Job Scheduling with Genetic Algorithm

Abstract

In this paper, we have used a Genetic Algorithm (GA) approach for providing a solution to the Job Scheduling Problem (JSP) of placing 5000 jobs on 806 machines. The GA starts off with a randomly generated population of 100 chromosomes, each of which represents a random placement of jobs on machines. The population then goes through the process of reproduction, crossover and mutation to create a new population for the next generation until a predefined number of generations are reached. Since the performance of a GA depends on the parameters like population size, crossover rate and mutation rate, a series of experiments were conducted in order to identify the best parameter combination to achieve good solutions to the JSP by balancing makespan with the running time. We found that a crossover rate of 0.3, a mutation rate of 0.15 and a population size of 100 yield the best results.