移动通信2025，Vol.49Issue(3)：86-91,138,7.DOI:10.3969/j.issn.1006-1010.20250124-0001
多层次算力网络下的资源管控优化技术
Resource Management and Optimization Technology Under Multi-level Computing Networks
摘要
Abstract
The computing power network,by integrating distributed computing resources,has garnered widespread attention for improving computational efficiency,reducing energy consumption,and supporting diverse needs.This paper discusses the current development status of computing power networks and the technical challenges they face,proposing an end-edge-cloud collaborative computing power network architecture that achieves efficient management of distributed computing resources.To optimize multi-level resource allocation,a resource scheduling algorithm based on IMPALA(Importance Weighted Actor-Learner Architecture)is proposed.Experimental results show that the proposed algorithm can more effectively optimize network resources and reduce network latency compared to baseline algorithms.The above research provides significant guidance for the efficient management of future computing power network resources.关键词
算力网络/IMPALA算法/资源调度Key words
computing power network/IMPALA algorithm/resources allocation分类
信息技术与安全科学引用本文复制引用
周宗杰,赵哲,宋西斐,黄航,冯杰..多层次算力网络下的资源管控优化技术[J].移动通信,2025,49(3):86-91,138,7.基金项目
中国博士后科学基金面上项目"边缘智能驱动的多层次网络多维资源管控技术"(2024M752525) （2024M752525）
