Multiagent reinforcement learning and self-organization in a network of agents
Proceedings of the 6th international joint conference on Autonomous agents and multiagent systems
| Hi-index | 0.00 |
The efficiency of service discovery in distributed systems relies on the collaboration of the agents and the structure of the relations established among them. Structural relations cannot be static, agents should be able to adapt their links as the domain conditions and their interests change. This self-organization considerably improves the performance of the service discovery process. We present a self-organization mechanism that facilitates the task of decentralized service discovery and improves its efficiency in dynamic environments. Each agent has local knowledge about their neighbors and the queries received during the discovery process. With this information, each agent is able to decide when it is more appropriate to modify its structural relations with its direct neighbors and what the most suitable acquaintances to replace them are.