首页|算网自智:研究进展与展望

算网自智:研究进展与展望

扫码查看
尽管算力网络可以通过网络连接并灵活地调度计算资源,但在实际落地应用中仍然缺乏智能性和灵活性.针对算力网络亟须提高智能化水平的问题,首先简要引出了算网自智的背景和概念;接着提出了一种新型的算网自智基本架构和工作机制,通过算网感知、智能决策、算力路由、自动化和自优化等关键技术实现算力网络多维资源的智能资源分配和服务调度;最后,对算网自智未来的主要研究方向和技术挑战作了展望.算网自智作为未来的必然发展趋势,可以很好地解决运维流程复杂、过度依赖人工经验、难以自适应动态变化的环境等一系列问题.
Computing-network intelligence:research progress and prospects
Although the computing power network(CPN)can connect via the Internet and flexibly schedule computing resources,it still lacks intelligence and flexibility in practical applications.To address the need for enhanced intelligence in computing power networks,the background and concept of the computing-network intelligence(CNI)were first intro-duced briefly.Then,followed by the proposal of a novel basic architecture and operational mechanism for CNI,intelli-gent resource allocation and service scheduling of multi-dimensional resources were achieved through key technologies such as computing-network perception,intelligent decision-making,computing power routing,automation,and self-optimization.Finally,future research directions and technical challenges were discussed.CNI is recognized as an inevi-table trend for the future,effectively addressing issues such as complex operational processes,excessive reliance on hu-man experience,and difficulties in adapting to dynamically changing environments.

computing-network convergencecomputing-network intelligencecomputing power networkautonomous network

朱海龙、杨帆、蒋如一、贾庆民、周晓茂、谢人超、黄韬

展开 >

北京邮电大学网络与交换技术国家重点实验室,北京 100876

紫金山实验室,江苏 南京 211111

算网融合 算网自智 算力网络 自智网络

国家自然科学基金资助项目国家自然科学基金资助项目

9226730192367104

2024

通信学报
中国通信学会

通信学报

CSTPCD北大核心
影响因子:1.265
ISSN:1000-436X
年,卷(期):2024.45(10)