Section: Partnerships and Cooperations

International Initiatives

AlDyNet

• Title: Algorithm for large and Dynamic Networks

• Inria principal investigator: Nicolas Nisse

• International Partner (Institution - Laboratory - Researcher):

  • Universidad Adolfo Ibañez, Santiago, Chile

    Facultad de Ingeniería y Ciencias

    Karol Suchan

• Duration: 2013 - 2015

• See also: http://team.inria.fr/coati/projects/aldynet/

• The main goal of this Associate Team is to study the structure of networks (modeled by graphs) to design both efficient distributed algorithms and reliable network topologies suitable to applications. We are interested both in large-scale (Facebook, Internet, etc.) and in smaller networks (e.g., WDM) that handle heavy traffic. More precisely, we aim at designing new techniques of distributed and localized computing to test structural properties of networks and to compute structures (e.g., decompositions) to be used in applications. Concerning the applications, we will first focus on routing and subgraph packing problems.

  There are two main objectives:

  • Find efficient localized algorithms to test certain graph properties or to prove that no such algorithms exist. We will formalize several distributed computing models and analyze which properties can and which cannot be tested in them.

  • Define graph properties – computable or approximable in distributed systems – such as structures/decompositions/representations. The driving idea is to combine several well studied graph properties in order to obtain more specific structures which we hope to be more easily computable.

  To verify the practical efficiency of our results, the designed algorithms will be implemented and compared to existing ones. For this purpose, a particular effort will be put to design and implement algorithms to generate graphs that satisfy properties of interest, in order to use them to test the algorithms.

  The originality of the proposal is to combine powerful tools of graphs theory (e.g., FPT complexity) and of combinatorial optimization (Mixed Integer Programming) with distributed computing. One challenge here is to balance between the degree of locality of desired algorithms and the relevance of properties that may be computed.

(https://gratel.labri.fr/pmwiki.php?n=Main.HomePage )

Inria FUNCAP (Inria-FAP):

ALERTE (ALgorithmes Efficaces pour les Réseaux de TElécommunications), with Pargo Team, Universidade Federal do Ceará, Brazil, 07/2011-07/2013.

Action ECOS-SUD:

ALgorithmes Distribués pour le calcul de la structure des réseaux, with Chile, 2013-2015.