Dynamic Choice of Agent Interaction Protocols
Proceedings of the IFIP 17th World Computer Congress - TC12 Stream on Intelligent Information Processing
Role-assignment in open agent societies
AAMAS '03 Proceedings of the second international joint conference on Autonomous agents and multiagent systems
Developing multiagent systems: The Gaia methodology
ACM Transactions on Software Engineering and Methodology (TOSEM)
Protocol Moderators as Active Middle-Agents in Multi-Agent Systems
Autonomous Agents and Multi-Agent Systems
Interaction Protocols as Design Abstractions for Business Processes
IEEE Transactions on Software Engineering
UML 2.0 and agents: how to build agent-based systems with the new UML standard
Engineering Applications of Artificial Intelligence
A dynamic joint protocols selection method to perform collaborative tasks
CEEMAS'05 Proceedings of the 4th international Central and Eastern European conference on Multi-Agent Systems and Applications
A unification-based approach to configure generic protocols into agent interaction models
International Journal of Agent-Oriented Software Engineering
A survey of flexible agent interaction approaches
Multiagent and Grid Systems
Hi-index | 0.00 |
This paper addresses the problem of dynamic protocol selection in open and heterogeneous multi-agent systems (MAS) for collaborative task performance. Protocols are interaction patterns which help agents coordinate their activities. Usually, agent designers select the protocols to be used for collaborative task performance. In open and heterogeneous MAS, such selection may lead to protocol mismatch at run time. Protocol selection mismatch may in turn result in a collaborative task not being performed, although it could have been performed had the protocols been selected properly. Our solution to this problem is to enable agents to dynamically select protocols. We first study the characteristics of protocols and the descriptions of collaborative tasks. Using these characteristics and descriptions, we devise mechanisms that enable agents to dynamically select protocols when they have to perform a task in collaboration. In this work we relax some restrictive assumptions made in prior work on protocol selection, and which may hinder collaborative task performance. In particular, we do not assume that agents trust one another. Additionally, because there is no protocol specification standard, we allow several protocol specification formalisms to coexist. Our method allows agents to select their protocols locally and individually, yet coordinate their selection with other agents' selections based on message exchange flows. Since the selection is performed locally, some errors may occur in the process. Our mechanisms provide means for detecting such errors and overcoming them. As we show, the complexity of our solution is empirically sufficiently low for practical use, and the level of success in coordinated protocol selection is very high.