电力物联网下基于云边协同的计算任务放置算法  

Computing Task Placement Algorithm Based on Cloud-edge Collaboration Under Power IoT

作　　者：张雅洁 陆旭 李曦[1] 张鹤立[1] 粘中元 慕春芳 ZHANG Yajie;LU Xu;LI Xi;ZHANG Heli;NIAN Zhongyuan;MU Chunfang(Beijing University of Posts and Telecommunications Pan Network Wireless Communication Laboratory,Haidian District,Beijing 100876,China;Information and Communication Branch of State Grid Inner Mongolia Eastern Power Co.,Ltd.,Hohhot 010020,Inner Mongolia Autonomous Region,China)

机构地区：[1]北京邮电大学泛网无线通信实验室,北京市海淀区100876 [2]国网内蒙古东部电力有限公司信息通信分公司,内蒙古自治区呼和浩特市010020

出　　处：《电力信息与通信技术》2024年第10期38-47,共10页Electric Power Information and Communication Technology

基　　金：国家电网有限公司总部科技项目资助“面向电力专网边缘业务保障的MEC关键技术研究与应用”(SGMDXT00JSJS2100034)。

摘　　要：电力无线网具有高可靠、安全性优势,但存在频段资源有限、输变电场景基站取能较为困难等诸多不利因素,基于云边协同的计算任务放置算法进行电力无线网的优化研究具有重要意义。云计算作为一种集中式的解决方案可以提供充足的计算资源,但是电力物联网设备与云服务器通信时存在低带宽和高时延的问题。由此,研究人员提出了边缘计算的概念,综合云计算和边缘计算的优点,云边协同逐渐以互补运作的模式得到广泛应用。文章提出一种云边协同场景下计算任务放置的改进优化算法,即基于文化基因(memetic algorithm,MA)的计算任务放置算法,以最小化电力物联网设备的能耗以及电力物联网应用程序的执行时间。基于MA的计算任务放置算法分3个阶段:预调度阶段、并行应用程序的计算任务放置阶段和故障恢复阶段。通过仿真结果验证,与现有算法对比,文章所提算法的性能包括带宽、最大迭代数、决策时间等方面都得到显著提高。Power wireless networks have the advantages of high reliability and security,but there are many unfavorable factors such as limited frequency band resources and difficulty in energy extraction of base stations in power transmission and transformation scenarios.As a centralized solution,cloud computing can provide sufficient computing resources,but power Internet of Things(IoT)devices often have the problems of low bandwidth and high latency when communicating with cloud servers.Therefore,the researchers put forward the concept of edge computing,which combines the advantages of cloud computing and edge computing,and cloud-edge collaboration is gradually widely used in a complementary operation mode.In this paper,an improved optimization algorithm for computing task placement in the cloud-edge collaboration scenario is proposed.The computing task placement algorithm based on memetic algorithm(MA),to minimize the energy consumption of power IoT devices and the execution time of power IoT applications.The MA-based computing task placement algorithm is divided into three stages:the pre-scheduling phase,the computing task placement phase of parallel applications,and the fault recovery phase.Through the simulation results,compared with the existing algorithms,the performance of the proposed algorithm in this paper is significantly improved,including bandwidth,maximum number of iterations,decision time.

关 键 词：电力物联网 云边协同 计算任务放置 能耗 时延 

分 类 号：TN915.853[电子电信—通信与信息系统]