A high-throughput computing environment for job shop scheduling (JSP) genetic algorithms

Genetic algorithms (GAs) have long been applied to tackling scheduling problems, particularly JSP, with varying degrees of success. However despite advances in GA theory, GAs remain a computationally expensive method of approaching scheduling problems (Wall, 1996). In particular for practical scheduling problems, such as those outlined in Madureira et al. (2001) and Noivo and Ramalhinho-Lourenco, GAs require extensive resources. We propose a grid based high-throughput computing framework that utilises spare computing capacity, which is distributed across a network, to address real scheduling problems. We use Web services as a gateway to this high-throughput computing environment.