Security-Driven Heuristics and A Fast Genetic Algorithm for Trusted Grid Job Scheduling
IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Papers - Volume 01
Hi-index | 0.00 |
In this article, we present a method of improving the genetic algorithms in the task scheduling of grid environment due to the dynamic variability characteristic of grid. First, we review the crossover probability Pc and mutation probability Pm, the key parameters affecting the performance of genetic algorithm. Next, using the adaptive thinking and population fitness which represents the performance of grid resource scheduling, we present an adaptive genetic algorithm, giving a reasonable way to select crossover probability and mutation probability. It helps Pc and Pm can be adjusted automatically with the change of the population fitness; therefore we can get a good resource scheduling. Finally, we describe the results of the test, showing that the improved adaptive genetic algorithms can make the grid resource scheduling have good population fitness.