Affiliations: Department of Computer Science and Engineering,
University of Nebraska-Lincoln, NE, USA
Note: [] Corresponding auhtor: Xin Li, 256 Avery Hall, University of
Nebraska-Lincoln, Lincoln, NE 68588-0115, USA. Tel.: +1 402 472 4257; Fax: +1
402 472 7767; E-mail: [email protected]
Abstract: In this paper, we present a coordination approach to resource
allocation problem in multiagent systems. Agents adaptively coordinate
resources among themselves to handle resource shortage crises resulted from
events they encounter in dynamic, uncertain, real-time, and noisy environments.
The coordination approach is implemented with a hybrid negotiation mechanism.
The hybrid negotiation mechanism combines competitive and cooperative
negotiations. In competitive negotiations, agents are self-interested and
negotiate to maximize their individual performance; while in cooperative
negotiations, agents are altruistic and negotiate to find a solution to help
others. We define a hybrid negotiation model based on the
Belief-Desire-Intention (BDI) architecture of agents, and implement the model
with a specific negotiation protocol and strategy. To help agents negotiate
better, we equip agents with profiling and learning capabilities. Agents
profile others and learn the negotiation experience to make decisions on with
whom to negotiate, and how to negotiate. We have implemented a multiagent
system for coordinating the CPU resource allocation among agents based on the
hybrid negotiation mechanism and conducted a series of experiments. The
experimental results show that our coordination approach to resource allocation
is able to reduce resource shortage crises, make the multiagent system adaptive
to the variation of load, and provide efficient resource coordination among
autonomous agents. The experimental results also show that the hybrid
negotiation mechanism is stable for resource coordination in complex
environments.
