Developed from a joint thesis in 2016, the collaboration between Télécom SudParis and Davidson consulting illustrates the potential of partnership research. With the AIDY-F2N project on Artificial Intelligence and Dynamic Modelling for More Flexible Future Networks, a joint laboratory (Labcom) co-funded by the French National Research Agency (ANR), the engineering school and technology consulting firm have hit a milestone and are making a significant contribution to 5G research, experimentation and development in France.
The networks of the future will have to be flexible and adapt to "real time" applications with very low latency, such as the control of robots or machine tools (Industry 4.0), whilst also using video telephony, and virtual or augmented reality, which require very high data rates.
“Thanks to virtualisation techniques,” says Hind Castel, professor at Telecom SudParis, in the Networks, Services and Telecommunications (RST) department, and researcher at the Samovar laboratory in the Methods team, “the same physical network can be shared by different logical networks with very different quality of service constraints.”
“In the Telecom SudParis laboratories, we have been working for a long time on network performance, in particular on modelling and optimization for architecture, protocol, technology and algorithmic intelligence issues. With AIDY-F2N, we plan to combine classical mathematical tools and artificial intelligence, more specifically machine learning,” adds Badii Jouaber, lecturer at Télécom SudParis in the field of wireless and mobile networks, coordinator of the AIDY-F2N project and head of the R3S team at Samovar.
“The objective is to better model and predict the behaviour of future networks and applications in order to optimize them. AI allows us to deal with certain problems that cannot be solved mathematically or to obtain faster results than with mathematical methods, which are reliable but resource-intensive. This requires theoretical work on AI methods, on their adaptation to network problems and on the optimization of their convergence.”
A fruitful collaboration
Co-founded by a Telecom SudParis alumnus, the Davidson service and consulting company has a number of partnerships with the engineering school, including a research and development focus on dynamic modelling for new generation networks. “The collaboration began with a thesis completed in 2016 at the school,” says Badii Jouaber. Since then, a trusting relationship has been established, both on a human and scientific level, and a determination to continue the collaboration has been confirmed.
The next step was to respond to a call for projects from the ANR, the second wave of Labcoms, joint laboratories bringing together public research organisations and SMEs. The success of the submission allowed Telecom SudParis and Davidson Consulting to create AIDY-F2N, a project dedicated to the dynamic modelling of next generation networks and services. “This is the first ANR Labcom at Telecom SudParis. We have just finished the 6-month start-up phase to put in place the people and the subjects, and we are now entering the 4-year productive phase. ”
The quality of this collaboration is in line with the Carnot Telecom and Digital Society label awarded to the school, and is linked notably to the balanced relationship between partners and the pooling of resources: “We are a real team” says Hind Castel. “And we also benefit from the combination of academic and industrial approaches, with theoretical scientific skills on topics such as AI on the one hand, and on the other hand knowledge of concrete use cases among operators, as well as conducting intelligence activities on standardization.”
An ambitious project
The AIDY-F2N project will consist of 2 platforms, located in Évry and Palaiseau, which are derived from the MOASIC5G open source platform, developed by Eurecom, and the THD (very high speed) platform, that already exists at Telecom SudParis which will be connected to it. The complex circuit resulting from the connection of these two platforms will make it possible to emulate services that are very demanding in terms of bandwidth (video and AR), notably by integrating equipment located at the edge of the network (edge computing). These topics “will be addressed in collaboration with our colleague Andréa Araldo, Senior Lecturer at the RST Department of Télécom SudParis, member of Samovar and collaborator on the project. This will allow us to test the flexibility of end-to-end network architectures, i.e. their ability to automatically reconfigure themselves to meet changing user needs in terms of connections. In addition to the different services and applications with different quality of service needs,” adds Hind Castel, “the experiments will also make it possible to integrate our models, mathematical tools and AI within the networks themselves and to study their effects.”
The integration of these platforms, thanks to Labcom, is an asset for other research projects or for teaching, which attests to the capacity of Telecom SudParis to develop high-level research partnerships.