TY - GEN
T1 - SMCoEdge
T2 - 23rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2023
AU - Xu, Changfu
AU - Li, Yupeng
AU - Chu, Xiaowen
AU - Zou, Haodong
AU - Jia, Weijia
AU - Wang, Tian
N1 - This work is supported in part by grants from the National Natural Science Foundation of China (No. 62272117) and the Joint Foundation of Guangzhou and Universities on Basic and Applied Basic Research (202201020126), the National Key R&D Program of China (2022YFE0201400), the Beijing Natural Science Foundation (No. 4232028), the National Natural Science Foundation of China (No. 62172046, 62372047), the Special Project of Guangdong Provincial Department of Education in Key Fields of Colleges and Universities (2021ZDZX1063), the Zhuhai Basic and Applied Basic Research Foundation (2220004002619), the Joint Project of Production, Teaching and Research of Zhuhai (2220004002686, ZH22017001210133PWC, and 2220004002686), the Guangdong Key Lab of AI and Multi-modal Data Processing, BNU-HKBU United International College (UIC), Zhuhai (No. 2020KSYS007), the UIC General project (No. R0200005-22), the UIC Start-up Research Fund (No. R72021202), the Science and Technology Projects of Social Development in Zhuhai (No. 2320004000213), the Guangdong Basic and Applied Basic Research Foundation (No. 2022A1515011583 and No. 2023A1515011562), the One-off Tier 2 Start-up Grant (2020/2021) of Hong Kong Baptist University (Ref. RC-OFSGT2/20-21/COMM/002), the Startup Grant (Tier 1) for New Academics AY2020/21 of Hong Kong Baptist University, National Natural Science Foundation of China (No. 62202402), the Germany/Hong Kong Joint Research Scheme sponsored by the Research Grants Council of Hong Kong and the German Academic Exchange Service of Germany (No. GHKBU203/22), and the Hong Kong RGC Early Career Scheme (No. 22202423)
Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.
PY - 2024/2/27
Y1 - 2024/2/27
N2 - Collaborative Mobile Edge Computing (MEC) has emerged as a promising solution for low service delay in computation-intensive Internet of Things (IoT) applications. However, current approaches typically perform offline task partitioning and offload each subtask to an Edge Server (ES) for processing. This leads to varying delays in subtask processing across different ESs, resulting in a high make-span of task offloading. To address this issue, we propose a novel approach called SMCoEdge, which utilizes simultaneous multi-ES offloading to minimize the make-span of task offloading for computation-intensive IoT applications. Specifically, we formulate our problem as a mixed integer non-linear programming problem and prove its NP-hardness. We then decompose our problem into two sub-problems of multi-ES selection and task allocation, and propose a Deep Reinforcement Learning-based Simultaneous Multi-ES Offloading (DRL-SMO) algorithm to effectively solve it. Additionally, we analyze the computation complexity of DRL-SMO. Our extensive simulation results demonstrate that SMCoEdge outperforms state-of-the-art approaches by reducing make-span by 18.93% while maintaining a low offloading failure rate.
AB - Collaborative Mobile Edge Computing (MEC) has emerged as a promising solution for low service delay in computation-intensive Internet of Things (IoT) applications. However, current approaches typically perform offline task partitioning and offload each subtask to an Edge Server (ES) for processing. This leads to varying delays in subtask processing across different ESs, resulting in a high make-span of task offloading. To address this issue, we propose a novel approach called SMCoEdge, which utilizes simultaneous multi-ES offloading to minimize the make-span of task offloading for computation-intensive IoT applications. Specifically, we formulate our problem as a mixed integer non-linear programming problem and prove its NP-hardness. We then decompose our problem into two sub-problems of multi-ES selection and task allocation, and propose a Deep Reinforcement Learning-based Simultaneous Multi-ES Offloading (DRL-SMO) algorithm to effectively solve it. Additionally, we analyze the computation complexity of DRL-SMO. Our extensive simulation results demonstrate that SMCoEdge outperforms state-of-the-art approaches by reducing make-span by 18.93% while maintaining a low offloading failure rate.
KW - Edge-edge collaboration
KW - Mobile edge computing
KW - Simultaneous multi-server offloading
KW - Task allocation
UR - http://www.scopus.com/inward/record.url?scp=85187657702&partnerID=8YFLogxK
U2 - 10.1007/978-981-97-0808-6_5
DO - 10.1007/978-981-97-0808-6_5
M3 - Conference proceeding
AN - SCOPUS:85187657702
SN - 9789819708079
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 73
EP - 91
BT - Algorithms and Architectures for Parallel Processing
A2 - Tari, Zahir
A2 - Li, Keqiu
A2 - Wu, Hongyi
PB - Springer Singapore
Y2 - 20 October 2023 through 22 October 2023
ER -