Modeling Distributed Scheduling via Fuzzy Constraint-Based Agent Negotiation

Abstract

This paper presents a general framework for modeling a distributed scheduling problem via fuzzy constraint-based agent negotiation. Fuzzy constraints, in this way, are used not only to represent the requirements that jobs being scheduled must satisfy, but also to specify the possibilities prescribing to what extent the solutions are suitable for scheduling to rank the solutions. Furthermore, fuzzy constraint-based agent negotiation provides a systematic method to gradually relax the requirements to generate a proposal, and then utilizes possibility functions to select an alternative that is subject to the others' acceptability. Each agent, on behalf of each entity involved in the scheduling, iteratively proposes its offers in order to gradually move toward a satisfactory schedule. The iterative nature of agent negotiation process forces the convergence between demand and offer. Thus, each agent, who is in charge of different aspects of the problem, not only distributively solves its problems to maximize its local objectives, but also works together with other agents to attain a globally beneficial schedule. Experimental results suggest that the proposed approach is focused not only on the minimization of parameters such as makespan and tardiness, but also on the economical effects to maximize the profits of the enterprise.