Julien Siebert

From SimpleWiki
Revision as of 20:08, 24 November 2009 by Simpleweb (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search
Student: Julien Siebert
Title: Multi-agent based modelization and simulation for dynamic,

autonomous and ubiquitous networks technologies evaluation and control

e-mail: julien.siebert@loria.fr
Affiliation: INRIA
Supervisor: Vincent Chevrier and Laurent Ciarletta
Start: 2007
End: 2010
Funding: Centre de Recherche INRIA Nancy - Grand Est


PhD project description

Peer to peer (P2P) systems are open, distributed and heterogeneous systems where the behavior of the users can impact the global performance of the network. We showed a real need to integrate into a single coherent model both a model of the user together with a model of the P2P systems in order to accuratly study the mutual influences of the user behavior and of the features of P2P system (protocol used, transfer delays, type of data exchanged, ...). We proposed a multi-model perspective to deal with this issue : we integrate and make interact several models, one of the user behavior as a reactive agent model, one of the overlay network (as the agent environment) and one of the protocol used (as the interaction model between agents). We implemented this proposal by adapting an existing simulator (PeerfactSim) and undertook experiments to study the influence of the rate of cooperation of user and the rate of pollution of data on the functioning of the network.

The goal of this thesis is to evaluate this new (multi-model) approach as a simulation model to evaluate the impact of users' behaviour in ubiquitous networks (P2P, adHoc, mesh...). It will allow us to foresee the behaviours which can be dangerous for the network's performance and to identify those which can be beneficial. This work could also make possible to quantify the dysfunctions based on the frequency/ratio of these behaviours. It will bring, in addition to results in relation to the problematic (the impact of user's behaviour in ubiquitous networks), answers to questions which go from multi-agent based modeling to dynamic networks' applications. Results we will obtain may open new ways to use these models for an autonomous and safe management of new ubiquitous networks, and to give new evaluation metrics for these networks and their services.


External links

  • [_URL_ Homepage] of Julien Siebert
  • Publications of Julien Siebert, as indexed by DBLP