A generic account of continuation-passing styles
POPL '94 Proceedings of the 21st ACM SIGPLAN-SIGACT symposium on Principles of programming languages
Multi-Agent Coordination through Coalition Formation
ATAL '97 Proceedings of the 4th International Workshop on Intelligent Agents IV, Agent Theories, Architectures, and Languages
The Contract Net Protocol: High-Level Communication and Control in a Distributed Problem Solver
IEEE Transactions on Computers
A Unification and Delegation Approach to Configure Generic Protocols for Agent Interactions
IAT '07 Proceedings of the 2007 IEEE/WIC/ACM International Conference on Intelligent Agent Technology
Task allocation via coalition formation among autonomous agents
IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 1
A modeling framework for generic agent interaction protocols
DALT'06 Proceedings of the 4th international conference on Declarative Agent Languages and Technologies
| Hi-index | 0.00 |
This paper discusses a new approach to dynamically configure generic protocols. Generic protocols are interaction patterns from which all application domain information has been removed. Configuring a generic protocol consists in refining the coarse-grained description of this protocol with the functionality an agent is endowed with for a specific application. In this process the general behaviors required of an agent to play a role in the protocol are unified with the more specific ones defined in the agent's functionality. Missing functionality are added as well when the initial functional model is incomplete: adaptation in case the unification fails. Dynamically configuring generic protocols is hindered by the "impossibility'' of dynamic adaptation, i.e., generate new functionality on the basis of the specifications of a generic protocol. In our previous research we developed a negotiation and delegation-based approach to overcome this limitation. However, our negotiation mechanism is limited both in its scope (one to one communications) and its completion in strictly one turn. In this paper we present a new alternative based on coalition formation. While the coalition is being formed, all the agents involved in the process are informed of the events which occur in its course. As well, we devise a fairly attractive reward mechanism to support the discussion which leads to the coalition formation in one turn. Finally, we draw on continuation techniques in programming languages and light weight mobility to effectively distribute the execution of the role among several agents.