Offre en lien avec l’Action/le Réseau : – — –/– — –
Laboratoire/Entreprise : Huawei Technologies, Paris Research Center
Durée : 2 years
Contact : jeremie.leguay@huawei.com
Date limite de publication : 2022-10-08
Contexte :
When applied to communication networks, traditional control and decision-making approaches for traffic engineering require a comprehensive knowledge of the system and of user behaviors, which is challenging in practice. Data-driven AI approaches do not have this drawback, but they offer no safety bounds and are difficult to interpret. In this context, we are conducting research to combine the best of both worlds. In particular, we are developing solutions based on Reinforcement Learning (RL) where the intelligence is shared between a global AI (at the central level) and local AIs (at the edge level) with the integration of safety bounds and the capability to interpret decisions.
Sujet :
In the context of ANR SAFE project, the postdoc researcher will focus on the design and implementation of (i) a testbed to validate AI-based control planes for traffic engineering and routing, and (ii) a semi-distributed reinforcement architecture which leverages efficient collaboration between global and local AIs with safety features. The postdoc will collaborate with partners of the project.
The work will mainly include:
• Design of a semi-distributed architecture based on RL
• Design and implementation of novel safety RL algorithms
• Implementation and evaluation using simulators or emulators (e.g., NS3, Mininet, GNS3)
• Test case generation and performance tests (scripts)
• Participation to scientific publications
Profil du candidat :
Ph.D. Degree in Networking, Computer Science, Electrical Engineering
Formation et compétences requises :
The following skills and experiences are highly desirable:
• Ph.D. Degree in Networking, Computer Science, Electrical Engineering
• Extensive experience with network simulators / emulators and RL applied to networking
• Experience with development tools: Visual Studio, SVN, GIT
• Background in Networking and RL
• English: Operational
Adresse d’emploi :
18 quai du point du jour, 92100 Boulogne Billancourt