Finding optimum neighbor for routing based on multi-criteria, multi-agent and fuzzy approach

In this paper, a hierarchical multi-agent based routing has been introduced. In dynamic situations, the previously planned entire optimum path may not stay optimum over time. Thus the approach in this paper routes a job to the next optimum neighboring node from the current position, instead of deciding over the entire path before the journey begins. Whenever there is a need to choose the next optimum node for routing or whenever a job enters the system, the master agent calls the worker agents. The worker agents run in parallel and return the results to the master agent. The worker agents are killed after their tasks are completed. The master agent takes decision based on the data delivered by the worker agents through a multi-criteria decision analysis technique known as PROMETHEE. A total of five worker agents are used for seven criteria and fuzzy approach is applied in a fuzzy shortest path algorithm performed by a worker agent and in fuzzy weight calculation in PROMETHEE. Three examples with three different kinds of networks have been used to show the effectiveness of the entire approach. The motivation of the idea introduced in this paper has come from the mating behavior of a spider known as Tarantula where the female spider sometimes eats the male spider just after mating.