Négociation de Ressources et Maximisation de Bien-être Social

Centralized approaches have several important drawbacks when employed to solve the resource allocation problem within an agent community. This paper seeks to provide scalable and distributed algorithms to solve this problem, avoiding centralized approach drawbacks. We propose adaptive and “anytime” algorithms which can be applied with any kind of contact network, any range for the utility values, and for the most important social welfare notions. In that purpose, we study various agent behaviors. In each case, we show that there exists a simple behavior which leads the negotiation process to a socially optimal allocation as an emergent phenomenon, or to a socially close allocation if the need arises.