Duration: between 16 and 18 months

Remuneration: depending on profile and experience, according to the University’s salary scale (from 2700 to 3600 € gross monthly). Position to be filled as soon as possible. The applications will be evaluated as soon as they arrive.

Location: IRIT laboratory, Paul Sabatier University (Toulouse, in France)

Following the ANR DATAZERO [1] project (2015-2019), this position is placed in the context of the ANR Datazero2 (2020-2024). The DATAZERO2 project (a project of the French National Research Agency) in partnership with IRIT (Toulouse), LAPLACE (Toulouse), FEMTO St (Belfort-Besancon) and the industrial partner EATON (Grenoble), aims at improving the operation and design of datacenters operated with local renewable energy sources. The postdoc will be supervised by IRIT’s SEPIA team. The SEPIA team has been working for several years on the optimization of the energy consumption of datacenters.

Research contribution:

The project aims to find solutions to operate datacenters powered on only by renewable sources. Different modules are working together to achieve this goal: an IT optimization, an electrical optimization and a negotiation module. Depending on the profile, experience and interest of the student, two research proposals could be studied:

1. working on the negotiation module. The idea of this module is to find a tradeoff between the power needed to run users tasks and the power available, under uncertainty conditions. Some leverages could be applied on IT applications (postponing, DVFS…) so that the energy budget of the execution could be changed. On the electrical part, storage capacities could also help fit the demand. Different solutions could be envisioned based on a multi-agents approach, game theory, or demand response. A solid work has already been published without considering uncertainty [2].
2. working on IT optimization. During the project, different optimization solutions have been studied by the team: offline IT scheduling under power constraints, online IT scheduling under power constraints. Uncertainties are also currently studied. In the context of this postdoc we would be interested in studying the integration of the datacenter to the city smart grid. An IT optimization could then be studied by considering the energetic mix at different times and trying to minimize the carbon footprint.

Skills, Expected abilities, one or more of the following

• Distributed systems
• Artificial Intelligence, in particular for Optimization, Game Theory, Multi-Agent System, …
• Fog/Edge computing

Application You can submit your application (CV/Cover Letter) to Georges Da Costa (georges.da-costa@irit.fr), Patricia Stolf (patricia.stolf@irit.fr) and Jean-Marc Pierson (pierson@irit.fr)

Reference

[1] Jean-Marc Pierson, Gwilherm Baudic, Stéphane Caux, Berk Celik, Georges Da Costa, Léo Grange, Marwa Haddad, Jerome Lecuivre, Jean-Marc Nicod, Laurent Philippe, Veronika Rehn-Sonigo, Robin Roche, Gustavo Rostirolla, Amal Sayah, Patricia Stolf, Minh-Thuyen Thi, Christophe Varnier. DATAZERO: DATAcenter with Zero Emission and RObust management using renewable energy. In: IEEE Access, IEEE: Institute of Electrical and Electronics Engineers, Vol. 7, july 2019. URL : http://doi.org/10.1109/ACCESS.2019.2930368

[2] Minh-Thuyen Thi, Jean-Marc Pierson, Georges Da Costa, Patricia Stolf, Jean-Marc Nicod, Gustavo Rostirolla, and Marwa Haddad. Negotiation game for joint it and energy management in green datacenters. Future Generation Computer Systems, 2019 URL : https://doi.org/10.1016/j.future.2019.11.018