DefinePK

Abstract

The proposed genetic algorithm could solve the problem of task scheduling with a new initialization strategy to generate the initial population and new genetic operators to ensure best task–resource mapping that preserves good characteristics of the found solutions. Genetic Algorithm for Task Scheduling (GeneTaS) uses bio-inspired genetic algorithm to find an optimal schedule and adapts new fitness function to find the suitability of task and resource pair for best allocation. The use of evolutionary operators; crossover and mutation are found to move the solution very close towards optimality. The proposed algorithm is implemented using Gridsim, a simulator for task allocation problems and tested with arbitrary task graphs that are generated using DAGitizer. In the experimental setups thousands of arbitrary task graphs are used and it is observed that the results of the proposed GeneTaS algorithm is found better than the compared scheduling algorithms, when scaled on performance metrices namely; makespan, resource utilization and speed up.