Sequence-Aware Online Container Scheduling with Reinforcement Learning in Parked Vehicle Edge Computing

doi:10.1109/TVT.2025.3554595

科研成果详情

题名	Sequence-Aware Online Container Scheduling with Reinforcement Learning in Parked Vehicle Edge Computing
作者	Wu，Jianqiu 1; Guo，Jianxiong 2,3; Tang，Zhiqing 2; Luo，Chuanwen 4,5; Wang，Tian 2; Jia，Weijia 2,3
发表日期	2025
发表期刊	IEEE Transactions on Vehicular Technology
ISSN/eISSN	0018-9545
摘要	Intelligent vehicles, often parked for long periods, are ideally suited to serve as computational nodes to expand the Mobile Edge Computing (MEC) infrastructure, with containerization significantly enhancing the system's load balancing, self-healing, resource isolation, and security. However, fluctuations in task demand and frequent container image downloads during peak hours create high loads on containerized nodes, as multiple mobile devices offload tasks simultaneously, leading to significant processing delays. Many existing studies make the simplified assumption of predefined patterns of task arrivals, which overlooks this issue and makes suboptimal decisions. In this paper, we consider a Parked Vehicles (PVs)-extended MEC scenario, where multiple devices request services on PVs functioning as edge servers, all controlled by a central base station. Task arrivals follow observed patterns based on long-term trends, such as peak and off-peak periods, resembling realistic arrival patterns rather than predefined ones. To optimize task offloading by identifying these patterns, we propose the Sequence-Aware Task Scheduling (SATS) algorithm, which is a policy gradient-based deep reinforcement learning approach that integrates Transformer and LSTM architectures to capture patterns in time-series task arrivals and relationships between nodes in a collaborative and containerized environment, thereby enhancing the efficiency of online task scheduling. The primary objective of SATS is to optimize the task offloading policy and minimize delay and energy consumption for all devices and PVs. Extensive numerical comparisons against baselines demonstrate the effectiveness and advantages of our algorithm.
关键词	Container LSTM Mobile Edge Computing Parked Vehicles Reinforcement Learning Transformer Vehicles Edge Computing
DOI	10.1109/TVT.2025.3554595
URL	查看来源
语种	英语English
Scopus入藏号	2-s2.0-105001232809
引用统计
文献类型	期刊论文
条目标识符	https://repository.uic.edu.cn/handle/39GCC9TT/13740
专题	北师香港浸会大学
通讯作者	Guo，Jianxiong
作者单位	1.Beijing Normal-Hong Kong Baptist University,Guangdong Key Lab of AI and Multi-Modal Data Processing,Department of Computer Science,Zhuhai,519087,China 2.Beijing Normal University,Advanced Institute of Natural Sciences,Zhuhai,519087,China 3.Beijing Normal-Hong Kong Baptist University,Guangdong Key Lab of AI and Multi-Modal Data Processing,Zhuhai,519087,China 4.Beijing Forestry University,School of Information Science and Technology,Beijing,100083,China 5.Engineering Research Center for Forestry-Oriented Intelligent Information Processing of National Forestry and Grassland Administration,Beijing,100083,China
第一作者单位	北师香港浸会大学
通讯作者单位	北师香港浸会大学
推荐引用方式 GB/T 7714	Wu，Jianqiu,Guo，Jianxiong,Tang，Zhiqinget al. Sequence-Aware Online Container Scheduling with Reinforcement Learning in Parked Vehicle Edge Computing[J]. IEEE Transactions on Vehicular Technology, 2025.
APA	Wu，Jianqiu, Guo，Jianxiong, Tang，Zhiqing, Luo，Chuanwen, Wang，Tian, & Jia，Weijia. (2025). Sequence-Aware Online Container Scheduling with Reinforcement Learning in Parked Vehicle Edge Computing. IEEE Transactions on Vehicular Technology.
MLA	Wu，Jianqiu,et al."Sequence-Aware Online Container Scheduling with Reinforcement Learning in Parked Vehicle Edge Computing". IEEE Transactions on Vehicular Technology (2025).