算力网络中一种新颖的看门狗故障检测协议  

A Novel Watchdog Fault-Detection Protocol for Compute First Networking

作　　者：梁轰 冯丽[1] 徐方鑫 李光程 周郭许 Liang Hong;Feng Li;Xu Fang-xin;Li Guang-cheng;Zhou Guo-xu(Faculty of Information Technology,Macao University of Science and Technology,Macao 999078;School of Automation,Guangdong University of Technology,Guangzhou 510006,China;Guangdong-Hong Kong-Macao Joint Laboratory for Smart Discrete Manufacturing,Guangzhou 510006,China)

机构地区：[1]澳门科技大学资讯科技学院,澳门999078 [2]广东工业大学自动化学院,广东广州510006 [3]粤港澳离散制造智能化联合实验室,广东广州510006

出　　处：《广东工业大学学报》2021年第6期35-46,共12页Journal of Guangdong University of Technology

基　　金：国家自然科学基金资助项目(61872451,61872452);澳门科学技术发展基金资助项目(0098/2018/A3,0037/2020/A1,0062/2020/A2)。

摘　　要：算力网络(Compute First Networking,CFN)是最新的分布式框架,可根据计算负载和网络状态为边缘计算智能地分配计算资源。它要求实时了解本地或远程计算资源的可用状态。本文首次提出集中式故障检测协议CFNWatchdog(简称为Watchdog),它可以很好地满足CFN的要求并及时回收故障所占用的资源。然后,从理论上分析各种参数(如检测阈值、任务处理时间和网络延迟)对Watchdog性能的影响。大量的仿真实验验证了本文提出的协议的有效性和理论模型的准确性。这项研究有助于边缘计算的参数优化配置和设计更好的故障检测协议。Compute first networking(CFN)is a latest distributed framework that intelligently allocates computing resources for edge computing according to computing load and network status.It requires real-time visibility of available statuses of local or remote computing resources.To the best of our knowledge,thisis the first endeavor to propose a centralized fault-detection protocol called CFN-Watchdog to well meet this CFN requirement and timely recycle resources occupied by faults.The impact of various parameters(e.g.,detection thresholds,task processing time,and network delay)on the Watchdog performance is then theoretically analyzed.Extensive simulations verify the effectiveness of our proposed protocol and the accuracy of our theoretical model.This study is very helpful to optimize parameter configurations and better design fault-detection protocols for edge computing.

关 键 词：边缘计算 算力网络 WATCHDOG 故障检测 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]