视觉端边云融合架构:面向超级智慧城市群演进的关键技术  

Visual end-edge-cloud fusion architecture:key technologies of future super metropolitan clusters

作　　者：纪雯[1,4] 杨哲铭 王智 郭斌 沈博[3,4] Wen JI;Zheming YANG;Zhi WANG;Bin GUO;Bo SHEN(Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China;Shenzhen International Graduate School,Tsinghua University,Shenzhen 518055,China;School of Computer Science,Northwestern Polytechnical University,Xi’an 710072,China;Peng Cheng Laboratory,Shenzhen 518055,China;School of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院计算技术研究所,北京100190 [2]清华大学深圳国际研究生院,深圳518055 [3]西北工业大学计算机学院,西安710072 [4]鹏城实验室,深圳518055 [5]中国科学院大学计算机科学与技术学院,北京100049

出　　处：《中国科学：信息科学》2024年第11期2518-2532,共15页Scientia Sinica(Informationis)

基　　金：北京市自然科学基金(批准号:L221004);国家重点研发计划(批准号:2023YFB4502805,2022YFE0125400,2022YFF0902403);国家自然科学基金(批准号:62072440)资助项目。

摘　　要：随着城市化建设进程的加速,超级智慧城市群的新格局正在形成,物联网、人工智能和大数据等技术正快速渗透至城市信息基础设施的建设,使智慧城市形成极具前景的新兴交叉研究领域.然而现有智慧城市解决方案通常是面向不同城市的定制化方案,在实际应用中缺乏对城市运行状态全面智能感知和多区域协同能力.本文提出了视觉端边云融合的架构,用于智慧城市群的协同发展.首先,将推动智慧城市信息基础设施的新质生产力凝练出算力、运力、数力、智力、视力的新五维特征.之后,通过构建层次化的城市内和城市间视觉端边云架构,解决数据互通瓶颈并助力打破城市间的信息孤岛,提升整体系统的灵活性和可靠性.同时,本文还提出了特征重构的资源共享技术、面向大规模群体的跨域协同优化技术、以及面向复杂城市网络的持续演化的关键技术,从而实现城市智慧连接,形成共通、共享、共赢的新型城市网络架构,以支持智慧城市群通过视觉数据进行跨场景、精准、动态实时的计算.本文旨在为智慧城市群的发展提供一个全面的技术架构和解决方案,以支撑超级智慧城市群的协同演进,推动智慧城市群的智能化进程.The super metropolitan cluster is rising with the acceleration of urbanization.Technologies such as the Internet of Things,artificial intelligence,and big data are rapidly permeating the construction of city infrastructure,making smart cities a promising new field of interdisciplinary research.However,existing smart city solutions are typically customized for individual cities and lack a comprehensive intelligent perception of city operation and multi-regional coordination capabilities in practical applications.This paper proposes a visual end-edge-cloud(E2C)fusion architecture for the collaborative development of smart city clusters.Firstly,the new quality productive forces promoting the information infrastructure of smart cities are distilled into five-dimensional features:computability,transmission-utility,data-utility,intelligence-utility,and vision-utility.By constructing a hierarchical visual end-edge-cloud architecture among cities and intercity connectivity,the proposed solution addresses data interworking bottlenecks and breaks down intercity information islands,enhancing the flexibility and reliability of comprehensive systems.Additionally,feature-reconstruction-based resource-sharing techniques,cross-domain collaborative optimization for large-scale groups,and evolving complex city networks are introduced.These innovations enable intelligent connectivity among city clusters,forming a city-network architecture characterized by interworking,commonality,and sharing.This architecture supports smart city clusters in performing cross-scenario,precise,and real-time dynamic computations using visual data.This paper aims to provide a comprehensive technical architecture and solution for the development of smart city clusters to support the coordinated evolution of super metropolitan clusters.

关 键 词：视觉端边云 端边云系统 智慧城市群 智能系统 视觉计算 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]