PhD defence of  Léo Weissbart

We are proud to announce the PhD defence of our member Léo Weissbart :

Date : Thursday, October 23rd 

Time : 14:30

Location : Aula Conference Centre, Mekelweg 5, Delft

Thesis title : "Side-Channel Analysis with Deep Learning: An Evergrowing Ally in Hardware Security Evaluation"

LINK to attend the defence online : https://nmclive.tudelft.nl/mediasite/play/80acc059734e470a941d58aa212bab871d

Abstract :  To ensure the physical security of devices, a rigorous evaluation campaign is applied using not only traditional Side-Channel Analysis (SCA) attacks (e.g., Differential Power Analysis (DPA)) through classical statistical analysis but also against machine learning and Deep Learning Side-Channel Analysis (DLSCA).  

This thesis investigates the use of deep learning in side-channel analysis of symmetric and public-key cryptography and other applications of side-channel analysis.   
A central insight of this work is the development of deep-learning algorithms to enhance tools for side-channel analysis evaluation.  

From this motivation, the thesis explores applications of attacks on physical devices using different cryptographic algorithms to improve the state-of-the-art of side-channel evaluation results.  

We first focus on the training of multilayer perceptron in the attack of protected implementation of AES and show high success rate for even models with few parameters, and the use of convolutional neural network on Ascon implementation that provide good candidates for the evaluation of symmetric key cryptography with DLSCA.
 
The application of convolution neural network also enhance known attacks on the symmetric key implementation of Curve25519 in which we present a single-trace attack and provide systematic approach for standard and protected implementations.  
In the security evaluation of non-cryptographic implementation, we apply deep-learning in a new testbed that evaluates TEMPEST attacks on screens of mobile devices and the use of side-channel analysis for the reverse engineering of neural networks implemented on GPU.