电力物联网中基于聚类的任务卸载在线优化方法

A Clustering-based Online Optimization Method for Task Offloading in Internet of Things for Electricity

随着电力物联网(electric Internet of Things,eIoT)技术的快速发展,海量电力设备在网络边缘环境中产生了丰富的数据。移动边缘计算(Mobile Edge Computing,MEC)技术在靠近终端设备的位置部署边缘代理可以有效减少数据处理延迟,这使其非常适用于延迟敏感的电力物联网场景。然而,目前的大多数研究没有考虑到部分边缘终端设备也可以作为代理设备提供计算服务,造成了资源浪费。为了充分利用移动边缘计算过程中边缘代理以及边缘终端设备的计算能力,提出了一种基于设备聚类的任务卸载方案。首先,基于分层DBSCAN(hierarchical density-based spatial clustering of applications with noise)算法,对系统中的静态和动态边缘设备进行聚类。其次,将任务卸载问题建模为多臂老虎机(Multi-Armed Bandits,MAB)模型,目标为最小化卸载延迟。再次,提出了一种基于自适应置信上限算法的算法来寻找簇内与簇间的卸载策略。最后,仿真结果表明,该方案在平均延迟方面表现出了更好的性能,并且设备簇的存活时间延长了10%~20%。

With the rapid development of electric Internet of Things(eIoT)technology,massive power devices generate rich data in network edge environments.Mobile Edge Computing(MEC)technique can effectively reduce data processing latency by deploying edge agents near to terminal devices,making it suitable for latency-sensitive eIoT scenarios.However,most current research has not considered that some edge terminal devices can also serve as agent devices to provide computing services,leading to resource waste.In order to fully utilize the computational power of edge agents and edge terminal devices in MEC,we propose a task offloading scheme based on device clustering in the eIoT scenario.Firstly,based on the hierarchical DBSCAN(hierarchical density-based spatial clustering of applications with noise)algorithm,we cluster the static and dynamic edge devices in the system.Then,we model the task offloading problem as a Multi-Armed Bandits(MAB)model with the objective of minimizing the offloading delay.Next,we propose an algorithm based on the adaptive upper confidence bound algorithm to find intra-cluster and cross-cluster offloading strategies.Finally,simulation results show that the proposed scheme exhibits better performance in terms of average delay and the survival time of device clusters is extended by 10%~20%.