DAG-Based Task Scheduling Optimization in Heterogeneous Distributed Systems

The scheduling of tasks with limited resources in cloud computing systems has been a popular research topic. One approach to addressing this problem is to employ dynamic voltage and frequency scaling (DVFS) techniques to further constrain energy consumption. In this paper, we investigate the scheduling of directed acyclic graph (DAG) tasks in heterogeneous distributed systems while considering both resource and energy constraints. We aim to decrease the duration required for task scheduling. To accomplish this, we propose a task scheduling framework that takes into account energy constraints, which provides an initial solution at the start. Additionally, we introduce a heuristic, the firefly algorithm, to further enhance the initial solution. Finally, we conduct experiments with various settings and parameters, and the experimental statistics demonstrate our suggested method exhibits a performance gain that is at least twice as significant as that of other benchmark algorithms.