基于深度强化学习的物联网服务协同卸载方法

扫码查看

原文链接

万方数据
维普

中文摘要：针对边缘计算中终端算力不足、资源有限和时延较大的问题,提出一种基于深度强化学习的物联网服务协同卸载方法.通过 3 种不同的卸载方式建立时延模型,挖掘服务之间的关联关系,对关联服务进行协同卸载,加入关联服务的通信时延以建立完善的卸载时延模型,结合整体模型考虑卸载率的取值以及关联服务如何协同卸载使时延最小,从而实现服务调用时延和服务间通信时延的最小化.试验结果表明,与其他算法相比,该算法在获取最优服务卸载策略的同时,系统总服务时延能降低 20%左右.

外文标题：A collaborative service offloading approach for Internet of Things based on deep reinforcement learning

外文摘要：Aiming at the problems of insufficient terminal computing power,limited resources and large delay in edge computing,a collaborative service offloading approach for Internet of Things based on deep reinforcement learning was proposed.The delay model was established through three different offloading methods,and the association relationship between services was mined.The associ-ated services were cooperatively offloaded,and the communication delay of the associated services was added to establish a perfect offloading delay model,and the value of the offload rate and how the associated services were cooperatively offloaded to minimize the delay was combined with the overall model.Therefore,the service request delay and the communication delay between services were minimized.The experiment results showed that our approach could reduce about 20%service delay in the system than other baseline algorithms on searching the optimal service offloading strategy.

外文关键词：

edge computingservice offloadinginteracting servicecollaborative offloadingdeep reinforcement learning

作者：

曹宇慧、黄昱泽、冯北鹏、张淼、郭珍珍

展开 >

作者单位：

重庆交通大学信息科学与工程学院,重庆 400074

关键词：

边缘计算服务卸载关联服务协同卸载深度强化学习

基金：

重庆市自然科学基金资助项目重庆市教育委员会科学技术研究计划青年项目重庆市教育委员会科学技术研究计划青年项目重庆市教育委员会科学技术研究计划青年项目国家自然科学基金资助项目

项目编号：

CSTB2022NSCQ-MSX0368KJQN202200702KJQN201900708KJQN20210073862101080

出版年：

2024

DOI：

10.6040/j.issn.1672-3961.0.2022.344

山东大学学报(工学版)

山东大学

山东大学学报(工学版)

CSTPCD北大核心

影响因子：0.634

ISSN：1672-3961

年,卷(期)：2024.54(1)

参考文献量4