An efficient differential evolution algorithm for task scheduling in heterogeneous cloud systems

Due to the ever-growing data and computing requirements of applications, it is very challenging for cloud scheduler to guarantee the optimal solution at a reasonable time. Although varieties of heuristics have been devised to solve the task scheduling problems in heterogeneous cloud systems, the results are still unsatisfactory, especially for large applications. Evolutionary algorithms outperform heuristics in terms of the quality of the solutions, however, they are often time-consuming and need lots of computing power. To address the above problems, this paper proposes an efficient differential evolution algorithm for task scheduling problems. This algorithm extends the canonical differential evolution in three aspects of hybrid initiation population, less greedy mutation and adaptive parameter adjustment. The results of the experiments indicate that our proposed algorithm consistently produces better solutions with smaller makespan and has the advantage of rapid convergence.