Efficient Multi-Agent Coordination Using Resource-Aware Junction Trees

In this paper we address efficient decentralised coordination for cooperative multi-agent systems (framed as DCOPs) by taking intoaccount the communication and computational resources of the system. We focus on techniques that exploit structural independenceamong agents' actions to provide optimal solutions to the coordination problem, and, in particular, we use the Generalized DistributedLaw (GDL) algorithm. In this settings, we propose a novel resourceaware heuristic to build junction trees and to schedule GDL computations across the agents. Our approach aims at minimising directlythe total running time of the coordination process, rather than thetheoretical complexity of the computation, by considering computational and communication capabilities of agents.

Authors

• Nicolas Stefanovitch
• Alessandro Farinelli University of Verona
• Alex Rogers University of Southampton
• Nicholas R. Jennings

Keywords

No keywords are indexed for this paper.