Energy-Efficient Scheduling Algorithms with Reliability Goal on Heterogeneous Embedded Systems

• Published in: 2021 IEEE Intl Conf on Parallel & Distributed Processing with Applications, Big Data & Cloud Computing, Sustainable Computing & Communications, Social Computing & Networking (ISPA/BDCloud/SocialCom/SustainCom) pp. 555 - 562
• Main Authors: Han, Yu, Hu, Wei, Liu, Jing, Gan, Yu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2021
• Subjects: Directed acyclic graph; Embedded systems; energy; Energy consumption; Fault tolerance; Fault tolerant systems; heterogeneous; reliability; Scheduling; Scheduling algorithms
• DOI: 10.1109/ISPA-BDCloud-SocialCom-SustainCom52081.2021.00082

In this paper, we are trying to find a algorithm for scheduling DAG (Directed Acyclic Graph) tasks in heterogeneous embedded systems to minimize energy consumption while meeting the reliability requirement. Like many traditional algorithms, we divide the task scheduling algorithm into two phases, the task priority calculation phase and the task allocation phase. In the task priority calculation phase, we proposes a priority calculation algorithm IOD based on the difference in task's input and output data. In the task allocation stage, we proposes a task allocation algorithm based on fault-tolerant technology of task replication and DVFS technology. Combining the two phase, we get three scheduling algorithms, IODS, IODQ and IODR. In the experimental part, we compare the performance of the algorithm proposed in this paper with existing research algorithms (EFSRG algorithm and HRRM algorithm). The analysis of experimental results shows that the IODS algorithm is a better choice.