Skip to Main content Skip to Navigation

Modèles de formation de coalitions stables dans un contexte ad-hoc et stochastique

Abstract : An ad-hoc and stochastic context prevents : 1- the existence of a global view of the system that reflects a complete image of the deployment environment ; 2- the existence of a priori knowledge because of the lack of a centralized structure, the dynamic of the tasks and the random availability of the entities. We proposed different strategies to facilitate the dynamic stabilization of the interactions between the agents and the convergence towards better coordination states. Our conception of alliances and recommendations allows an agent to evolve independently, to dynamically identify reliable neighboring agents with whom to cooperate and to form Nash-stable or Core stable coalitions according to the requirements of the deployment environment. To face with the challenges of the correlation between local behavior of the agents and the properties of their environment, we use in an original way the Markovian models. We also focused on taking into account the interdependencies between the agents to increase their efficiency in order to optimize the imputed costs of the ad-hoc components where the agents are deployed. This led us to propose both mechanisms, the S-NRB (Sequential Non-return Broadcast) and the P-NRB (Parallel Non-return Broadcast) for distributed coordination seeking to maximize the social welfare of the agents. To highlight the intrinsic properties of our methods, our whole proposals have been studied theoretically and experimentally through our simulator
Document type :
Complete list of metadata

Cited literature [105 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, June 23, 2015 - 6:27:13 PM
Last modification on : Tuesday, June 1, 2021 - 2:08:10 PM
Long-term archiving on: : Tuesday, September 15, 2015 - 10:15:51 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01167110, version 1


Pascal François Mbissane Faye. Modèles de formation de coalitions stables dans un contexte ad-hoc et stochastique. Autre [cs.OH]. Université Claude Bernard - Lyon I; Université Cheikh Anta Diop (Dakar), 2015. Français. ⟨NNT : 2015LYO10062⟩. ⟨tel-01167110⟩



Record views


Files downloads