Jobs

Offre de thèse en Réailté Virtuelle et Visualisation 3D pour la Cyber Sécurité

Cyber Sécurité en Réalité Virtuelle : améliorer le processus de détection d’intrusion, d’investigation et de décision via l’utilisation de techniques de visualisations 3D immersives

Il s'agit de proposer de nouveaux moyens d'aide à la détection d'intrusion basés sur la détection de signaux dits « faibles » en provenance de capteurs de types variés, allant par exemple des fichiers de traces associés à des pare-feux jusqu'à des capteurs de mouvements (présence) dans des locaux. Ces moyens seront basés sur de nouvelles métaphores 3D de représentations de ces signaux permettant à des utilisateurs de « monitorer » les éléments à surveiller. Pour leur faciliter la tâche, on leur donnera des moyens pour s'immerger dans un univers virtuel 3D dans lequel on devra proposer des moyens adaptés pour naviguer dans ces représentions qui pourront parfois s'inspirer du réel (comme la modélisation 3D de bâtiments à surveiller) mais qui devront aussi intégrer des représentations plus abstraites à ces environnements 3D de façon à caractériser visuellement ces signaux faibles.

Date de démarrage : dès que possible

Financement : Chaire Cyber CNI, 1450€ net mensuel

Conditions particulières : le candidat devra être de nationalité française

Direction de la thèse : Thierry Duval (IHSEV) et Nora Cuppens (IRIS)

Lieu : IMT Atlantique campus de Brest

 

Download : Descriptif complet du sujet (84.01 ko)

PhD Position in Robotics and Machine Learning

ROGAN - RObot manipulation learning by demonstration using Generative Adversarial Networks

A PhD position is available in the domain of learning by demonstration for the transfer of human manipulation skills to humanoid service robots for personal assistance scenarios. The thesis is in the context of a bilateral collaboration between IMT Atlantique (team HAAL of LabSTICC, Brest, France) and the University of Adelaide (Australian Institue of Machine Learning). The 3-year doctoral scholarship aims to advance the state-of-the-art in imitation learning using deep learning and generative adversarial networks (GAN) in particular.

Expected starting date of the thesis is end of 2019/early 2020, with shared stays of the candidate between France and Australia.

Link to the full details on the complete subject of the thesis as well as the candidature procedure
Download : Complete subject (213.01 ko)

Post-doc

Multiphysics modelling and simulation (electromagnetic, thermal and mechanical) for 3D-printing microwave design

This post-doctoral proposal concerns the Rapid IMPACT project funded by the DGA. The main objective of the project is to design microwave devices based on 3D-printing technology. Two sides of multiphysics will be analysed in this project. The first one concerns the system reliability with the classical tests (thermal stress, etc.). The second one, more upstream research, consists in the analysis of thermal and mechanical effects induced by the microwave signal (coupling between electromagnetism, thermic and mechanics).

The first part of the work is then dedicated to the reliability tests on a multiphysics simulation software (ANSYS-Multiphysics) and these tests will be compared to measurements. It will allow us to calibrate the software for the second step of the project.

In this second part, the key point is the board heating induced by microwave signal power. Electrical properties of materials can change with heating and size can be modified by thermal expansion. Then, the impact of thermal and mechanical effects on microwave devices will be also studied. Devices will be realized by considering several 3D-printing technologies.

European Union citizenship

Download : fiche_de_poste_multiphysics.pdf (85.22 ko)

Post-doc

Tuneable Antenna feeds based on Semi-Conductor Distributed Doped Area (ScDDA)

This post-doctoral proposal concerns the MCM-ITP An-Dro. The main objective of the project is to design tunable microwave devices based on a global approach mixing semiconductor and electromagnetism theories. A novel and innovative solution to co-design tunable microwave devices (such as switches, tunable filters, reconfigurable antennas…) has been developed [1]–[6]. The co-design method makes possible to build both the tunable element and the passive distributed component on the same silicon substrate. With a bias voltage, the tunable element which is an N+PP+
junction, acts as an ON / OFF switches and makes an electrical short-circuit in the substrate thickness. Thanks to the semiconductors properties, this codesign between the passive and active parts allows to integrate the active element in the substrate, removing the constraints related to the tuning elements that are usually reported components and drilling of via holes. In a same design flow, this co-design permits to optimize the microwave function in terms of size and quality factor. The concept permits to choose the size, position and shape of the doped areas to offer a greater flexibility in the tunable devices design. In this context, the first part of the work is dedicated to the development of the modelisation and the design flow for tunable devices mixing semiconductor and electromagnetism concepts by the use of Comsol multiphysics or combining SILVACO 3D and HFSS. Several technology such GaAs, InP and SiC will be studied. Then devices will be fabricated in the University of Sheffield. Measurements will be compared with simulations to enhance the modelisation and the design flow.

Download : fiche_de_poste_andro.pdf (99.08 ko)