Journée annuelle des Doctorants du pôle

Programme Détaillé de la journée du 11 juin - IMT-Atlantique Salle B03-133:

9:30 - 10:00 : Accueil - Cafe, Vienoiseries

10:00- 10:15 : Welcome word - Lab-STICC presentation / S. Houcke

10:15 - 10:45 Pitch :
Morlier Jérémie - Hardware Software Co-Design For Artificial Intelligence / Vincent Gripon and Mathieu Léonardon
Comtet Sylvain - Blind analysis of communications signals for satellite intelligence / S. Houcke and FX Socheleau
MezaouLi Mohammed - Real-time AI-Based Power Assisted Malware Predictor / Y. Nasser and S. Saoudi
Baltagi Abdel Salam (distance) - Real-time Power-based Hardware Trojan Detection for Automotive Applications / A. Baghdadi and Y. Nasser
Huu Nath Huynh (distance) - Model-Based Deep Learning for Signal Processing and Telecommunication

10:45 - 11:05 : Chove Thomas

Titre de la thèse : Communications acoustiques sous-marines à haute efficacité spectrale Encadrants : Christophe Laot (IMT Atlantique, Lab-STICC), Pierre-Jean Bouvet (ISEN, L@bISEN), Nicolas le Josse (Thales) Titre de la présentation : Massive MIMO Channel Model for Underwater Acoustic Communications Abstract :

In order to conceive systems able to explore the underwater environment, the design of efficient underwater wireless communications is a major stake. However, the underwater acoustic channel is considered as one of the most difficult environments for communications, due to multiple limitations (frequency band, latency, Doppler effect. . . ). Massive MIMO (multiple inputs, multiple outputs) systems, that consists of transmitting and receiving signals simultaneously from and to a large number of transducers is a promising approach to improve spectral efficiency enabling high data rate transmission.
Due to the expensiveness of at-sea experimentations, it is crucial to model the underwater acoustic channel, making possible to develop new communication techniques. We here present a channel model for underwater acoustic massive MIMO (multiple inputs, multiple outputs) systems. This model takes into account the spatial coherence of the channel, which has a major impact on the channel capacity and on the methods that will be chosen for MIMO communications.
An analysis on the impact of array correlation on the theoretical achievable rate is performed, using Shannon’s capacity: in order to take full advantage of the multiple transducers, it is necessary to keep an inter-transducers distance equal to or greater than the spatial coherence radius.

11:05 - 11:25 : Linda Kanaan

Title : Interference Management for robust satellite AIS receivers in dense maritime traffic areas

Supervisors: Prof. Karine AMIS and Prof. Frédéric GUILLOUD.

Abstract:
The main challenge in satellite AIS reception is the higher occurrence of AIS packet collisions especially in dense maritime areas. To enhance the success rate of detected AIS messages from a mixture, interference management using successive interference cancellation techniques is studied. The main focus is to improve the demodulation of the signal of interest to ensure a robust reconstruction and thus efficient interference cancellation. For this purpose, different methods exploiting the a priori information of the AIS signals are investigated to improve the performance of different demodulators: optimal coherent detector, differential detector and the coherent detector with Laurent decomposition and factor graph.

11:25 - 11:45 Melainine Cheikh
Title : Self-Configurable Receiver for Underwater Acoustic Communications using Bayesian Optimization.

Encadrants : François-Xavier Socheleau et Arnaud Jarrot de SLB.

Abstract :
Underwater communication receivers typically rely on the fine-tuning of numerous hyperparameters to perform optimally. This fine-tuning process is challenging and time consuming, and must be carried out by domain experts. Using a receiver with a decision-feedback equalizer (DFE), it is shown that finding optimal hyperparameters can be formalized as a black-box optimization problem for which an objective function is minimized. This objective function is designed to address the degenerate states of the DFE, known as “DFE hallucinations”, where the DFE produces small apparent mean square errors yet very high bit error rate. We are proposing a method for automating the tuning of hyperparameters using a tree-structured parzen estimator (TPE) approach, an algorithm belonging to the large family of Bayesian optimization algorithms. Results obtained from synthetic and real channel demonstrate the efficiency of
the method, allowing to significantly reduce the packet error rate in most of our simulations.

Break Lunch 12:00 - 13:30 We will have lunch box and if the weather is fine we could go to the Dellec for lunch.

13:30 - 14:00 : Pitch
Zenia Massinissa - Conception et décodage de codes correcteurs d’erreurs quantiques courts / G. Burel et Y. Saouter
Abdallah Abdallah - Implementation of a simplified NB-LDPC decoder / Emmanuel Boutillon, Bertrand Le Gal.
Mohammad Bazzal - Novel algorithms for ultra-high throughput trellis-based coding/decoding / C. ABDEL NOUR, C. DOUILLARD, S. WEITHOFFER
Lucas Fentanes Machado - Distributed learning on connected devices with limited resources / Amer Baghdadi, Jean-Philippe Diguet, Damith Ranasinghe, Ehsan Abbasnejad
Ardianto Wibowo - Distributed intelligence on a group of autonomous systems under resource and communication constraints / Amer Baghdadi (IMT Atlantique), Paulo E Santos (Flinders University), Karl Sammut (Flinders University), Jean-Philippe Diguet (CNRS)

14:00 - 14:20Aline Habib

Phd Title: Reconfigurable Intelligent Surfaces for millimeter wave communications in Metros

Supervisors: Charlotte Langlais, Marion Berbineau (Université Gustave Eiffel), Ammar El Falou (King Abdullah University of Science and Technology)

Abstract:
Railway environments are complex and challenging from a radio point of view due to the various obstacles that can create high radio attenuation. In addition, there is a continuous need for high data rates to serve large numbers of passengers and to meet critical communications needs. Therefore, exploring millimeter wave (mmWave) bands is crucial to meet this demand. However, the high penetration loss makes mmWave very sensitive to blocking, which limits its coverage area. One promising, efficient, and low-cost solution is the reconfigurable intelligent surface (RIS). RIS is an electromagnetic-based reconfigurable structure that turns the random nature of the propagation channel into a controllable and programmable radio environment thanks to a controlled phase shift. In this context, the thesis aims to solve the blocking problem of mmWave in tunnels, especially when a masked train blocks the direct link between the transmitter (Tx) and the receiver (Rx). A ray-tracing simulator is used to model the geometric environment realistically. The objective is to determine the feasibility of using RIS in tunnels in a particular scenario where the direct link between Tx and Rx is blocked due to a masked train and determine the optimal location of the RIS in the tunnel to obtain good performance. We, therefore, compare the power received by a mobile relay located at the top of the train when the RIS is located in the wall, the ceiling, and on the train.

14:20 -14:40Ahmad Ismail
Title : Learning to design Error Correcting Codes and Decode them Smarter
Supervisors: Raphael Le Bidan; Elsa Dupraz; Charbel Abed El Nour

14:40 - 15:00Albane Colin de Verdière
Title : Conception et architecture FPGA d'une chaîne de synchronisation performantes
Encadrants : Matthieu ARZEL, Raphaël LE BIDAN et Alain THOMAS (SAFRAN)

According to ESA, the RF band for earth observation will be saturated in 2025. Because of that, it has become necessary to use optical signal in this field. Now some questions are rising, like what will be the atmosphere impact on the signal ? What type of transmission chain will be needed ? But also what kind of signal processing will be used at reception and what will be the cost of this processing?

During this PhD, we will answer this last question. The objective of the thesis is to design a coherent DP-QPSK receiver processing chain for Earth observation and for scaling up at an acceptable hardware cost.

A 10Gbps demonstration using OOK and coherent BPSK will be taking place in 2024 with TELEO project. During this first year of the thesis, we have tried to design a high-performance synchronization chain with acceptable complexity on the trans-atmospheric channel applied to the TELEO project, which will enable us to obtain real data on which to base the increase in capacity.

15:00 - 15:20Emmanuelle Bodji
Title : Optimisation conjointe à l'aide d'un apprentissage par renforcement d'un code correcteur d'erreurs et de son décodeur pour la transmission mobile
Encadrant : Emmanuel Boutillon
Abstract : Dans le canal de communication de la 5e génération de réseaux mobiles, les codes LDPC sont choisis pour les données utilisateur (grande taille de données) et ils pourraient potentiellement être conservés pour la 6G. Les codes LDPC offrent de bonnes performances pour les grandes tailles de données, mais ne sont pas aussi performants dans le cas de petites tailles de données. Nous avons essayé de résoudre ce problème en introduisant l'intelligence artificielle pour améliorer les performances des matrices LDPC de 5G tout en conservant leur structure.

15:30 - ?? Discussions ouvertes.