A Grouping-Based Multi-task Scheduling Strategy with Deadline Constraint on Heterogeneous Edge Computing

In heterogeneous edge computing, multiple tasks often compete for limited computing resources on the same edge server. These tasks request different edge computing services and usually have a deadline. Efficiently scheduling them is a complex and challenging problem. In this paper, we first develop a model for grouping and mapping limited edge computing resources. Then, we mathematically describe the multi-task scheduling problem with deadline constraints. Third, we propose a grouping-based multi-task scheduling strategy called GMTSS, which includes task regrouping and priority sorting, a resource-aware greedy scheduling algorithm, and a task adjusting method. Task regrouping and priority sorting are designed to balance the efficiency and fairness of scheduling multiple tasks. The greedy scheduling algorithm assigns tasks to an optimal node based on the status of resource groups. Additionally, task adjusting aims to achieve a better scheduling scheme that will meet the maximum number of deadlines or higher long-term satisfaction of system service, called LTSS. We conduct large-scale simulations, and the experimental results clearly show that our proposed GMTSS outperforms the current state-of-the-art benchmark strategy in terms of task completion rate within deadlines and LTSS. Furthermore, GMTSS performs well in terms of task completion time.