|
Work item:
|
Y.Sup.ICN-DTI
|
|
Subject/title:
|
Supplement to ITU-T Y.3086: Information-centric networking: Additional capabilities and functional enhancements for distributed training and inference
|
|
Status:
|
Under study
|
|
Approval process:
|
Agreement
|
|
Type of work item:
|
Supplement
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2027-10 (High priority)
|
|
Liaison:
|
-
|
|
Supporting members:
|
Shanghai Advanced Research Institute, Chinese Academy of Sciences (SARI, CAS),
|
|
Summary:
|
As machine learning (ML) and artificial intelligence (AI) applications grow, training and inference are transitioning from centralized processing to distributed execution across terminal, edge, and cloud environments. [ITU-T Y.3086] outlines the general requirements for Information-Centric Networking (ICN) to support ML, focusing on data, model, and service life-cycle management. Although Y.3086 already touches on aspects of distributed execution, some provisions remain principle-oriented and are difficult to implement consistently in cross-node collaborative networks. Without changing the baseline framework of Y.3086, further clarifications and implementation-oriented supplements are needed, grounded in the execution-layer characteristics of distributed training and inference, so that Y.3086 can better support runtime object interactions, state exchanges, and chained execution across continuously coordinated multi-node scenarios. In distributed environments, training typically follows long-term iterative optimization centered on data aggregation and parameter synchronization, while inference forms context-driven, multi-stage collaborative execution chains. Accordingly, this Supplement builds on the ICN-ML framework of Y.3086, provides supplementary clarifications for relevant capabilities, and offers compatible, overlayable additional capabilities and functional enhancements for references.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-03-06 15:54:10
|
|
Last update:
2026-03-10 09:32:18
|
|