Skip to main content

Liaison statement
New Technical Reports ITU-T YSTR.Af-KBSC “Architectural Framework for Knowledge-Based Semantic Communication over Public IMT Networks” and ITU-T YSTR.SemGenAI “Framework for Semantic Communication Pipe-line with Generative AI-based Reconstruction”

Additional information about IETF liaison relationships is available on the IETF webpage and the Internet Architecture Board liaison webpage.
State Posted
Submitted Date 2025-09-03
From Group ITU-T-SG-13
From Contact tsbsg13@itu.int
To Group opsawg
To Contacts Joe Clarke <jclarke@cisco.com>
Benoît Claise <benoit@everything-ops.net>
Cc Benoît Claise <benoit@everything-ops.net>
Joe Clarke <jclarke@cisco.com>
Mahesh Jethanandani <mjethanandani@gmail.com>
Scott Mansfield <Scott.Mansfield@Ericsson.com>
Operations and Management Area Working Group Discussion List <opsawg@ietf.org>
Mohamed Boucadair <mohamed.boucadair@orange.com>
Response Contact marco.carugi@gmail.com
fuyuexia@chinamobile.com
Technical Contact marco.carugi@gmail.com
Action Holder Contacts Scott Mansfield <Scott.Mansfield@Ericsson.com>
Purpose For information
Attachments ITU-T YSTR.SemGenAI: “Framework for Semantic Communication Pipe-line with Generative AI-based Reconstruction”.
ITU-T YSTR.Af-KBSC: “Architectural Framework for Knowledge-Based Semantic Communication over Public IMT Networks”
Body
ITU-T Working Party 1/13 would like to inform ITU-T SG17, 3GPP SA1, 3GPP SA2
and IETF about the initiation of the two draft new Technical Reports
YSTR.Af-KBSC “Architectural Framework for Knowledge-Based Semantic
Communication over Public IMT Networks” and YSTR.SemGenAI “Framework for
Semantic Communication Pipe-line with Generative AI-based Reconstruction”.
These draft new Technical Reports have been initiated at the ITU-T Working
Party 1/13 plenary on 25 July 2025.

ITU-T YSTR.Af-KBSC aims to specify the framework for knowledge-based semantic
communication over public IMT networks, including requirements, functional
architecture, procedures, and security considerations. Centered on the network
entity named Semantic Knowledge Management Function (SKMF), the framework
enables the network to manage source, channel, and task knowledge bases, and to
generate and distribute customised Semantic Encoders/Decoders. This framework
supports efficient semantic representation, adaptive transmission, and
task-driven optimisation, thereby improving resource efficiency and service
experience for data-intensive applications such as XR and multi-modal
communication.

ITU-T YSTR.SemGenAI aims to present a framework for integrating a Semantic
Communication (SemCom) pipe-line with Generative AI (GenAI) powered
regeneration of the source input at the receiving end-point, while indicating
the key entities and capabilities that the underlying network service needs to
enable for realization of such a framework. It presents the interactions among
major entities including the multiple levels of Knowledge Bases (KB), KB
orchestration, local edge computing, model tuning, etc. It recommends the
requirements to be satisfied by the underlying network service. This document
also provides a few example application scenarios that benefit from such
integration of GenAI based reconstruction capability with a SemCom pipeline.

ITU-T Working Party 1/13 looks forward to keeping continued collaboration and
exchange with you on the topic of semantic communications.