通信学报 ›› 2021, Vol. 42 ›› Issue (5): 149-163.doi: 10.11959/j.issn.1000-436x.2021067

• 学术论文 • 上一篇    

面向海洋观监测传感网的计算卸载方法研究

苏新1, 薛淏阳1, 周一青2, 朱金秀1   

  1. 1 河海大学物联网工程学院,江苏 常州 231022
    2 中国科学院计算技术研究所,北京100190
  • 修回日期:2021-01-05 出版日期:2021-05-01 发布日期:2021-05-01
  • 作者简介:苏新(1986- ),男,河北霸州人,博士,河海大学教授、硕士生导师,主要研究方向为通信与计算融合、移动通信技术、边缘/雾计算、智慧海洋等
    薛淏阳(1996- ),男,山西临汾人,河海大学硕士生,主要研究方向为海洋网络、边缘/雾计算、计算卸载等
    周一青(1975- ),女,浙江杭州人,中国科学院计算技术研究所研究员、博士生导师,主要研究方向为通信与计算融合、移动边缘计算、存储通信、干扰管控等
    朱金秀(1972- ),女,江苏常州人,博士,河海大学副教授、硕士生导师,主要研究方向为神经网络、边缘/雾计算、数字通信等
  • 基金资助:
    国家重点研发计划基金资助项目(2021YFE0105500);国家自然科学基金资助项目(61801166)

Research on computing offloading method for maritime observation monitoring sensor network

Xin SU1, Haoyang XUE1, Yiqing ZHOU2, Jinxiu ZHU1   

  1. 1 College of IoT Engineering, Hohai University, Changzhou 213022, China
    2 Institute of Computing Technology, Chinese Academy of Sciences, Beijing 100190, China
  • Revised:2021-01-05 Online:2021-05-01 Published:2021-05-01
  • Supported by:
    The National Key Research and Development Program of China(2021YFE0105500);The National Natural Science Foundation of China(61801166)

摘要:

针对海洋网络节点间计算能力与通信资源的差异性,提出了一种基于海洋网络连通概率的边缘计算节点选取方法。根据海洋近岸与远岸的网络节点密度不同,分别建立2种卸载模型。在近岸场景下,提出多节点协同的卸载方法,利用基于海洋多节点协同卸载遗传算法求解;在远岸场景下,提出可容错的卸载方法,利用基于分组交叉学习粒子群算法求解。仿真结果表明,所提方法较传统方法可降低30%以上的网络延时并可节约20%以上的网络成本,极大提高了各类海事应用的用户体验。

关键词: 海洋网络, 多接入边缘计算, 计算卸载, 遗传算法, 粒子群优化

Abstract:

Considering the differences in computing capacity and communication resources of the maritime network nodes, a maritime network connectivity probability based method was proposed for selecting edge computing service nodes.Because of the different node densities in the near-shore and far-shore scenarios, two offloading models were established accordingly.In the near-shore scenario, a multi-node cooperative offloading method was proposed by using the genetic algorithm based on maritime multi-node cooperative offloading.In the far-shore scenario, a fault-tolerant offloading method was proposed based on the particle swarm algorithm with grouping cross learning.Simulation results show that compared with conventional methods, the proposed methods save over 30% network delay and reduces about 20% network costs, which can greatly enhance the maritime user experiences.

Key words: maritime network, multi-access edge computing, computing offloading, genetic algorithm, particle swarm optimization

中图分类号: 

No Suggested Reading articles found!