ETSI IEEE AI 代理会议.pdf

1、Presented by:ETSIRaymond C.Forbes ENI ChairSeptember 2025Work in ETSI on AI;focus on AI AgentsWork in ETSI on AI;focus on AI Agents ETSI 20252The presentation will give an overview of the Work in ISG ENI on AI Agents,the new work just started,Proof of Concept project and specification of AI Agents,i

2、nvite co-operation,and the discussion in ETSI to consolidate the migration to a standardization group.Dr Ray ForbesWork in ETSI on AI;focus on AI Agents ETSI 20253List of Supporters:Rapporteur:Raymond Colin ForbesPublication:10.02.2025 https:/docbox.etsi.org/ISG/ENI/Open/ENI GR 051 DetailsENI GR 051

3、 OverviewISG ENI Group Report 051 Study on AI Agents based Next-generation Network SlicingScope:AI-native 6G core network architecture Multiple AI agents for intent processing Customized network slice generation as a use case for multi-agent collaborationArchitecture Overview:Deutsche Telekom,Future

4、wei,CNIT,Telefonica,Motorola Mobility Huawei Technologies(UK)ETSI 2025.All rights reserved.Control:The“brain”of an Agent,responsible for decision-making logic(planning,reasoning,tools selection)Tools:The set of external capabilities that an Agent can invoke.Memory:Collects and stores short-term cont

5、ext and long-term knowledge.Control bus:Mechanism for communication between internal module.Background:AI Agent architecture in ENI 051Reference:ETSI GR ENI 051 Study on AI Agents based Next-generation Network SlicingSummary:The AN Agent shows the specific workflow within the Agent.The 051 Agent arc

6、hitecture encapsulates these complex cognitive and perceptual processes within core modules such as control and memory modules.For designing a private network multi-agent system,the AN agent architecture can be employed to define the distinct agents required for the design,with the 051 Agent serving