| Popis: |
The workflow scheduling problem is one of the most important challenges in cloud computing, which should be considered by cloud data providers in data centers. Since goals other than makespan should be considered in the real-world scheduling and these goals are contradictor in most cases, the workflow scheduling problem is a NP-Hard problem. Considering the complexities of the scheduling problem, multi-objective metaheuristic algorithms are a good option for solving such problems. These algorithms help service providers find a set of optimal tradeoffs of solutions. An important criterion in finding this tradeoff is diversity in the choice of solutions. By adding this criterion to the multi-objective evolutionary algorithms, one can achieve a set of optimal solutions. Given the importance of this criterion, we extended the Black Hole heuristic algorithm and then presented a new multi-objective algorithm based on the diversity criteria for workflow scheduling in the cloud environment. The purpose of the proposed algorithm is to search for the problem space and find the non-dominated Pareto front to optimize the Resource efficiency, Makespan and Cost factors in the cloud environment. In order to achieve this goal, we not only increased the diversity of solutions, but also modified the layout type of initial solutions based on the priority of requests in the workflow, and created a more appropriate initial generation to enhance the purposefulness of our search. The results obtained from the WorkflowSim show that the proposed method improves the balanced and unbalanced workflow better than do the known algorithms of SPEA2 and NSGA2 and PBHO. |
