This chapter presents a generalized messaging infrastructure that can be used for distributed agent systems. The principle of agent feedback messaging, upon which the infrastructure is built, is presented with examples. Agent feedback messaging allows agents to function as an intelligent agent ecosystem that spans multi-node computing clusters and facilitates agent communications in a way that mimics naturally occurring biofeedback mechanisms. An implementation of agent feedback messaging, AFM, is also described. Biosimulation and a solution to the travelling salesman problem are also presented as examples.
Agent Communications And The Principle Of Feedback
Agent feedback messaging is an approach to agent communications that relies on a message transport mechanism which functions independently from the activities of the agents themselves. The idea, in a nutshell, is to remove the burden of communications management from the agents and relegate it to a reliable message delivery subsystem. At the same time, the message delivery subsystem is built to guarantee that messages between agents are delivered according to a well defined set of rules, so that message delivery is both reliable and predictable.