A Chaotic Genetic Algorithm for Fuzzy Grid Job Scheduling

In this paper, we present a multi-constraint evolutionary algorithm based scheduler for fuzzy grid job scheduling. A chaotic genetic algorithm (CGA) is proposed to schedule jobs with uncertain operation time and flexible deadline on grid. The uncertainty is modeled by fuzzy set based execution time (FSET) model. Chaos is incorporated into standard genetic algorithm by logistic function, a simple equation involving chaos. The convergence of CGA is controlled by the three famous characteristic of logistic function: convergent, bifurcating, and chaotic during evolution. Instead of producing a single optimal solution, CGA provides a set of quasi-optimal resolutions. It´s flexible for users to make the final decision according to their preferences. In order to evaluate the performance of CGA, an entropy based statistical method is introduced. Experimental results show that in terms of searching quasi-optimal resolutions, CGA proves to be superior to the standard genetic algorithm