Constructing agents blackboard communication architecture based on graph theory

Abstract

In the agent blackboard communication architecture, agents do not interact with each other directly but through blackboard. The blackboard architecture includes central blackboard architecture and distributed one. In blackboard communication architecture, the location of central blackboard (or distributed sub-blackboards) and communication topology among sub-blackboards are two important issues that can influence the agent communication performance very much. However, there are few works about such issues; and in the existing agent systems, the central blackboard (or distributed sub-blackboards) is (or are) usually randomly located in the underlying network. To solve such problem, this paper presents a model for constructing agent blackboard communication architecture based on graph theory. The model computes the location of central blackboard or sub-blackboards based on median location method, and computes the communication topology among sub-blackboards based on Steiner Tree method; the model also applies graph theory to the construction of blackboard architecture's adaptation mechanism for dynamic topology and the realization of the blackboard architecture's fault-tolerance ability. At last, several case studies and simulation experiments are conducted, which prove that the presented model can construct the effective agent blackboard communication architecture.