考虑任务迁移的配电网边缘计算节点部署方法

Edge computing node deployment method for distribution network considering task migration

需求侧部署边缘计算节点能有效降低电力网络的数据传输和存储压力,提高电力业务的服务质量。目前多从电网拓扑的维度确定边缘节点的部署位置,以网格化的方法划定各节点服务范围,各节点的工作过程相互独立,因此边缘节点选址定容过程灵活性较低,同时可能造成设备计算资源的浪费。为此文中提出一种考虑任务迁移的边缘计算节点部署方法。首先,基于边缘设备特点和居民区空间特征提出一种考虑任务迁移的边缘计算架构;其次,结合居民节点空间信息、用电规律形成特征数据,利用改进的密度峰值分析算法确定边缘节点部署的数量、地址及服务范围;最后,设计启发式算法实现边缘节点间的任务迁移,保证各节点的计算资源得到均衡利用,提高系统的稳定性。以南京市某居民区为例设计仿真实验,结果表明所提边缘节点部署方法能有效降低居民节点原始数据传输成本,任务迁移算法能有效改善边缘设备计算资源使用均衡度,提升区域内边缘计算服务的执行效率。

Deploying edge computing nodes on the demand side can effectively reduce the pressure of data transmission and storage of the power network and improve the quality of electric power service.At present,the deployment location of edge nodes is mostly determined from the dimension of grid topology,and the service scope of nodes is defined by the grid method.The working process among nodes is independent.Therefore,the flexibility of the location and capacitydetermination of edge nodes is low,and it may cause waste of equipment computing resources.Therefore,an edge computing node deployment method considering task migration is proposed in this paper.Firstly,an edge computing architecture considering task migration is proposed based on the characteristics of edge devices and the spatial characteristics of residential areas.Secondly,feature data is formed according to spatial information and load characteristics of residential nodes.Then the number,address and service scope of nodes are determined by using the improved density peak analysis algorithm.Finally,a heuristic algorithm is designed to realize the task migration among the edge nodes to ensure the balanced utilization of computing resources of nodes and improve the stability of the system.A residential area in Nanjing is taken as an example to design simulation experiments.The results show that the proposed edge nodes deployment method can effectively reduce the data transmission costs of residential nodes.Also,the task migration algorithm can effectively balance the computing resources among edge devices and improve execution efficiency of edge computing services.