The VAADER research team of IETR Rennes is seeking for candidates for a master internship titled "Implementation of Deep-Learning Based Restoration and Interpretation Methods for Electro-Magnetic Intercepted Images".
Please find the complete proposal at: https://florianlemarchand.github.io/
All electronic devices produce electro-magnetic (EM) emanations that not only interfere with radio devices but also compromise the data handled by the information system. A third party may perform a side-channel analysis and recover the original information, hence compromising the system privacy. Such attack performed on the screen of an information system allow an attacker to reconstruct the displayed signal from tens of meters. For information systems processing sensitive data, it is important to audit the risk of compromising. The work conducted on these attacks in our team aims to enhance these audits.
When retrieving visual information from an EM signal, an important part of the original information is lost through the leakage and interception process. This loss leads to a drop of the signal to noise ratio (SNR) and a deterioration of spatial coherence into the reconstructed images, making it difficult to interpret .
Deep learning algorithms have recently revolutionized most signal processing problems.
These algorithms build from data have an extreme ability to fit complex problems.
Different trained models exist that solve image denoising and restoration problems. These models are mainly designed for well-behaved known noise models like additive white Gaussian noise (AWGN). Thus, they do not adapt well to other noise models like the one generated by the EM interception.
The VAADER team from IETR (Institut d'Electronique et de Télécommunications de Rennes) holds an expertise in image processing and embedded low energy processing. This internship will act as a support for the PhD thesis of Florian Lemarchand on image restoration in a context of images intercepted from side channel signals. The objectives of the internship are thus closely related to research objectives and will aim at obtaining novel image restoration methods.
The candidate will be provided the chance to follow the seminars and scientific animations of the VAADER team to discover state of the art works in signal processing and embedded systems.
The objective of the internship is to study novel methods to improve image restoration and interpretation in the context of intercepted data.
The missions will include:
The candidate is expected to be a master student with background in image processing and deep learning.
Python language and previous experiences with learning frameworks like PyTorch or Tensorflow are required, C++ is a plus.
(c) GdR 720 ISIS - CNRS - 2011-2020.