Chen Ming, Qi Ping, Chu Yangyang, Wang Bo, Wang Fucheng, Cao Jie
Tongling University, Tongling, 244061, China.
Anhui Engineering Research Center of Intelligent Manufacturing of Copper-based Materials, Tongling, 244061, China.
Heliyon. 2024 Jun 10;10(12):e32399. doi: 10.1016/j.heliyon.2024.e32399. eCollection 2024 Jun 30.
Recent years, edge-cloud computing has attracted more and more attention due to benefits from the combination of edge and cloud computing. Task scheduling is still one of the major challenges for improving service quality and resource efficiency of edge-clouds. Though several researches have studied on the scheduling problem, there remains issues needed to be addressed for their applications, e.g., ignoring resource heterogeneity, focusing on only one kind of requests. Therefore, in this paper, we aim at providing a heterogeneity aware task scheduling algorithm to improve task completion rate and resource utilization for edge-clouds with deadline constraints. Due to NP-hardness of the scheduling problem, we exploit genetic algorithm (GA), one of the most representative and widely used meta-heuristic algorithms, to solve the problem considering task completion rate and resource utilization as major and minor optimization objectives, respectively. In our GA-based scheduling algorithm, a gene indicates which resource that its corresponding task is processed by. To improve the performance of GA, we propose to exploit a skew mutation operator where genes are associated to resource heterogeneity during the population evolution. We conduct extensive experiments to evaluate the performance of our algorithm, and results verify the performance superiority of our algorithm in task completion rate, compared with other thirteen classical and up-to-date scheduling algorithms.
近年来，边缘云计算由于融合了边缘计算和云计算的优势而受到越来越多的关注。任务调度仍然是提高边缘云计算服务质量和资源效率的主要挑战之一。尽管已有多项研究探讨了调度问题，但在其应用方面仍存在一些需要解决的问题，例如忽略资源异构性、仅关注一种类型的请求。因此，在本文中，我们旨在提供一种考虑异构性的任务调度算法，以提高具有截止期限约束的边缘云计算的任务完成率和资源利用率。由于调度问题具有NP难特性，我们利用遗传算法（GA）（最具代表性且应用广泛的元启发式算法之一）来解决该问题，分别将任务完成率和资源利用率作为主要和次要优化目标。在我们基于GA的调度算法中，一个基因表示其对应的任务由哪个资源处理。为了提高GA的性能，我们提出利用一种倾斜变异算子，在种群进化过程中使基因与资源异构性相关联。我们进行了大量实验来评估我们算法的性能，结果验证了与其他十三种经典和最新调度算法相比，我们的算法在任务完成率方面具有性能优势。
