applications - ISITA 2018
With the matter of secure communication between devices, and especially for IoT devices, more and more applications need trustful protocols to communicate using public key cryptography. Elliptic curve cryptography is nowadays a very secure and efficient public key cryptography method. One of the most recent and secure curve is Curve25519 and one of its failure is attack on low-order elements during a Diffie-Hellman key exchange. This document demonstrates that an attack using an order 4 point is possible on an embedded system with a simple power analysis, pointing out every IoT using Curve255119 as a cryptographic method, a potential target to side-channel attacks.