Scheduling Scientific Workflows using Imperialist Competitive Algorithm

Computational grid is a growing technology in large-scale computing environment, as it provides more flexibility and easier access to distributed computing resources at a lower cost. In grid computing, applications are regarded as workflows. Scheduling complex workflows plays an important role in the performance of grid applications. In this paper we propose a meta-heuristic scheduling method based on a novel evolutionary algorithm called Imperialist Competitive Algorithm (ICA). ICA enhances the global search capability and it balances the exploration and exploitation abilities of the scheduling algorithm. Many of existing workflow scheduling algorithms can only tackle the problems with a single QoS parameter. Proposed scheduling algorithm deals with two important problems: cost optimization under deadline constraint, and execution time optimization under budget constraint. Experiment results illustrate the algorithm performance and feasibility for scheduling workflow applications and solving constrained satisfaction problems. Furthermore, solved examples represent that ICA can be effectively used to solve optimization problems with discrete variables.