摘要
Abstract
5G networks utilize artificial intelligence(AI)technologies to accelerate the intelligent development of mobile communica-tions and various vertical industries.However,the external AI services of 5G networks face the drawbacks of cross-domain scheduling dif-ficulties,as network bandwidth resources,computing resources,AI models,and data resources belong to different domains.In addition,the nodes of edge computing lack a coordination mechanism in the application of the 5G field.To address the above issues,it is proposed to introduce an AI service management network element in the 6G core network to be responsible for the unified scheduling of AI tasks,computing power,bandwidth,and AI models,forming a new architecture for external AI services.With the support of the computing pow-er network,it provides more stable and efficient services for AI algorithms.First,the AI service models and architectures in 5G networks are summarized and analyzed,and the characteristics and defects of current edge computing are also concluded.Subsequently,in re-sponse to the requirements of intelligence,low latency,and real-time performance of 6G's external AI services,a new network architec-ture is proposed.Finally,the development prospects of the integration of AI inherent in 6G and computing power networks,as well as the potential difficulties and challenges,are summarized.关键词
智慧内生/6G/AI/算力网络Key words
Native intelligence/6G/AI/Edge computing分类
信息技术与安全科学
