Postdoctoral Position in Traffic Analysis

“Intensive computing, scientific visualisation and high-speed transnational transfer of peta-scale data: application to prediction of airflow through the upper airways”

Host institution : INRIA RESO team at LIP, École Normale Supérieure de Lyon, France. Fellowship : PETAFLOW project funded by the French National Agency of Research (ANR) Duration: 12 months Starting date : anytime from now on.

Supervisor : Paulo Gonçalves (INRIA)

Context and topic : Petaflow is an international research project between french teams: Gipsa-Lab (leader), MOAIS (INRIA Grenoble) and RESO (ENS Lyon, INRIA Grenoble) and Japanese laboratories from the Universities of Osaka and of Kyoto and the NICT. The core of this project is to demonstrate the technological and economical advantages of a resources sharing system. As a use case to prove the concept, we consider a challenging biomedical application that consists in predicting the airflow in human upper airways. While the mathematical model is calculated on a high performance computing facility in Osaka, the resulting image (a peta-scale data matrix) is visualized on the Grimage platform located in Grenoble. In this work, we focus on the adequate network infrastructure able to handle the specific constraints of this distributed application: High performance computing, long-distance connection, high-speed links, large scale data transfer and time- sensitivity.

The objective of the postdoctoral work is to propose and to evaluate data transfer solutions that guarantee optimal performance with respect to the application specificities and that adapt to the network context.

Description of work : Traffic analysis. The goal of this task is to equip the link between France and Japan with a dedicated metrology device. The metrology platform MetroFlux, jointly developed by the hosting group RESO and the AIST (Japan), will be used and will allow for fine-grain capture at packet level. From these measurements, the candidate will then perform a thorough analysis of the end-to-end path, as well as the statistical analysis of the traffic produced by an interactive application exchanging massive data over long- distance, high-speed links. Statistical analysis of traffic patterns should lead to characterizing the impact of temporal and spatial constraints due to the interactivity of the application. Then, we will strive to adapt services and/or transport protocols to abide with these characteristics.

Requested skills : Statistical signal analysis (modelling, estimation, sampling theory, data fusion)

Contact : send application (PhD summary, resume, motivation and recommendation letters) to : paulo.gonç

Leave a Reply




You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Rejoindre ResCom

Pour s'inscrire à ResCom il vous suffit d'envoyer un mail de demande d’adhésion à l'adresse avec dans l'objet de votre e-mail subscribe asr.rescom Prénom Nom.

Vous pouvez aussi joindre le comité scientifique pour toute demandes/questions à partir de la page contact.

IL est aussi conseillé de s'inscrire à la liste di GDR asr-forum.

Pour plus d'information sur les listes, regardez la page "contacts".

Une aide sur sympa est aussi disponible.