Two New Parallel Algorithms for Solving the Job Shop Problem Working Paper

In this paper the job shop scheduling problem is considered. Since this well known problem is NP hard nding an optimal solution for a given instance is often very diicult. Although newest hardware solutions are used it is nearly impossible to solve even instances of moderate size (like 15 jobs processed on 15 machines) in a reasonable time. Due to this fact several heuristic algorithms have been developed in the last 25 years. But nding good results for large job shop problems becomes very time consuming while regarding a realistic time limit leads to poor results. Due to this the use of distributed systems seems to be a promising approach. Nowadays many companies have connected their personal computers to local area networks (LANs). These LANs can be used as a parallel computer. In this paper parallel concepts for two well known sequential algorithms for solving the job shop problems are presented. They are additionally designed for a practical use in business. Therefore they use no special attribute of speciic parallel systems. It can be shown that the parallel version delivers comparable results in clearly better time limits as the sequential heuristics or clearly better results in the same time limit.