Resource Allocation in Distributed Factory Scheduling

Abstract

A distributed factory scheduling model that is based on the concept of constraint-directed search within a centralized framework is presented. A set of texture measures that quantify several characteristics of the problem space being searched has been developed into heuristics that direct the searches conducted by scheduling agents. The heuristics have been used to direct the searches of a centralized activity-based scheduler, achieving good schedule quality and minimizing the likelihood of backtracking. The model discussed uses textures to direct distributed scheduling, optimizing decisions in the global search space by allowing agents to focus and direct searches in their individual search spaces. Experimental results are presented and confirm the effectiveness of the approach.