- Internship offer 2021/2022 - MCMC receivers for Faster-than-Nyquist transmission with imperfect synchronization
- Internship offer 2021/2022 - Variational Bayes phase tracking
- PhD offer 2022/2025 - Faster-than-Nyquist receivers based on Monte Carlo methods
- PhD offer 2022/2025 - Reliable transport of aeronautical (air-ground) communications
- PhD offer 2022/2025 - Secret Key Generation from shared randomness
- Waveform design: efficiency, robustness and security
We propose “information shaping” functions suited to realistic propagation channels. We target spectral or energy efficiency, while taking into account robustness and security constraints. We develop for instance wiretap channel codes, faster-than-Nyquist transmission, and multifunctional waveforms (e.g., radar and communications). We also consider machine learning to design optimal waveforms along with their decoders (e.g., auto-encoders).
- Advanced receiver techniques
We design advanced receivers to face challenging observation models. We rely on Bayesian or supervised learning estimation/detection techniques (e.g., Monte-Carlo Markov chain, neural networks). We are also interested in reduced-complexity schemes such as turbo-equalization or variational Bayesian filtering for phase tracking.
- Efficient resource allocation in a multiuser environment
We study dynamic spectral allocation/usage techniques in multiuser scenarios (cooperative or non-cooperative). A first theoretical approach includes the derivation of fundamental limits (e.g., capacity regions). Secondly, we design channel random access methods and network coding tailored to aeronautical and space systems.
Examples of on-going research topics
Spectrum sharing between radar and communications
A single waveform can fulfill simultaneously both functions to save spectral resources and ease integration of underlying electronic building blocks (e.g., shared front-end and antennas).
Physical layer security
Randomness caused by the propagation channel and by radio-frequency components can be exploited to ensure secured communications (including key generation methods), thereby supplementing traditional cryptographic techniques.
Wideband radars, usually used in radar imaging, can also be used for moving target detection. A nonambiguous mode can be obtained by simultaneously exploiting target range migration and sparsity of the scene.
Machine learning for physical layer
Neural networks can support end-to-end communication functions (or transmitter/receiver building blocks), at a very-low computational cost or for non-closed-form channel models.
- RAdio Logicielle pour la Formation (RALF)
- Vehicular radar/communication transceiver station (RadCom)
More info and contacts