|
Work item:
|
Y.OnDevice-AIoT-arch
|
|
Subject/title:
|
Functional architecture for on-device Artificial Intelligence of Things enablement
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2028-Q2 (Medium priority)
|
|
Liaison:
|
ITU-T SG13, ITU-T SG17, ISO/IEC JTC1 SC42
|
|
Supporting members:
|
ETRI, Daejeon University, Malaysian Communications and Multimedia Commission (MCMC)
|
|
Summary:
|
This draft Recommendation specifies the functional architecture for on-device AIoT enablement, establishing a framework to integrate autonomous perception and reasoning into resource-constrained IoT environments. It addresses the technical challenges of executing AI within strict hardware limits, such as constrained CPU, memory, and energy, by defining modular functional entities for local, closed-loop inference.
The architecture standardizes the lifecycle management of on-device models, including quantization and resource-aware orchestration, to ensure operational resilience and ultra-low latency independent of network connectivity. By embedding principles of privacy-by-design and providing interoperable interfaces for device-to-cloud collaboration, this draft Recommendation provides a common architecture framework for a scalable and trustworthy on-device AIoT ecosystem where intelligence is distributed efficiently at the data source.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-05-27 13:51:54
|
|
Last update:
2026-07-16 16:33:28
|
