|
Work item:
|
Y.DSE-LISF
|
|
Subject/title:
|
Reference architecture of data sharing and exchange based on lightweight intelligent software framework for Internet of things devices
|
|
Status:
|
Under study
|
|
Approval process:
|
AAP
|
|
Type of work item:
|
Recommendation
|
|
Version:
|
New
|
|
Equivalent number:
|
-
|
|
Timing:
|
2026-12 (Medium priority)
|
|
Liaison:
|
ITU-T SG13, SG16, SG17
|
|
Supporting members:
|
ETRI; Korea (Rep. of); Daejeon University
|
|
Summary:
|
Artificial intelligence has been growing rapidly through deep learning technology, and large amounts of training data are needed to create accurate learning models. To overcome this problem, it has become necessary to share and exchange not only training data but also the learning models (e.g., federated learning model). Additionally, federated learning may be used to collect data from multiple IoT devices and to share and collaborate a learning model with the assistance of a server via exchanging model parameters rather than directly sharing decentralized datasets, which can break data islands and meanwhile bring the advantage of data privacy protection and legal compliance. Nevertheless, the problem of cultural and social bias in the training data itself still needs to be resolved.
This draft Recommendation specifies the data (i.e, sensing data, management data and learning model) sharing and exchange principles between a server/edge system and IoT devices equipped with the lightweight intelligent software framework (LISF) that supports Internet of things (IoT) applications requiring intelligent processing and enables such processing to work on resource-limited IoT devices. This draft also identifies general requirements and provides a functional architecture for the data sharing and exchange based on the LISF.
|
|
Comment:
|
-
|
|
Reference(s):
|
|
|
Historic references:
|
|
Contact(s):
|
|
| ITU-T A.5 justification(s): |
|
|
|
|
First registration in the WP:
2024-07-17 14:50:00
|
|
Last update:
2024-07-17 14:57:07
|
