|
Work item:
|
Y.IoT-AMDM
|
|
Subject/title:
|
Metadata for AI model deployment in heterogeneous IoT environments
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2028-Q4 (Medium priority)
|
|
Liaison:
|
ITU-T SG13, ITU-T SG21, ISO/IEC JTC 1/SC 42
|
|
Supporting members:
|
Korea (Rep. of), ETRI, Daejon University
|
|
Summary:
|
On-device AI inference is increasingly adopted in Internet of things (IoT) environments, where requirements such as low latency, data privacy, and offline operation make local execution preferable. IoT environments, however, encompass a wide variety of devices with different hardware architectures, memory capacities, and operational constraints, making the determination of whether a given AI model can be deployed on a given device inherently complex and device-specific.
Existing AI model representation formats describe model structure and parameters but do not capture hardware compatibility profiles or IoT-specific operational constraints. A metadata structure that captures an AI model’s requirements alongside the corresponding IoT device capability profile provides the common information basis needed to evaluate model-device suitability across heterogeneous environments.
To address this need, this draft Recommendation specifies metadata for AI model deployment in heterogeneous IoT environments. The specification is anticipated to cover, but is not limited to, model identification and provenance, hardware compatibility, IoT-specific deployment context, and resource requirements. It is designed to be implementation-agnostic, providing a common interoperability basis irrespective of the actor that creates or consumes the metadata.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2026-05-27 13:51:06
|
|
Last update:
2026-06-16 16:47:20
|
|
