边缘计算中的算网融合趋势及研究进展

Trend of Computing-Network Convergence and Research Progress of Computing Power Network in Edge Computing

随着边缘计算业务的逐步展开,各类行业用户对于边缘计算性能的需求也日益明确,特别是在多个边缘计算节点可选的情况下,用户希望能够根据其位置到计算节点之间的时延以及用户自身业务需求情况,综合选择计算与网络资源,即将算力资源与网络资源融合后提供给上层业务调用,也就是算网融合服务。为了实现此目标,需要解决算力度量、算力感知、算力路由、算力交易以及算力编排等系列难题,因此算力网络技术应运而生。本文概要性地介绍了边缘计算推动下的算网融合趋势、面临的问题与挑战,以及由此而产生的算力网络等新型技术。

With the development of edge computing services, users in various industries have more specific requirements for edge computing performance. When multiple edge computing nodes are available, users hope to select computing and network resources according to the latency between user’s location to the computing node also their service requirements. In other words, computing and network convergence services indicates that the converged computing resources and network resources are provided to the upper-level services. To achieve this goal, a series of problems need to be solved, such as computing power measurement, computing power awareness, computing power routing, computing power transaction and computing power scheduling. Therefore, computing power network and other technologies came into being. This paper briefly introduces the trend of computing-network convergence driven by edge computing, the problems and challenges faced, and the resulting new technologies such as computing power network.