Location: INRIA Sophia Antipolis (French Riviera): http://www-sop.inria.fr/

                 Planète Project-team: http://planete.inria.fr

Duration:  1 year up to 2 years (starting as soon as possible)

Context:
The Planète project-team conducts research in the domain of networking, with an emphasis on designing, implementing, and evaluating Internet protocols and applications. The main objective of the project-team is to propose and study new architectures, services and protocols to support efficient and secure communication through the Internet. One of its research axis focuses on the design of an evaluation environment for the future Internet. We are highly involved in the elaboration of ns-3, an efficient and realistic network simulator that includes accurate models and allows integrating real applications and communication stacks. In parallel, an important part of our activities focuses on experimental evaluation, which complements theoretical modelling and simulation. We contribute to the PlanetLab/OneLab international experimental research network in the area of platforms federation, wireless extensions and benchmarking of network protocols. The work will be done within the context of the OneLab2 European project (http://www.onelab.eu/).

Summary:

Evaluation of new network protocols and architectures are at the core of networking research. This evaluation is usually performed using simulations, emulations and/or experimental platforms. Each evaluation technique has strengths and weaknesses. However, there is currently no way to combine them in a scientific experimental workflow. To achieve the integration of experimental workflows among the various evaluation platforms, it is essential to be able to reproduce network experiments and to compare the conclusions extracted from experiments run on different platforms, or on the same platform but with different conditions. The objective of this work is to settle the foundations of a new discipline called network benchmarking, which addresses how to perform reproducible and comparable experiments. For instance, to compare two experiments, one needs to identify which parameters of the environment have an impact on the benchmarking metrics used to evaluate a protocol. For a given benchmarking metric (defined by the service expected from a protocol), one needs to evaluate the impact of each parameter on this metric. As it is not possible to evaluate all parameters and all values of each parameter, it is critical to perform a statistical analysis of well chosen parameters. Extensive experiments will be required to evaluate which parameters have an impact on the considered benchmarking metrics and to perform a strong statistical analysis of the results.

To apply:

  Send resume and letters of recommendations to Arnaud Legout, Walid Dabbous and Thierry Turletti

• Email: arnaud.legout@sophia.inria.fr  dabbous@sophia.inria.fr and turletti@sophia.inria.fr .

• Address:

  INRIA, Planète Project-team

  2004, route des Lucioles