移动边缘计算中基于贡献度激励的端池化解决方案

来源：计算机技术与发展杂志2024年第03期北京时间：

作者：阳柳;章立群;林晓勇

单位：南京邮电大学 通信与信息工程学院,江苏 南京 210003 Author(s): YANG Liu;ZHANG Li-qun;LIN Xiao-yong School of Communication & Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China 关键词: 移动边缘计算;计算力池;端池化;分簇;激励 Keywords: mobile edge computing;computation pool;terminal pooling;clustering;emulated 分类号: TP391 DOI: 10. 3969 / j. issn. 1673-629X. 2024. 03. 013 摘要: 经典的移动边缘计算是将计算任务从云计算迁移到移动边缘,终端用户的计算任务请求可智能选择在云端和边缘处理,这都取决于强大的基站通信能力。 在基站覆盖力不足,外节点无法接入基站,乃至基站通信缺失的场景下,计算任务无法通过云计算与边缘计算进行处理。 因此,提出一种基于贡献度激励的端池化解决方案。 该方案由终端提供自身可用计算力,通过端池化帮助基站完成任务计算,针对终端的管理,提出基于算力池最大化的动态分簇算法,该算法利用不同终端作为簇首的差异性聚簇,得到综合算力池最大时的分簇方案;针对基站覆盖力不足的情况,外围节点根据其历史贡献度指标,通过移动自组织网络与内部节点连接,该激励策略能提升内部节点的连接意愿,以此提高外节点的接入率,扩宽基站覆盖范围,解决基站弱覆盖的问题。 仿真结果表明,对比其他方案,CETP 方案能在云计算与边缘计算无法实施的情况下,利用端池化过程得到最大算力池,能以最短的时延完成计算任务。 Abstract: The traditional mobile edge computing is to migrate computing tasks from cloud computing to mobile edge. The computingtask requests of end users can be intelligently?selected in the cloud and edge processing,depending on the strong communication capabilityof the base station. In the scenario where the coverage of the base station is insufficient,external nodes cannot access the base station,oreven the communication of the base station is missing, computing tasks cannot be processed through cloud computing and edgecomputing. Therefore,a contribution based incentive end pooling solution is proposed,in which the terminal provides its own availablecomputing power?and helps the base station complete task calculations through end pooling. For terminal management, a dynamicclustering algorithm based on computing power pool maximization is proposed. This algorithm utilizes different terminals as cluster headsfor differential clustering,and obtains the clustering scheme when the comprehensive computing power pool is maximum. In response tothe insufficient coverage of base stations,external nodes are connected to internal nodes through mobile ad hoc networks based on theirhistorical contribution indicators. This incentive strategy can enhance the connection willingness of internal nodes,thereby improving theaccess rate of external nodes,expanding the coverage range of base stations,and solving the problem of weak base station coverage. Thesimulation results show that compared with other schemes,　CETP can use the end pooling process to obtain the maximum computingpower pool when cloud computing and edge computing cannot be implemented,and can complete computing tasks with the shortest delay.

摘要：经典的移动边缘计算是将计算任务从云计算迁移到移动边缘,终端用户的计算任务请求可智能选择在云端和边缘处理,这都取决于强大的基站通信能力。 在基站覆盖力不足,外节点无法接入基站,乃至基站通信缺失的场景下,计算任务无法通过云计算与边缘计算进行处理。 因此,提出一种基于贡献度激励的端池化解决方案。 该方案由终端提供自身可用计算力,通过端池化帮助基站完成任务计算,针对终端的管理,提出基于算力池最大化的动态分簇算法,该算法利用不同终端作为簇首的差异性聚簇,得到综合算力池最大时的分簇方案;针对基站覆盖力不足的情况,外围节点根据其历史贡献度指标,通过移动自组织网络与内部节点连接,该激励策略能提升内部节点的连接意愿,以此提高外节点的接入率,扩宽基站覆盖范围,解决基站弱覆盖的问题。 仿真结果表明,对比其他方案,CETP 方案能在云计算与边缘计算无法实施的情况下,利用端池化过程得到最大算力池,能以最短的时延完成计算任务。

关键词：移动边缘计算;计算力池;端池化;分簇;激励