Towards Real World Scheduling Systems

Abstract

It has been shown within the last two decades that local-search based heuristics can fit the needs arising in many fields of combinatorial optimization. Mainly attracted by the difficulty in NP-hardness, academic research has concentrated in very simple models. Unfortunately, these model often do not contain the essentials of problems met in the real world. In the field of scheduling, an intensive discussion comprised the design of problem representations and search operators. But the suitability of the proposed techniques has been verified solely with respect to makespan minimization problems. Although other criteria of schedules are of more practical need, there is only little research on local search oriented towards pursuing these objectives. The transfer of experiences made for the JSP therefore remains an open challenge.