A new agent characterization model and grouping method for multi-agent system

Abstract

In ubiquitous environment a number of agents dynamically collaborate with each other. Therefore, it is very important to effectively model their characteristics and group them accordingly. In this paper we introduce a new agent modeling and grouping method based on the supplier/demander model. The degree of matching between the agents is modeled by UoP (Utility of Predicate), UoA (Utility of Agent), and UoC (Utility of Community). An experiment reveals that the proposed scheme drastically decreases useless data stream and inter-broker communications compared to random grouping.