A Hybrid Task Scheduling Algorithm Based on Task Clustering

Abstract

Based on the problem of task communication overhead being higher than the task execution time has a direct negative impact on the makespan of task scheduling in the current scheduling algorithms. In this paper, we propose a novel hybrid task scheduling algorithm based on task clustering (HTSTC). The algorithm uses task clustering technology to integrate tasks that meet the conditions into one cluster and uses task duplication method in the phase of processor selection. The algorithm effectively reduces the task communication overhead, and advances the start time of the successor tasks. In the layering and task priority calculation phase, HTSTC takes into account both the task communication overhead and task execution cost on different processors. The proposed algorithm effectively shortens the makespan of task scheduling. Experiments show that HTSTC has superior performance when compared to HEFT and CPOP, two of the currently leading algorithms.