Work item:
Y.NCE-DAICC
Subject/title:
Network capability enhancement for distributed artificial intelligent computing centers in NGNe
Approval process:
AAP
Type of work item:
Recommendation
Version:
New
Equivalent number:
-
Timing:
2025-10 (High priority)
Liaison:
IETF CATS
Supporting members:
China Unicom, China Telecom, University of Electronic Science and Technology of China
Summary:
Intelligent computing is a kind of computing mode based on artificial intelligence technology, mainly used to process complex, high-dimensional, dynamic, and unstructured data and issues, which plays an important role in the field of meteorological data analysis, intelligent security, autonomous driving, and so on. Artificial Intelligence Computing Center (AICC) has received widespread attention as an infrastructure for intelligent computing, more and more intelligent computing tasks are no longer limited to a single AICC, but require collaboration from multiple AICCs. Therefore, AICC needs to be interconnected to solve the problem of insufficient capacity of a single AICC.
The above scenarios put forward higher capabilities and requirements for transmission networks to serve the AICCs. The network is required to provide greater bandwidth, lower latency, high determinacy, more flexible adaptive collaboration capability with the demand of the intelligent computing service, and also to solve issues like slow feedback and packet loss in long distance transmission.
Therefore, this recommendation analyses typical scenarios, provides framework, network enhancement requirements and capabilities for distributed artificial intelligent computing centers in NGNe.
Comment:
-
Historic references: