一种快速低耗的多维异构雾控制系统规划方法  

A Fast and Low-cost Programming Method for Multi-dimensionally Heterogeneous Fog Control Systems

作　　者：胡程 颜学明 金耀初 HU Cheng;YAN Xueming;JIN Yaochu(School of Information Science and Technology,Guangdong University of Foreign Studies,Guangzhou 510006,China;Guangzhou Key Laboratory of Multilingual Intelligent Processing,Guangzhou 510006,China;School of Engineering,Westlake University,Hangzhou 310024,China)

机构地区：[1]广东外语外贸大学信息科学与技术学院,广东广州510006 [2]广州市非通用语种智能处理重点实验室,广东广州510006 [3]西湖大学工学院,浙江杭州310024

出　　处：《信息与控制》2025年第1期79-96,共18页Information and Control

基　　金：国家自然科学基金项目(62202111);广东省基础与应用基础研究基金项目(2021A1515110754);国家留学基金项目(202307150168)。

摘　　要：对于雾环境下控制系统的低耗构建问题,受雾环境下基础设施显著差异性的硬件资源、网络通信条件及差异化价格的影响,在没有部署完所有应用并且完成系统构建之前,都难以对系统构建的总消耗代价进行直接评估与对比。而在探寻系统的构建方案时,进行应用部署的设施组合数量达到了幂指函数级。若通过对比全部的可行方案来确定出最优的低耗构建方案,随着系统规模的提升,其耗时将变得无法接受。为克服这些挑战,通过考虑系统对基础设施多方面即多维的资源需求,提出一种快速低耗的多维异构雾控制系统规划方法。首先,阐明了雾计算环境适用的多维异构雾控制系统定义,该定义下基础设施有着多方面(多维)异构的硬件资源与通信链路。其次,基于此系统定义,以满足多维服务资源及通信质量要求为目标,明确了该系统对基础设施的多维需求约束。最后,提出具体规划方法,对问题进行综合多维建模得到一个最优规划的表达形式,并通过多约束规划求解得到最优方案。通过该方法得到的最优构建方案,既确定了系统构建的基础设施配置,又明确了全部应用的具体部署安排。在具有普遍代表性案例实验下的结果表明,所提出方法以平均973 ms,同量级于一个传统线性复杂度(局部最优)方法的求解时间,快速得到普遍优于其它全部对比方法的一个系统总构建代价优化的方案;以此方案构建雾控制系统,可将总消耗代价平均降至传统/典型方法的72.5%以下,而获得平均约30%的资源盈余,可满足数量上平均超额60.65%的额外应用部署。For the problem of low-cost control system construction in fog environments,due to the significantly heterogeneous hardware resources and network communication conditions,along with the different prices of the infrastructure in fog environments,it is difficult to directly assess and compare the total construction cost until all the applications are deployed and the system is finally established.While,the number of facility combinations for deploying applications reaches a power exponential level,when searching system construction solutions.If an optimal low-cost construction solution is determined by comparing all feasible solutions,the time spend will become unbearable with the increasing scale of the system.To overcome these challenges,a fast and low-cost multi-dimensionally heterogeneous fog control system programming method is proposed by considering the system′s multi-aspect,i.e.,multi-dimensional resource requirements on facilities.First,the system definition for a multi-dimensionally heterogeneous fog control system under fog computing environments is clarified,where the facilities are heterogeneous in multi-aspect(multi-dimensional)hardware resources and communication links.Second,based on the system definition,the system′s multidimensional demand constraints on facilities are specified,with the objective of satisfying the multidimensional service resource and communication quality requirements.Finally,a specific programming method is proposed to comprehensively and multi-dimensionally model the problem into an optimal programming expression,and derive a solution through multi-constraint programming.The solution derived by this method not only determines the facility configuration for the system construction,but also specifies the specific deployment for all the applications.The experimental results under a representative case study show that,the proposed method can fast derive a total-cost-optimized system construction solution that generally outperforms all other comparison methods,with the solution

关 键 词：异构雾控制系统规划 低耗系统架构 多维资源需求感知 雾应用调度 

分 类 号：TP271[自动化与计算机技术—检测技术与自动化装置]