|
Work item:
|
Y.FN-WDI
|
|
Subject/title:
|
Requirements and framework of future networks for supporting wide-area distributed inference
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
10-2027 (Medium priority)
|
|
Liaison:
|
ITU-T SG2, SG11, SG12, IETF CATS
|
|
Supporting members:
|
China Mobile, China Telecom, China Unicom, Huawei Technologies, ZTE
|
|
Summary:
|
With the widespread application of Large Language Model (LLM) and the rapid development of emerging technologies such as AI Agent and embodied intelligence, the demand for AI inference in society shows an exponential growth. However, large-scale AI inference has extremely high requirements for computing power, latency, etc., and the requirements of users or tasks will be differentiated. The traditional single-point inference approach faces significant challenges in terms of performance, cost, and scalability. Therefore, deploying inference service in a wide-area distributed manner has become a way to address these challenges. It can efficiently integrate and coordinate computing resources, improve performance, and intelligently select the optimal path and nodes according to the requirements. This method requires fundamental evolution to future network architectures and capabilities. This Recommendation aims to specify the requirements and framework of future networks for supporting wide-area distributed inference.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2025-11-18 17:21:31
|
|
Last update:
2025-11-18 17:28:01
|
