An Approach to Optimized Genetic Algorithm for Task Scheduling in Cloud Computing

Cloud computing is a nascent technology in distributed computing. There are a multitude of researches on the issue of scheduling in cloud computing. Cloud task scheduling is an NP-hard optimization problem and many meta-heuristic algorithms have been proposed to solve it. This paper presents an optimized hybrid algorithm for task scheduling based on genetic algorithm and threshold accepting method to minimize both total executing time and cost. By advantages of both algorithms, probability and speed of convergence to the optimum solution is improved. Also, this strategy avoids sinking into local optima and population diversity is increased. Furthermore, the proposed approach can be implemented on both dependent and independent tasks. By virtue of comparing proposed approach with the genetic simulated annealing and improved particle swarm optimization, the experiment results show the hybrid approach not only has better scheduling performance but also runs faster than the other algorithms in a large scale. In addition, the experimental results indicate that the proposed algorithm can substantially achieve both minimal cost and minimal time.

Keywords:

Cloud computing , distributed computing , genetic algorithm , np-hard optimization , task scheduling , threshold accepting