← Reading listAgendaReviewdatatracker ↗ draft-an-nmrg-i2icf-cits-03 · none · 38 pp · 2026-11-02 · offline copy



Internet Research Task Force                                  B. An, Ed.
Internet-Draft                                                      KETI
Intended status: Informational                             J. Jeong, Ed.
Expires: 6 May 2027                              Sungkyunkwan University
                                                                 P. Park
                                                                 S. Jang
                                                                    KETI
                                                         2 November 2026


  An Agentic AI Framework for Software-Defined Vehicles in Cooperative
                   Intelligent Transportation Systems
                      draft-an-nmrg-i2icf-cits-03

Abstract

   This document specifies an Agentic AI Framework for orchestrating,
   managing, and monitoring AI Computing Functions (ACFs) for Software-
   Defined Vehicles (SDVs) in Cooperative Intelligent Transportation
   Systems (C-ITS).  AI agents as ACFs communicate with each other via
   the Agent-to-Agent (A2A) protocol.  SDVs decouple vehicular functions
   from underlying hardware and expose them as software-driven,
   continuously updatable services; this architectural shift makes them
   natural participants in an agentic control plane that spans vehicles,
   roadside infrastructure, and mobile networks.  AI agents are co-
   located with each functional entity of the C-ITS architecture (e.g.,
   C-ITS Center, Government Public Center, C-ITS Infra, Mobile Network
   Operator (MNO), Roadside Unit (RSU), SDV, and Vulnerable Road User
   (VRU)) and interact over an A2A protocol overlay to advertise
   capabilities, discover peers, negotiate resources, and coordinate the
   placement and execution of ACFs across multi-domain networks.  In the
   context of Vehicle-to-Everything (V2X) communications, this agentic
   overlay enables efficient management of Vehicle-to-Vehicle (V2V) and
   Vehicle-to-Infrastructure (V2I) communications and their integration
   with C-ITS by leveraging in-network computing to optimize real-time
   communication, streamline traffic management, and enhance data
   processing and security services at the network edge.  The proposed
   framework aligns with the up-to-date agentic AI and A2A applicability
   to network management, and adapts these concepts to the specific
   operational and safety requirements of SDV-centric C-ITS deployments.

Requirements Language

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in RFC 2119 [RFC2119].





An, et al.                 Expires 6 May 2027                   [Page 1]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at https://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   This Internet-Draft will expire on 6 May 2027.

Copyright Notice

   Copyright (c) 2026 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents (https://trustee.ietf.org/
   license-info) in effect on the date of publication of this document.
   Please review these documents carefully, as they describe your rights
   and restrictions with respect to this document.  Code Components
   extracted from this document must include Revised BSD License text as
   described in Section 4.e of the Trust Legal Provisions and are
   provided without warranty as described in the Revised BSD License.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   3
   2.  Agentic AI Framework and Interfaces for SDV in C-ITS  . . . .   5
     2.1.  Agentic AI Framework for SDV, C-ITS, and MNO
           Networking  . . . . . . . . . . . . . . . . . . . . . . .   5
     2.2.  Agentic AI Interfaces . . . . . . . . . . . . . . . . . .  11
   3.  Use Cases for Agentic AI in SDV and C-ITS . . . . . . . . . .  12
   4.  A2A Use Cases for SDV-Centric C-ITS Integration . . . . . . .  14
   5.  Agent Cards for SDVs and C-ITS Entities . . . . . . . . . . .  17
   6.  YANG-based Structured Data for A2A Communication in SDV and
           C-ITS . . . . . . . . . . . . . . . . . . . . . . . . . .  19
   7.  Operational Considerations  . . . . . . . . . . . . . . . . .  30
     7.1.  Agent Skills as Expertise Expansion . . . . . . . . . . .  30
     7.2.  Knowledge Base as Ground Truth Data . . . . . . . . . . .  31
     7.3.  Event-driven A2A for Safety Signalling  . . . . . . . . .  31
     7.4.  Scalability . . . . . . . . . . . . . . . . . . . . . . .  31



An, et al.                 Expires 6 May 2027                   [Page 2]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


     7.5.  Latency and Performance Constraints . . . . . . . . . . .  31
     7.6.  Reliability and Failure Handling  . . . . . . . . . . . .  32
     7.7.  Agent Lifecycle and Resource Management . . . . . . . . .  33
     7.8.  Inter-Domain Operational Challenges . . . . . . . . . . .  33
   8.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .  33
   9.  Security Considerations . . . . . . . . . . . . . . . . . . .  33
   10. References  . . . . . . . . . . . . . . . . . . . . . . . . .  34
     10.1.  Normative References . . . . . . . . . . . . . . . . . .  34
     10.2.  Informative References . . . . . . . . . . . . . . . . .  35
   Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . .  37
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  37

1.  Introduction

   This document proposes an Agentic AI framework for coordinating AI
   Computing Functions (ACFs) across Software-Defined Vehicles (SDVs),
   roadside infrastructure, and mobile networks in Cooperative
   Intelligent Transportation Systems (C-ITS).  An ACF is a processing
   function that uses AI techniques, for example perception inference,
   sensor fusion using an AI model, or traffic prediction.  An AI agent
   is the entity that interprets an intent, selects and coordinates
   functions or tools, and monitors the resulting task.  An agent and an
   ACF are therefore distinct roles.

   In-network and edge computing provide relevant deployment and
   resource-management foundations.  Prior work describes use cases for
   In-network Computing Functions called ACFs
   [I-D.jeong-nmrg-agentic-ai-framework][I-D.yao-tsvwg-cco-problem-statement-and-usecases][I-D.irtf-coinrg-use-cases].
   However, computation inside a network is not necessarily AI
   computation.  Message forwarding, deterministic aggregation, network-
   slice configuration, and key rotation can remain conventional
   supporting services or tools invoked by an agent.  This document does
   not classify them as ACFs solely because they participate in an
   agentic workflow.

   The proposed framework addresses registration, configuration,
   orchestration, and monitoring of ACFs and their supporting services
   across vehicles and infrastructure.  The Interface to Network
   Security Functions (I2NSF) framework [RFC8329] provides relevant
   management concepts for Network Security Functions (NSFs), but an AI
   agent's coordination role is distinct from the function that it
   invokes.  The following architecture applies agent-to-agent
   coordination to C-ITS and MNO integration.  The use cases describe
   intended behaviors, not measured performance or deployment results.

   A distinguishing focus of this document is the Software-Defined
   Vehicle (SDV) as the primary endpoint of the Agentic AI framework.
   In an SDV, traditionally hardware-bound vehicular functions (e.g.,



An, et al.                 Expires 6 May 2027                   [Page 3]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   perception, planning, control, infotainment, and diagnostics) are
   realized as software components running on a consolidated, zonal
   compute platform and are continuously updated over the air.  This
   software-defined nature aligns naturally with the agentic AI model:
   each SDV can host one or more AI agents that expose vehicle
   capabilities (e.g., compute, sensing, connectivity, and safety
   classes) as A2A skills, negotiate ACF placement with peer SDV, RSU,
   MNO, and Center agents, and dynamically adapt to changing road,
   network, and mission contexts.  Conversely, the surrounding C-ITS and
   MNO infrastructure must evolve from static, message-based V2X
   delivery toward an intent-driven, agent-mediated control plane that
   can match the update cadence and behavioral diversity of fleets of
   SDVs.  The Agentic AI framework defined here treats SDVs as first-
   class agentic participants rather than passive V2X clients.

   In C-ITS, distributed agents across vehicles, Road-Side Units (RSUs),
   and traffic management backends increasingly need to collaborate
   directly in an _Agent-to-Agent (A2A)_ manner for capability
   advertisement, peer discovery, task delegation, and state exchange.
   Emerging IETF work on Artificial Intelligence (AI) agent protocols
   formalizes such A2A interactions (e.g., messaging, capability
   schemas, and discovery) [I-D.rosenberg-ai-protocols], while the IRTF
   NMRG explores applicability of A2A to multi-domain network management
   and automation
   [I-D.jeong-nmrg-agentic-ai-framework][I-D.yang-nmrg-a2a-nm].  This
   approach provides a standards-based application/control-plane overlay
   for ACFs: vehicle, infrastructure, and cloud agents can discover and
   configure ACFs on programmable network devices, coordinate V2V/V2I
   compute placement, and exchange low-latency signals for admission
   control, safety services, and analytics, thereby reducing end-to-end
   delay and network overhead while improving interoperability and
   manageability in C-ITS deployments.

   The A2A paradigm itself has been actively discussed across recent
   IRTF NMRG meetings (e.g., IETF 124 and IETF 125), where a body of
   work on agentic AI, Large Language Model (LLM)-assisted operations,
   and agent-mediated control planes for networks has emerged.
   Representative directions include problem statements and motivations
   for agentic AI in network management [I-D.hong-nmrg-agenticai-ps],
   agentic AI architectural principles for autonomous networks
   [I-D.jadoon-nmrg-agentic-ai-autonomous-networks], agentic network
   architectures supporting agent interconnection and multi-level
   inference [I-D.chuyi-nmrg-agentic-network-inference], agent-driven
   Network Digital Twin (NDT) architectures
   [I-D.wmz-nmrg-agent-ndt-arch], agentic sensing and network
   optimization [I-D.bernardos-nmrg-agentic-network-optimization], and
   the applicability of model and tool-access protocols such as the
   Model Context Protocol (MCP) to network management



An, et al.                 Expires 6 May 2027                   [Page 4]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   [I-D.yang-nmrg-mcp-nm].  In parallel, industry initiatives (e.g.,
   Google's A2A protocol) have contributed reference patterns for agent
   identity, capability cards (Agent Cards), task lifecycles, and secure
   peer-to-peer messaging that can be adopted as the application-layer
   substrate of an A2A overlay for C-ITS.

   Building on these trends, this document positions A2A as the primary
   coordination layer of the Agentic AI framework for SDV-centric C-ITS.
   Specifically, each major entity in the architecture (e.g., Center
   Cloud agents, C-ITS Infra agent, MNO agent, RSU agent, SDV agents,
   and VRU agent) hosts an AI agent that exposes standardized capability
   descriptions and exchanges intent-based requests over A2A.  SDV
   agents in particular act as the bridge between in-vehicle software
   stacks (e.g., zonal controllers, AI/ADAS workloads, and IVN buses)
   and the external A2A overlay, so that vehicle-side software updates
   and capability changes are immediately reflected in the agent's
   advertised Agent Card.  The Agentic AI interfaces (from I1 to I10)
   defined in this document are realized as A2A interactions between
   these agents, while the underlying transport, security, and policy
   enforcement continue to rely on existing C-ITS, V2X, and mobile
   network mechanisms.  Therefore, this split keeps the agentic control
   plane aligned with the NMRG agentic-AI direction while preserving
   compatibility with deployed C-ITS, SDV, and MNO infrastructures.

2.  Agentic AI Framework and Interfaces for SDV in C-ITS

   This section presents the detailed design of the Agentic AI framework
   and interfaces for Software-Defined Vehicles (SDVs) in C-ITS and MNO
   networking.

2.1.  Agentic AI Framework for SDV, C-ITS, and MNO Networking

   Figure 1 shows the Agentic AI framework for SDV, C-ITS, and MNO
   networking.  In this framework, every major functional entity hosts
   an AI _Agent_ that participates in an A2A overlay.  SDV agents are
   first-class peers of infrastructure and operator agents: they
   advertise their in-vehicle software capabilities, sensing and compute
   resources, supported autonomy levels, and ACF requirements through
   signed Agent Cards.  The A2A overlay provides capability
   advertisement (e.g., Agent Cards), peer discovery, intent/task
   delegation, and state exchange among agents, and is used to register,
   configure, monitor, and coordinate ACFs across the architecture.  The
   A2A protocol runs as an application/control-plane overlay on top of
   the existing I1 to I10 interfaces; wired interfaces are typically
   depicted with solid links and wireless or A2A signalling with dashed
   links.





An, et al.                 Expires 6 May 2027                   [Page 5]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   * Central Cloud: A system that comprehensively controls the entire
   C-ITS (Cooperative Intelligent Transport Systems) environment.  It
   manages information from various C-ITS centers, including regional
   centers and highway centers, and facilitates and oversees the
   connection between C-ITS data from the Government Public Center and
   end users.  Additionally, it provides security functions through an
   integrated cybersecurity system.  Within the Central Cloud, two
   cooperating agents (a C-ITS Center Agent and a Government Public
   Center Agent) participate in the A2A overlay to negotiate cross-
   domain data sharing and to delegate ACF configuration requests
   downward to the C-ITS Infra and MNO agents.

   * C-ITS Center: The C-ITS Center is a comprehensive term that
   encompasses both the Region Center and the Highway Center.  It serves
   as the central hub for managing and coordinating intelligent
   transportation systems across various environments, including urban
   regions and highways.  By integrating data from Region Centers and
   Highway Centers, the C-ITS Center ensures efficient traffic
   management, real-time data processing, and seamless communication
   between infrastructure and connected or autonomous vehicles.

   * Region Center: The Region Center refers to local centers
   established at key locations.  These regional centers are connected
   to Roadside Units (RSUs) and function as one of the C-ITS Centers.
   Each regional C-ITS center collaborates with the Government Public
   Center to share collected data, ensuring seamless integration and
   data exchange between local infrastructure and centralized management
   systems.

   * Highway Center: The Highway Center operates similarly to the Region
   Center but is managed separately due to the unique characteristics of
   highways, which span multiple regions rather than being confined to a
   single city.  Given the higher traffic volume on highways compared to
   regular roads, there is a significant increase in data generation,
   necessitating dedicated network management for highway environments.
   Highways are equipped with a greater number of RSUs than general
   roads, enabling the delivery of critical information to autonomous
   vehicles.  As a result, the Highway Center focuses on managing areas
   that require more real-time processing to support safe and efficient
   autonomous driving.

   * Government Public Center: The Government Public Center is a C-ITS
   information provision system managed by the government.  Due to the
   nature of road traffic infrastructure, it is challenging for private
   companies to manage this data effectively, and concerns over
   reliability make it difficult for users to utilize privately managed
   data.  The Government Public Center ensures the delivery of highly
   reliable, government-provided data to users, enabling them to



An, et al.                 Expires 6 May 2027                   [Page 6]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   effectively utilize infrastructure-based information.  It oversees
   the provision and management of trustworthy data essential for safe
   and efficient transportation systems.

   * C-ITS Data Linkage System: The C-ITS Data Linkage System is a
   platform designed to provide C-ITS data to external users.  By
   offering data through methods such as Open APIs, this system connects
   C-ITS infrastructure information with users, enabling seamless access
   to real-time traffic and transportation data.  It facilitates the
   integration of C-ITS data into various applications and services,
   supporting the development of innovative mobility solutions and
   enhancing the overall efficiency and safety of transportation
   systems.

   * Cyber Security System: The Cyber Security System is responsible for
   managing the security of communications between Software-Defined
   Vehicles (SDV), Vulnerable Road Users (VRU), RSU, Mobile Network
   Operators (MNO), and C-ITS infrastructure.  Security technologies are
   fundamentally integrated into all communications to ensure encrypted
   data transmission.  Outgoing data is encrypted using a public key,
   while receiving devices decrypt the data using a private key to
   securely access the information.  The Cyber Security System oversees
   the protection of both private and public keys across all modules,
   ensuring robust security against potential exposure and safeguarding
   the integrity and confidentiality of transmitted data.  In the A2A
   overlay, the Cyber Security System provides agent identity issuance,
   capability attestation, mutual authentication between agents, and
   revocation/rotation of agent credentials, so that all A2A
   interactions across I1 to I10 are cryptographically protected.

   * C-ITS Infra: The C-ITS Infrastructure is a system designed to
   collect and provide various types of information, including traffic
   signal data, roadside environment information, VRU data, and RSU
   data.  The specific C-ITS information available may vary depending on
   the devices and equipments installed on the road.  This
   infrastructure enables real-time data exchange between the
   transportation system and connected or autonomous vehicles,
   supporting safer and more efficient traffic management.  A C-ITS
   Infra Agent fronts this subsystem in the A2A overlay, brokering
   interactions between Center Cloud agents (downstream policy and ACF
   configuration) and RSU/SDV/MNO agents (upstream telemetry and
   capability advertisement).

   * RSU: The RSU is a device that connects the C-ITS Infrastructure
   with SDVs.  Through the RSU, SDVs can transmit and receive data
   between vehicles via V2V and between vehicles and infrastructure via
   V2I.  RSUs play a critical role in enabling real-time communication,
   providing essential information such as traffic signals, road



An, et al.                 Expires 6 May 2027                   [Page 7]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   conditions, and safety alerts, thereby enhancing the safety and
   efficiency of autonomous and connected vehicle operations.  Each RSU
   hosts a lightweight RSU Agent that participates in the A2A overlay to
   publish locally available ACFs (e.g., AI-based perception offload)
   and supporting tools (e.g., aggregation and filtering), discover SDV/
   VRU agents within its coverage, and accept intent-based tasks
   delegated by the C-ITS Infra Agent or by neighboring SDV agents.

   * SDV1 and SDV2: SDV1 and SDV2 are examples depicted in the diagram,
   but in real-world scenarios, there can be an arbitrary number of
   vehicles.  A Software-Defined Vehicle (SDV) is a vehicle whose
   functions (perception, planning, control, infotainment, diagnostics,
   and connectivity) are realized as software components running on a
   consolidated, zonal compute platform and are continuously updatable
   over the air.  An SDV exposes two main communication interfaces: an
   External Communication Interface that enables communication with
   external systems such as RSUs, other vehicles (V2V), and
   infrastructure (V2I/V2N) for seamless interaction within the C-ITS
   ecosystem; and an Internal Vehicle Network (IVN) Interface that
   manages internal communication within the vehicle, connecting various
   onboard systems and components to ensure smooth operation and
   integration of vehicle functionalities.  This dual-interface
   structure allows SDVs to efficiently exchange data both externally
   with the C-ITS infrastructure and internally for optimized vehicle
   control.  Because vehicle-side software, models, and skills evolve
   continuously, an SDV's externally visible capabilities are
   intentionally agentic: the SDV hosts an in-vehicle SDV Agent that
   acts as the vehicle's representative in the A2A overlay, publishes a
   signed Agent Card that reflects the currently installed software/
   skills, autonomy level, and safety class, negotiates ACF placement
   with RSU, MNO, and Center agents, and coordinates directly with other
   SDV agents for cooperative perception, platooning, and safety tasks.
   The SDV Agent also mediates between the IVN (zonal controllers,
   sensors, actuators) and the external A2A overlay, ensuring that only
   authorized, policy-bound intents reach in-vehicle subsystems.
















An, et al.                 Expires 6 May 2027                   [Page 8]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   * IVN-Network1 and IVN-Network2: IVN-Network1 and IVN-Network2 are
   examples, but in practice, the internal communication system of a
   vehicle can consist of N different networks.  These networks are part
   of the In-Vehicle Network (IVN), which facilitates communication
   within the vehicle.  In an SDV (Software-Defined Vehicles), the IVN
   is designed based on a Zonal Architecture, where communication
   interfaces connect various devices and components within specific
   zones of the vehicle.  This architecture improves data transmission
   efficiency, reduces wiring complexity, and enhances the integration
   of advanced systems for autonomous driving and vehicle control.
   Through this zonal design, SDVs can effectively manage high-speed
   data exchange between sensors, controllers, and actuators, supporting
   real-time processing and safer driving operations.

   * VRU: A VRU refers to users who can communicate either with an MNO
   or directly with SDVs.  VRUs typically include pedestrians, cyclists,
   and motorcyclists who are more susceptible to traffic accidents due
   to their limited protection.  By connecting with MNO networks, VRUs
   can receive real-time safety alerts and traffic information.
   Additionally, direct communication with SDVs enables VRUs to exchange
   critical safety data, such as location and movement intentions, which
   helps autonomous and connected vehicles detect and respond to nearby
   vulnerable users, ultimately enhancing road safety.  A VRU Agent,
   typically hosted on the VRU's personal device, participates in the
   A2A overlay with a constrained capability profile, announcing
   presence and intent (e.g., crossing, joining traffic) to nearby SDV
   and MNO agents under privacy-preserving identifiers.

   * MNO: An MNO is a service provider that owns and manages wireless
   communication infrastructure, including network towers, core
   networks, and data centers.  In the context of C-ITS, MNOs play a
   critical role in enabling real-time communication between vehicles,
   infrastructure, and VRUs by providing seamless connectivity through
   cellular networks (e.g., LTE, and 5G).  MNOs facilitate the
   transmission of safety messages, traffic updates, and vehicle data,
   ensuring low-latency, high-reliability communication essential for
   autonomous driving and connected vehicle ecosystems.  Additionally,
   MNOs collaborate with C-ITS infrastructure to enhance data security
   and manage network resources for efficient traffic management and
   mobility services.  The MNO hosts an MNO Agent that exposes network-
   side capabilities (e.g., slice selection, QoS modulation, edge
   compute placement) over A2A, so that SDV, VRU, and Center agents can
   request connectivity and ACF execution in an intent-based, cross-
   domain manner.







An, et al.                 Expires 6 May 2027                   [Page 9]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


                                   Center Cloud
        ********************************************************************
        *                                                                  *
        *   +------------------+-----+      +-----+---------------------+  *
        *   | C-ITS Center     |     |      |     | Government Public   |  *
        *   |------------------|     |  I1  |     | Center              |  *
        *   | Region Center    |Agent+<....>+Agent|---------------------|  *
        *   |------------------|     |      |     | C-ITS Data Linkage  |  *
        *   | Highway Center   |     |      |     |---------------------|  *
        *   |                  |     |      |     | Cyber Security Sys  |  *
        *   +------------------+--+--+      +--+--+---------------------+  *
        *                         ^            ^                           *
        **************************:************:****************************
                                  : I2         : I3
                                  v            v
                     +------------+----+     +-+------------------------+
                     |      Agent      |     |          Agent           |
                     |-----------------|I4   |--------------------------|
             I7 +...>|   C-ITS Infra   |<...>|           MNO            |
                :    +--------+--------+     +----------------+---------+
                :             ^              ^                ^
                :             : I5        I6 :             I6 :
                :             :              :                :
                v             v              v                v
          +-----+-----+ I8  +-+--------------++ I9  +-----+---+-+
          | RSU |Agent|<...>|      Agent      |<...>|Agent| VRU |
          +-----+-----+     |-----------------|     +-----+-----+
                            | SDV1            |
                            | IVN-Network1    |
                            +-------+---------+
                                    ^
                                    : I10
                                    :
                                    v
                              +-----+-------+
                              |    Agent    |
                              |-------------|
                              | SDV2        |
                              | IVN-Network2|
                              +-------------+

  <...> A2A Protocol

            Figure 1: Agentic AI Framework and Interfaces







An, et al.                 Expires 6 May 2027                  [Page 10]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


2.2.  Agentic AI Interfaces

   According to the framework described in the previous section, there
   are major interfaces for Agentic AI coordination in C-ITS and MNO
   networking.  Each interface listed below is realized as an A2A
   interaction between the two corresponding agents: the two agents
   authenticate each other using credentials issued by the Cyber
   Security System, exchange Agent Cards to learn each other's ACF-
   related capabilities, and then use intent-based request/response or
   publish/subscribe messages to register, configure, monitor, or invoke
   ACFs.  The interface descriptions below therefore cover both the
   underlying data exchange and the A2A semantics layered on top.

   Interface 1 (I1): This is the registration interface between the
   C-ITS Center and the Government Public Center.  It facilitates the
   exchange of C-ITS infrastructure data, such as traffic information
   and real-time road conditions, ensuring the Government Public Center
   can provide accurate and trustworthy data to external users.  This
   interface also supports secure data sharing through standardized
   protocols and encryption.

   Interface 2 (I2): This interface connects the C-ITS Center with the
   C-ITS Infra.  It is responsible for distributing infrastructure data,
   such as traffic signal information, road environment data, and RSU
   status, from the C-ITS Center to the C-ITS Infra for real-time
   processing and delivery to connected vehicles.  It ensures continuous
   data flow for effective traffic and infrastructure management.

   Interface 3 (I3): This is the data exchange interface between the
   Government Public Center and the MNO (Mobile Network Operator).  It
   enables the secure transmission of C-ITS data to MNOs, allowing
   mobile networks to deliver critical traffic and safety information to
   VRUs and vehicles.  This interface must ensure data integrity and
   security during transmission.

   Interface 4 (I4): This interface connects the C-ITS Infra with the
   MNO.  It supports the sharing of network resources and real-time
   communication between infrastructure components and mobile networks.
   This connection allows for efficient distribution of data, such as
   traffic alerts and safety notifications, to mobile users and
   vehicles.

   Interface 5 (I5): This is the communication interface between the
   C-ITS Infra and SDVs.  It enables bidirectional data exchange,
   allowing SDVs to receive real-time infrastructure information (e.g.,
   traffic signals, road hazards) and transmit vehicle status data back
   to the infrastructure.  This interface is critical for supporting V2I
   communications.



An, et al.                 Expires 6 May 2027                  [Page 11]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   Interface 6 (I6): This interface connects the MNO with both VRUs and
   SDVs.  It is used to deliver real-time safety messages, navigation
   updates, and other critical data.  It also allows VRUs and SDVs to
   send status or emergency signals back to the network.  This interface
   must ensure low-latency and secure data transmission to prevent
   accidents and improve traffic efficiency

   Interface 7 (I7): This is the management interface between the RSU
   and the C-ITS Infra.  It facilitates the configuration, monitoring,
   and management of RSUs to ensure stable communication between
   roadside infrastructure and vehicles.  It also handles firmware
   updates and diagnostics for RSUs.

   Interface 8 (I8): This interface supports V2I communication between
   SDVs through the RSU.  It allows SDVs to exchange critical
   information such as speed, direction, and emergency signals, enabling
   collision avoidance and cooperative driving.  This interface must
   provide real-time and reliable data exchange in dynamic traffic
   environments.

   Interface 9 (I9): This is the communication interface between SDVs
   and VRUs.  It ensures that vulnerable road users receive immediate
   safety notifications from nearby vehicles and infrastructure.  For
   example, SDVs can warn pedestrians of approaching vehicles or detect
   VRU movements in blind spots, enhancing road safety.

   Interface 10 (I10): This is the external and internal communication
   interface between multiple SDVs It enables secure and efficient
   communication within the vehicle's zonal architecture, facilitating
   seamless data exchange between various internal systems (e.g.,
   sensors, controllers) and supporting autonomous driving functions.

3.  Use Cases for Agentic AI in SDV and C-ITS

   This section introduces practical use cases of the Agentic AI
   framework in the context of Software-Defined Vehicles (SDVs), C-ITS,
   and MNO networking.  These use cases focus on how the software-
   defined and continuously updatable nature of SDVs, combined with the
   A2A overlay, enables End-to-End (E2E) communication, cybersecurity,
   and cooperative driving services that adapt to vehicle, road, and
   network context in real time.

   * Real-Time Data Processing for SDV: The Agentic AI framework enables
   seamless communication between SDVs and C-ITS infrastructure through
   interfaces such as I5 (C-ITS Infra <-> SDV) and I8 (V2V Communication
   via RSU).  Real-time data such as traffic signals, road conditions,
   and obstacle detection are transmitted to SDVs for immediate
   processing.  By offloading certain data processing tasks to network



An, et al.                 Expires 6 May 2027                  [Page 12]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   devices (e.g., RSUs), SDVs can reduce internal computational load,
   allowing faster decision-making for functions like emergency braking
   or lane changes.  This distributed data processing model improves the
   overall safety and efficiency of autonomous driving.

   * E2E Communication for Cooperative Driving: The integration of MNO
   networks with C-ITS through interfaces like I4 (C-ITS Infra <-> MNO)
   and I6 (MNO <-> VRU/SDV) allows for reliable and low-latency E2E
   communication.  This connectivity is essential for cooperative
   driving scenarios, where multiple SDVs coordinate lane changes,
   merging, or platooning in real time.  The Agentic AI framework
   ensures that the network can dynamically manage traffic loads and
   prioritize safety-critical data transmission, enabling vehicles to
   share and act on real-time information seamlessly.

   * Enhanced Cybersecurity for C-ITS and MNO Integration: Given the
   extensive data exchange between vehicles, infrastructure, and network
   operators, cybersecurity is a critical component.  The Cyber Security
   System within the Agentic AI framework, managed through interfaces
   like I3 (Government Public Center <-> MNO) and I10 (Internal SDV
   Communication), provides E2E encryption and secure key management.
   Private keys are stored securely in the cloud and can be updated via
   Over-The-Air (OTA) mechanisms if compromised.  If a critical security
   breach occurs, the system can initiate a global reset to reissue
   encryption keys, ensuring system-wide security integrity.  This
   proactive approach minimizes the risk of cyberattacks on connected
   vehicles and infrastructure.

   * Dynamic Resource Allocation for High-Density Traffic Environments:
   In high-traffic conditions such as highways or urban intersections,
   efficient data management is crucial.  The Agentic AI framework,
   through I7 (RSU <-> C-ITS Infra) and I9 (SDV <-> VRU), enables
   dynamic resource allocation.  For example, RSUs can prioritize data
   transmission for emergency vehicles or redirect network resources to
   manage traffic congestion.  This adaptive data flow management
   reduces latency and prevents network bottlenecks, ensuring that all
   vehicles and infrastructure components receive critical information
   in real time.













An, et al.                 Expires 6 May 2027                  [Page 13]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   * Edge Computing for Latency-Sensitive Applications: Edge computing
   capabilities are integrated into the Agentic AI framework using RSUs
   and Programmable Network Devices (PNDs) to handle latency-sensitive
   tasks.  Interfaces like I1 (C-ITS Center <-> Government Public
   Center) and I8 (SDV <-> SDV via RSU) allow certain computational
   tasks such as object detection or predictive path planning to be
   processed at the network edge rather than relying on centralized
   cloud servers.  This significantly reduces response time for
   autonomous driving actions and enhances road safety by enabling
   faster vehicle reactions.

   * These use cases demonstrate how the Agentic AI framework can
   enhance the performance, security, and reliability of intelligent
   transportation systems by integrating C-ITS infrastructure with MNO
   networks.  By supporting real-time data processing, secure
   communication, and dynamic resource management, the framework
   addresses the complex demands of modern SDVs and connected mobility
   solutions.

4.  A2A Use Cases for SDV-Centric C-ITS Integration

   In emerging SDV ecosystems, seamless coordination among vehicle
   agents, C-ITS infrastructure agents, and MNO agents becomes essential
   for achieving real-time awareness and adaptive network optimization.
   The _A2A_ communication paradigm enables direct interaction among
   these heterogeneous entities without relying solely on centralized
   control mechanisms.  For instance, SDV agents can dynamically
   negotiate data offloading, perform network slicing, or compute
   resource scheduling with MNO agents in response to vehicular mobility
   and network conditions, while roadside infrastructure agents exchange
   situational context such as congestion level, edge compute
   availability, or safety alerts with SDV agents to support cooperative
   perception and hazard prediction.  These A2A interactions facilitate
   distributed decision-making across application and network layers,
   forming the basis for intelligent coordination and resilient service
   continuity in connected mobility.

   When integrated with the proposed Agentic AI framework, such A2A-
   based orchestration allows SDVs, RSUs, and MNO backends to jointly
   manage ACFs, thereby ensuring low-latency, secure, and reliable
   service delivery across multi-domain C-ITS environments.  This
   approach follows recognized architectures and standards used in C-ITS
   and V2X systems, including frameworks for security management and
   network data analytics.  Leveraging these industry standards, A2A
   coordination can extend beyond individual domains to enable intent-
   driven, cross-layer management of communication, computation, and
   data analytics among SDV, C-ITS, and MNO ecosystems.




An, et al.                 Expires 6 May 2027                  [Page 14]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   The following use cases illustrate concrete A2A interaction patterns
   on top of the Agentic AI framework.  They are intentionally aligned
   with the agentic-AI direction discussed in the IRTF NMRG at IETF 124
   and IETF 125, and with the broader industry A2A reference patterns
   (e.g., Agent Cards, task lifecycles, and secure peer-to-peer
   messaging) so that C-ITS deployments can interoperate with general-
   purpose AI agent ecosystems.

   * Capability Advertisement and Discovery (I2, I4, I5, I7): C-ITS
   Infra Agent, MNO Agent, RSU Agents, and SDV Agents publish Agent
   Cards describing the ACFs they host (e.g., AI-based perception
   offload) and their supporting skills (e.g., message aggregation,
   slice selection, and OTA key rotation).  Center Cloud agents index
   these cards via I2/I3 and use them to plan ACF placement.  Discovery
   is intent-driven: an SDV Agent requesting "low-latency cooperative
   perception within 100 ms" is matched with nearby RSU/MNO agents whose
   Agent Cards advertise compatible ACFs.

   * Cooperative Perception via SDV-to-SDV and SDV-to-RSU A2A (I8, I10):
   SDV agents directly negotiate sensor data exchange, fusion
   responsibility, and confidence reporting with neighboring SDV agents
   (I10) and RSU agents (I8).  Heavy fusion ACFs can be delegated to the
   RSU Agent when the SDV is compute-constrained, while the SDV Agent
   retains decision authority.  This pattern mirrors the agent-
   interconnection and multi-level inference approach discussed in
   [I-D.chuyi-nmrg-agentic-network-inference].

   For cooperative perception, an SDV Agent selects an advertised RSU or
   MEC processing function and sends the region of interest, processing
   deadline, and required output format.  The provider checks the
   request conditions and available resources, and returns acceptance
   with a task identifier or rejection with a reason.  After acceptance,
   the provider invokes the ACF and reports its processing status.  A
   completed result includes the observation time, source, and quality
   information so that the SDV can combine it with its own observations.

   Agent coordination uses an IP-capable path, including connectivity
   through LTE or 5G mobile networks.  LTE-V2X and NR-V2X PC5 can carry
   vehicle and roadside data in the same deployment.  The coordination
   path and the observation or result data path can differ, depending on
   the available connectivity and application mapping.










An, et al.                 Expires 6 May 2027                  [Page 15]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   * Intent-Based Slice and Edge Negotiation with MNO (I4, I6): An SDV
   Agent or a VRU Agent expresses a high-level intent (e.g., "platooning
   session, 50 vehicles, end-to-end latency budget 20 ms, reliability
   99.999 percent") to the MNO Agent via A2A.  The MNO Agent translates
   the intent into slice configuration and edge ACF placement,
   consistent with the agentic network optimization patterns described
   in [I-D.bernardos-nmrg-agentic-network-optimization].

   * Cross-Domain Safety Coordination (I3, I6, I9): When a Government
   Public Center Agent detects a hazardous event, it issues an A2A task
   to the MNO Agent and C-ITS Infra Agent, which fan out the task to
   local RSU and SDV agents.  VRU agents in the affected area receive
   targeted alerts via I6/I9 under privacy-preserving identifiers
   provided by the Cyber Security System.

   * Agentic Network Digital Twin for C-ITS (I1, I2, I4): Center Cloud
   agents maintain an agentic NDT of the C-ITS network and use it to
   simulate ACF placement before pushing configurations to C-ITS Infra
   and MNO agents.  This use case aligns with the agent-driven NDT
   architecture in [I-D.wmz-nmrg-agent-ndt-arch] and with the problem
   statement for agentic AI in network management in
   [I-D.hong-nmrg-agenticai-ps].

   * Tool-Augmented Agent Operations via MCP (all interfaces): Agents
   may expose their local management tools (e.g., RSU firmware update,
   SDV diagnostic, slice reconfiguration) as MCP-style tools, and other
   agents may invoke them under A2A authorization.  The applicability of
   MCP to such network management tasks is discussed in
   [I-D.yang-nmrg-mcp-nm]; the Agentic AI framework reuses this pattern
   to keep tool invocation auditable and policy-bound.

   * Autonomous Reaction and Human-in-the-Loop Oversight: Following the
   agentic-AI architectural principles in
   [I-D.jadoon-nmrg-agentic-ai-autonomous-networks], Agentic AI agents
   operate within explicit autonomy levels.  Routine reconfiguration of
   ACFs or supporting services (e.g., adjusting aggregation thresholds
   at an RSU) runs autonomously, while safety-critical reconfiguration
   (e.g., revoking a compromised agent or resetting cryptographic
   material) requires escalation to a Center Cloud agent and, where
   appropriate, a human operator.











An, et al.                 Expires 6 May 2027                  [Page 16]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


5.  Agent Cards for SDVs and C-ITS Entities

   Following the A2A protocol model adopted by [I-D.yang-nmrg-a2a-nm],
   every agent in the Agentic AI framework MUST publish an Agent Card: a
   JSON metadata document that describes the agent's identity, endpoint,
   advertised ACF-related capabilities, skills, and authentication
   requirements.  Agent Cards are the basis for capability discovery and
   intent matching across I1 to I10.  Agent Cards are signed and
   attested by the Cyber Security System and SHOULD be discoverable by
   peer agents either via a well-known URI on the hosting entity or via
   a Center Cloud registry.

   The following non-normative examples illustrate Agent Cards for
   representative C-ITS entities.  The illustrative "capabilities" field
   lists ACF and supporting-service classes exposed by the agent, and
   "skills" enumerate concrete actions invocable through A2A messages.

{
  "name": "RSUAgent-01",
  "description": "RSU agent for AI perception offload and message aggregation.",
  "url": "https://rsu-01.cits.example.net/a2a/tasks",
  "capabilities": ["PerceptionOffload", "MessageAggregation",
                   "V2X-Relay"],
  "skills": [
    {
      "id": "offload_perception",
      "name": "Cooperative perception offload",
      "description":
        "Run sensor fusion on behalf of an SDV within RSU coverage.",
      "inputModes":  ["application/json", "text/structured"],
      "outputModes": ["application/json", "text/status"]
    },
    {
      "id": "aggregate_cam",
      "name": "CAM/BSM aggregation",
      "description":
        "Aggregate and forward V2X awareness messages to the C-ITS Infra.",
      "inputModes":  ["application/json"],
      "outputModes": ["text/status"]
    }
  ]
}

               Figure 2: Example Agent Card: RSU Agent







An, et al.                 Expires 6 May 2027                  [Page 17]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


{
  "name": "SDVAgent-VIN-XYZ",
  "description": "In-vehicle SDV agent for cooperative driving over A2A.",
  "url": "https://sdv-xyz.veh.example.net/a2a/tasks",
  "capabilities": ["CooperativePerception", "Platooning",
                   "Telemetry", "OTA-KeyRotation"],
  "skills": [
    {
      "id": "request_perception",
      "name": "Request cooperative perception",
      "description":
        "Request fused object lists from nearby RSU or SDV agents.",
      "inputModes":  ["application/json"],
      "outputModes": ["application/json"]
    },
    {
      "id": "join_platoon",
      "name": "Join platoon",
      "description": "Negotiate joining a platoon with neighboring SDV agents.",
      "inputModes":  ["application/json"],
      "outputModes": ["text/status"]
    }
  ]
}

               Figure 3: Example Agent Card: SDV Agent

























An, et al.                 Expires 6 May 2027                  [Page 18]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


{
  "name": "MNOAgent-OpA",
  "description": "MNO agent for C-ITS slice selection and edge ACF placement.",
  "url": "https://mno-a.example.net/a2a/tasks",
  "capabilities": ["NetworkSlicing", "EdgeACFPlacement",
                   "QoSModulation"],
  "skills": [
    {
      "id": "select_slice",
      "name": "Slice selection",
      "description":
        "Select or instantiate a network slice matching the requested intent.",
      "inputModes":  ["application/json"],
      "outputModes": ["application/json"]
    },
    {
      "id": "place_edge_acf",
      "name": "Edge ACF placement",
      "description":
        "Place an ACF on an MEC node close to the requesting SDV/VRU agent.",
      "inputModes":  ["application/json"],
      "outputModes": ["text/status"]
    }
  ]
}

               Figure 4: Example Agent Card: MNO Agent

   For a cooperative-perception capability, the Agent Card's skill
   identifier and name identify the offered function, while its input
   and output modes describe the supported formats.  The associated ACF
   descriptor additionally identifies the processing region and
   endpoint.  The "acf-descriptor" container below represents this
   information.  A request refers to the advertised capability using
   "capability-id".  The descriptor can be carried in or referenced from
   the Agent Card according to the selected A2A version.

6.  YANG-based Structured Data for A2A Communication in SDV and C-ITS

   While the A2A protocol natively supports unstructured textual content
   for inter-agent exchange, the orchestration of ACFs and supporting
   services in C-ITS demands clarity, unambiguous intent transmission,
   and machine-interpretable data, especially when safety-critical
   actuation (e.g., emergency rerouting, key revocation, slice
   reconfiguration) is involved.  Natural-language-only payloads carry
   the inherent ambiguity of human language and can lead to inconsistent
   ACF configuration or large-scale service degradation, which is
   unacceptable for C-ITS.



An, et al.                 Expires 6 May 2027                  [Page 19]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   This document adopts the approach of [I-D.yang-nmrg-a2a-nm], which
   uses YANG [RFC7950] as the structured data layer for A2A payloads,
   with NETCONF [RFC6241] and RESTCONF [RFC8040] as candidate underlying
   management protocols when an agent needs to actuate an ACF on a
   programmable network device.  Building on that approach, this
   document defines a non-normative YANG module, "ietf-cits-acf-intent",
   whose data nodes are carried inside the "data" part of A2A messages
   exchanged across I1 to I10.  The module provides a machine-
   interpretable representation of C-ITS ACF intents and supplements the
   natural-language "text" parts that the same A2A message may carry.
   The intent is that an SDV, RSU, MNO, or Center agent that receives an
   A2A message can validate the "data" part against this YANG module
   before acting upon it, and can use existing NETCONF or RESTCONF
   tooling to persist or actuate the intent on a programmable network
   device.

   In the C-ITS context, YANG-modeled payloads carried as the "data"
   part of A2A messages are RECOMMENDED for the following content
   categories:

   * ACF descriptors (capability schemas of RSU/MNO/SDV ACFs), and their
   supporting services, modeled by the "acf-descriptor" container of
   "ietf-cits-acf-intent".

   * C-ITS intent objects (e.g., platooning, cooperative perception,
   hazard alert) expressed as structured intents under the "cits-acf-
   intents" container of "ietf-cits-acf-intent".

   * Telemetry and incident reports between RSU/SDV agents and Center
   Cloud agents, reusing the YANG model in
   [I-D.ietf-nmop-network-incident-yang] where applicable.

   * Cryptographic material lifecycle events (rotation, revocation)
   managed by the Cyber Security System.

   The cooperative-perception exchange uses the existing intent object
   for Request, Response, and Result information.  The Request uses
   "intent-id" as its request identifier and adds "capability-id" and
   "output-format" to the existing region and deadline fields.  The
   Response carries the same "intent-id", an accepted or rejected
   status, and a "task-id" when accepted.  A rejected request includes a
   "reason" and does not allocate a task identifier.

   After execution, the Result retains the request and task identifiers
   and reports completed or failed status.  The added "result" container
   carries a result reference, output format, observation time, source
   identifiers, and a quality metric and value.  The metric identifies
   how to interpret the quality value.  The SDV checks these fields and



An, et al.                 Expires 6 May 2027                  [Page 20]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   the result's age before using the information.  For cooperative
   perception, "response-deadline" denotes the latest acceptable receipt
   time of the processing result; admission is reported before that
   time.

   The added fields are optional in the module skeleton so that existing
   intent objects remain representable.  A cooperative- perception
   request supplies the selected capability and output format.  Its
   provider assigns a task identifier after acceptance and supplies the
   result information on completion.  These data fields supplement the
   existing A2A message and task mechanisms.

   The module name, namespace, prefix, and local identifiers below use
   ACF terminology.  These illustrative names are a proposal, not
   registered or finalized identifiers.  The model includes supporting
   service classes such as NetworkSlicing and OTAKeyRotation; those
   classes do not imply that the underlying operation uses AI.  Renaming
   the earlier module is a schema and namespace change, not a backward-
   compatible alias.  Implementations using earlier names would require
   an explicit migration mapping.

   The high-level tree structure of the "ietf-cits-acf-intent" module,
   expressed using the YANG tree diagram notation, is shown in Figure 5.
   The corresponding YANG module skeleton is shown in Figure 6.  Both
   artifacts are non-normative and are intended as a starting point for
   a companion YANG document.

   module: ietf-cits-acf-intent

     +--ro acf-descriptor
     |  +--ro capability* [capability-id]
     |     +--ro capability-id      string
     |     +--ro function-name?     string
     |     +--ro endpoint?          string
     |     +--ro input-format*      string
     |     +--ro output-format*     string
     |     +--ro service-region*    string
     |     +--ro region-reference?  string
     +--rw cits-acf-intents
        +--rw intent* [intent-id]
           +--rw intent-id            string
           +--rw type                 identityref
           +--rw requester            string
           +--rw target-acf           identityref
           +--rw capability-id?      string
           +--rw output-format?      string
           +--rw safety-class?        enumeration
           +--rw priority?            uint8



An, et al.                 Expires 6 May 2027                  [Page 21]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


           +--rw response-deadline?   yang:date-and-time
           +--rw interface?           enumeration
           +--rw constraints
           |  +--rw max-latency-ms?      uint32
           |  +--rw min-reliability?     decimal64
           |  +--rw min-confidence?      decimal64
           |  +--rw region-of-interest?  string
           |  +--rw bandwidth-mbps?      uint32
           |  +--rw max-data-age-ms?    uint32
           |  +--rw region-reference?   string
           |  +--rw quality-metric?     string
           +--rw participants*        string
           +--ro task-id?            string
           +--ro reason?             string
           +--ro result
           |  +--ro result-reference?  string
           |  +--ro output-format?     string
           |  +--ro observation-time?  yang:date-and-time
           |  +--ro source-id*         string
           |  +--ro quality-metric?    string
           |  +--ro quality-value?     decimal64
           +--ro status?              enumeration
           +--ro last-updated?        yang:date-and-time

     notifications:
       +---n acf-intent-event
          +--ro intent-id    -> /cits-acf-intents/intent/intent-id
          +--ro event-type   enumeration
          +--ro task-id?     string
          +--ro reason?      string
          +--ro details?     string

            Figure 5: YANG Tree Diagram of ietf-cits-acf-intent

module ietf-cits-acf-intent {
  yang-version 1.1;
  namespace "urn:ietf:params:xml:ns:yang:ietf-cits-acf-intent";
  prefix "cits-acf";

  import ietf-yang-types {
    prefix yang;
  }

  organization
    "IETF NMRG (Network Management Research Group)";
  contact
    "WG Web:   <https://datatracker.ietf.org/rg/nmrg/>
     Editor:  Byoungman An <bman@keti.re.kr>



An, et al.                 Expires 6 May 2027                  [Page 22]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


     Editor:  Jaehoon Paul Jeong <pauljeong@skku.edu>";
  description
    "This YANG module defines a data model for representing
     Cooperative Intelligent Transportation System (C-ITS)
     AI Computing Function (ACF) intents that are carried
     as the 'data' part of Agent-to-Agent (A2A) messages between
     SDV, RSU, MNO, VRU, and Center agents of the Agentic AI framework.";

  revision 2026-11-02 {
    description
      "Align the illustrative module, namespace, and identifiers
       with ACF terminology in the Agentic AI framework;
       distinguish ACFs from supporting services.";
    reference "draft-an-nmrg-i2icf-cits-03";
  }

  revision 2026-10-27 {
    description
      "Add capability, request, response, and result fields for
       SDV-to-RSU/MEC cooperative perception.";
    reference "draft-an-nmrg-i2icf-cits-03";
  }

  revision 2026-02-05 {
    description "Initial revision.";
    reference "draft-an-nmrg-i2icf-cits-02";
  }

  /* Identities for ACF intent types and target ACF classes */

  identity intent-type {
    description "Base identity for C-ITS ACF intent types.";
  }
  identity cooperative-perception {
    base intent-type;
    description "Cooperative perception intent (e.g., SDV-RSU).";
  }
  identity platooning {
    base intent-type;
    description "Platooning session intent among SDVs.";
  }
  identity hazard-alert {
    base intent-type;
    description "Hazard alert dissemination intent.";
  }
  identity slice-request {
    base intent-type;
    description "Network slice/edge ACF placement intent to MNO.";



An, et al.                 Expires 6 May 2027                  [Page 23]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


  }
  identity ota-key-rotation {
    base intent-type;
    description "OTA key rotation intent managed by the Cyber
                 Security System.";
  }

  identity acf-class {
    description "Base identity for ACF or supporting-service classes.";
  }
  identity PerceptionOffload    { base acf-class; }
  identity MessageAggregation   { base acf-class; }
  identity NetworkSlicing       { base acf-class; }
  identity EdgeACFPlacement     { base acf-class; }
  identity OTAKeyRotation       { base acf-class; }

  /* Top-level data tree */

  container acf-descriptor {
    config false;
    description
      "Advertised ACFs and supporting services of the hosting agent.";
    list capability {
      key "capability-id";
      description "One computing function offered by the agent.";
      leaf capability-id {
        type string;
        description "Identifier of the advertised function or service.";
      }
      leaf function-name {
        type string;
        description "Name of the offered processing function.";
      }
      leaf endpoint {
        type string;
        description "Endpoint for requesting the function.";
      }
      leaf-list input-format {
        type string;
        description "Supported input representations.";
      }
      leaf-list output-format {
        type string;
        description "Supported output representations.";
      }
      leaf-list service-region {
        type string;
        description "Regions for which processing is offered.";



An, et al.                 Expires 6 May 2027                  [Page 24]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


      }
      leaf region-reference {
        type string;
        description "Coordinate or map reference for the regions.";
      }
    }
  }

  container cits-acf-intents {
    description
      "Collection of C-ITS ACF intents currently known to the
       hosting agent.";
    list intent {
      key "intent-id";
      description "A single C-ITS ACF intent.";

      leaf intent-id {
        type string;
        description "Globally unique intent identifier.";
      }
      leaf type {
        type identityref { base intent-type; }
        mandatory true;
        description "Intent type (e.g., cooperative-perception).";
      }
      leaf requester {
        type string;
        mandatory true;
        description "Agent name of the requesting agent.";
      }
      leaf target-acf {
        type identityref { base acf-class; }
        mandatory true;
        description "Targeted ACF or supporting-service class.";
      }
      leaf capability-id {
        type string;
        description "Selected capability advertised by the provider.";
      }
      leaf output-format {
        type string;
        description "Requested result representation.";
      }
      leaf safety-class {
        type enumeration {
          enum QM;     enum ASIL-A; enum ASIL-B;
          enum ASIL-C; enum ASIL-D;
        }



An, et al.                 Expires 6 May 2027                  [Page 25]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


        description "ISO 26262 ASIL classification of the intent.";
      }
      leaf priority {
        type uint8 { range "0..7"; }
        description "Relative priority (0 = lowest).";
      }
      leaf response-deadline {
        type yang:date-and-time;
        description "Latest acceptable response time.";
      }
      leaf interface {
        type enumeration {
          enum I1; enum I2; enum I3; enum I4; enum I5;
          enum I6; enum I7; enum I8; enum I9; enum I10;
        }
        description "Agentic AI interface over which the intent flows.";
      }
      container constraints {
        description "Operational constraints attached to the intent.";
        leaf max-latency-ms     { type uint32; units "milliseconds"; }
        leaf min-reliability    { type decimal64 { fraction-digits 5; } }
        leaf min-confidence     { type decimal64 { fraction-digits 2; } }
        leaf region-of-interest { type string; }
        leaf bandwidth-mbps     { type uint32; units "Mbit/s"; }
        leaf max-data-age-ms {
          type uint32; units "milliseconds";
          description "Maximum observation age at result receipt.";
        }
        leaf region-reference {
          type string;
          description "Coordinate or map reference for the region.";
        }
        leaf quality-metric {
          type string;
          description "Definition used to interpret min-confidence.";
        }
      }
      leaf-list participants {
        type string;
        description "Other agents participating in this intent.";
      }
      leaf task-id {
        type string;
        config false;
        description "Provider-assigned identifier after acceptance.";
      }
      leaf reason {
        type string;



An, et al.                 Expires 6 May 2027                  [Page 26]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


        config false;
        description "Reason for rejection, failure, or expiry.";
      }
      container result {
        config false;
        description "Information returned for a completed task.";
        leaf result-reference {
          type string;
          description "Reference to the processing result.";
        }
        leaf output-format {
          type string;
          description "Representation of the returned result.";
        }
        leaf observation-time {
          type yang:date-and-time;
          description "Earliest observation contributing to result.";
        }
        leaf-list source-id {
          type string;
          description "Sources contributing to the result.";
        }
        leaf quality-metric {
          type string;
          description "Definition used to interpret quality-value.";
        }
        leaf quality-value {
          type decimal64 { fraction-digits 5; }
          description "Quality under the identified metric.";
        }
      }
      leaf status {
        type enumeration {
          enum requested; enum accepted; enum running;
          enum completed; enum failed;   enum revoked;
          enum rejected;  enum expired;
        }
        config false;
        description "Lifecycle status of the intent.";
      }
      leaf last-updated {
        type yang:date-and-time;
        config false;
        description "Time of the last status update.";
      }
    }
  }




An, et al.                 Expires 6 May 2027                  [Page 27]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


  /* Event-driven A2A notification */

  notification acf-intent-event {
    description
      "Notification emitted when an intent changes state, used by
       event-driven A2A pub/sub channels (see Section
       'Event-driven A2A for Safety Signalling').";
    leaf intent-id {
      type leafref {
        path "/cits-acf:cits-acf-intents/cits-acf:intent"
           + "/cits-acf:intent-id";
      }
    }
    leaf event-type {
      type enumeration {
        enum accepted; enum rejected; enum progress;
        enum completed; enum failed;   enum revoked;
        enum expired;
      }
    }
    leaf task-id {
      type string;
      description "Task identifier assigned on acceptance.";
    }
    leaf reason {
      type string;
      description "Reason for an unsuccessful outcome.";
    }
    leaf details { type string; }
  }
}

         Figure 6: YANG Module Skeleton: ietf-cits-acf-intent

   The following non-normative example illustrates an A2A message sent
   from an SDV Agent to an RSU Agent over I8 to request a cooperative
   perception ACF.  The "data" part carries an instance of the "ietf-
   cits-acf-intent" YANG module shown above, encoded as JSON per
   [RFC7951], using the media type "application/yang-data+json" defined
   in [RFC8040], while the "text" part preserves a human-readable
   summary.

POST /a2a/tasks HTTP/1.1
Host: rsu-01.cits.example.net
Content-Type: application/json
{
  "jsonrpc": "2.0",
  "method": "message/send",



An, et al.                 Expires 6 May 2027                  [Page 28]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


  "params": {
    "message": {
      "message_id": "c1ts-7f3e-4a21-9b12-0e2d6a8c4f10",
      "context_id": "intersection-42:approach-east",
      "role": "ROLE_USER",
      "parts": [
        {
          "text":
            "Cooperative perception within 100 ms: intersection-42 eastbound.",
          "media_type": "text/plain"
        },
        {
          "data": {
            "ietf-cits-acf-intent:cits-acf-intents": {
              "intent": [
                {
                  "intent-id": "intent-9921",
                  "type":
                    "ietf-cits-acf-intent:cooperative-perception",
                  "requester": "SDVAgent-VIN-XYZ",
                  "target-acf":
                    "ietf-cits-acf-intent:PerceptionOffload",
                  "capability-id": "offload_perception",
                  "output-format": "application/json",
                  "safety-class": "ASIL-B",
                  "priority": 6,
                  "response-deadline": "2026-02-10T06:00:00Z",
                  "interface": "I8",
                  "constraints": {
                    "max-latency-ms": 100,
                    "min-confidence": "0.85",
                    "quality-metric": "fusion-confidence-v1",
                    "max-data-age-ms": 150,
                    "region-reference": "intersection-map-v1",
                    "region-of-interest":
                      "intersection-42:east"
                  },
                  "participants": ["RSUAgent-01"]
                }
              ]
            }
          },
          "media_type": "application/yang-data+json",
          "metadata": {
            "yang_module": "ietf-cits-acf-intent",
            "revision": "2026-11-02"
          }
        }



An, et al.                 Expires 6 May 2027                  [Page 29]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


      ]
    },
    "configuration": {
      "accepted_output_modes": ["application/yang-data+json",
                                "text/status"],
      "blocking": false,
      "history_length": 3
    },
    "metadata": {
      "request_type": "acf_invocation",
      "priority": "high",
      "response_deadline": "2026-02-10T06:00:00Z"
    }
  }
}

   Figure 7: A2A Message Example with YANG-structured Data (I8: SDV
                               to RSU)

   The YANG module shown above is non-normative and is expected to be
   refined in a companion YANG document, reusing IETF building blocks
   for incident management [I-D.ietf-nmop-network-incident-yang] and
   existing C-ITS/V2X data dictionaries.

7.  Operational Considerations

   The introduction of an A2A overlay within the Agentic AI framework
   has several operational implications that need to be considered when
   deploying the architecture in large-scale, safety-critical C-ITS
   environments.  This section is aligned with the operational
   considerations discussed in [I-D.yang-nmrg-a2a-nm] and adapts them to
   C-ITS.

7.1.  Agent Skills as Expertise Expansion

   General-purpose AI agents do not natively possess C-ITS domain
   expertise (e.g., V2X message semantics, RSU operational procedures,
   MNO slice templates).  C-ITS operators SHOULD package such expertise
   as Agent Skills (e.g., the "diagnose_rsu_link_loss" or
   "build_platoon_session" skill) so that a generic Agentic AI agent can
   be specialized for a given deployment without re-training the
   underlying model.  Skills SHOULD be versioned and signed so that the
   Cyber Security System can attest the skill set advertised in an Agent
   Card.







An, et al.                 Expires 6 May 2027                  [Page 30]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


7.2.  Knowledge Base as Ground Truth Data

   To mitigate hallucinations of LLM-based agents, deployments SHOULD
   provide a shared, machine-interpretable knowledge base covering C-ITS
   topology, RSU/SDV configuration baselines, MNO slice templates,
   regulatory policies, and historical incident records.  Agents
   retrieve from this knowledge base when reasoning over A2A tasks.  The
   Model Context Protocol (MCP) [I-D.yang-nmrg-mcp-nm] is a candidate
   substrate for standardized knowledge-base access in this framework.

7.3.  Event-driven A2A for Safety Signalling

   The task-based A2A request/response model is sufficient for slow-loop
   ACF configuration but insufficient for safety-critical C-ITS events
   (e.g., emergency electronic brake, hazard detection).  For such
   events, agents SHOULD also support an event-driven publish/subscribe
   pattern, where a Hazard Agent publishes to a topic on a message
   broker and interested SDV, RSU, MNO, and Center agents subscribe to
   receive near-real-time notifications.  This pattern reuses the event-
   driven A2A architecture described in [I-D.yang-nmrg-a2a-nm].

7.4.  Scalability

   A national-scale C-ITS deployment may include thousands of RSUs,
   millions of SDVs and VRUs, and multiple cooperating MNOs.  A2A
   messaging volume is therefore significantly higher than static RPC-
   based interfaces.  Operators SHOULD:

   * Adopt hierarchical/federated agent structures so that local
   decisions (e.g., intersection-level cooperative perception) remain
   local to RSU and SDV agents.

   * Apply rate-limiting, backoff, and prioritization on A2A messages,
   with the highest priority reserved for safety signalling.

   * Localize Agent Card registries per region (e.g., per Region Center
   or Highway Center) to avoid global discovery hotspots.

7.5.  Latency and Performance Constraints

   Many C-ITS use cases (cooperative perception, platooning,
   intersection management) impose tight end-to-end timing budgets (tens
   of milliseconds).  Implementations SHOULD define performance
   envelopes for:

   * Maximum A2A message processing latency for each interface (I1 to
   I10).




An, et al.                 Expires 6 May 2027                  [Page 31]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   * Timeout and retry behavior, with deterministic fallback to direct
   controller APIs or pre-provisioned static ACF configurations when A2A
   interactions cannot meet the deadline.

   * Bounded LLM inference time when an agent relies on generative
   reasoning; safety-critical paths SHOULD NOT depend on unbounded LLM
   inference.

7.6.  Reliability and Failure Handling

   A2A workflows in C-ITS may span vehicles entering and leaving
   coverage, intermittent wireless links, and partial cloud failures.
   Implementations SHOULD provide:

   * Workflow checkpointing so that long-lived tasks (e.g., a platoon
   session) can survive agent restarts.

   * Idempotent retry semantics for ACF invocations.

   * Transactional ACF reconfiguration with confirmed-commit-like
   semantics (analogous to NETCONF confirmed-commit) when multiple
   devices must be updated atomically.

   * Circuit breakers and automatic escalation to a human operator when
   sustained failures threaten safety.

   For cooperative-perception requests, the provider uses the
   authenticated requester and "intent-id" to recognize a retry of the
   same request.  While the request record is retained, it returns the
   recorded decision or task status without starting a second execution.
   Reuse of that identifier with different request conditions is
   rejected with a reason.  Retention covers at least the request's
   response deadline.

   If the deadline is reached before completion, the provider reports
   expired status and a reason.  Failed or expired tasks retain the
   assigned task identifier.  The SDV does not use a result received
   after its deadline or maximum data-age limit.  For a completed task,
   the returned output format matches the accepted request, and its
   source and quality information remain associated with the same
   request and task identifiers.










An, et al.                 Expires 6 May 2027                  [Page 32]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


7.7.  Agent Lifecycle and Resource Management

   Agentic AI agents are long-running production components and require
   lifecycle management: onboarding and registration through the Cyber
   Security System, controlled model/skill updates (with rollback),
   decommissioning, and revocation of compromised agents through the
   same OTA mechanism used for cryptographic key rotation.  Operators
   SHOULD monitor agent resource consumption (CPU, memory, inference
   workload), especially on constrained RSU and SDV platforms.

7.8.  Inter-Domain Operational Challenges

   C-ITS is inherently multi-domain: Government Public Center, Region/
   Highway Centers, multiple MNOs, vehicle OEMs, and third-party service
   providers may each operate independent agent populations.  Cross-
   domain A2A interactions therefore require:

   * Federated trust establishment among Cyber Security Systems of
   different domains.

   * Interoperable Agent Card schemas and YANG models for ACF intents.

   * Policy reconciliation (e.g., differing privacy regulations for VRU
   data) before sharing tasks across administrative boundaries.

   * Auditing and accountability records for every cross-domain A2A
   invocation, retained by the receiving domain.

8.  IANA Considerations

   TBD.

9.  Security Considerations

   The Agentic AI framework for C-ITS and MNO Networking offers numerous
   advantages for various applications.  However, due to the framework's
   extensive connectivity between diverse vehicles, devices, centers,
   clouds, and VRUs, a vast amount of information and functionalities
   are exposed during network configuration, leading to potential
   security risks.  To ensure the overall security of the entire system,
   the following measures are recommended: First, the application
   development system should be controlled by the same service providers
   (e.g., cloud service providers or network operators) that own the
   network and computing infrastructure.  Second, devices within the
   cloud center should be pre-configured with security zones to isolate
   traffic, preventing it from affecting other network traffic.  Third,
   encryption keys for each device should be centrally managed by the
   cloud center.  In the event of key exposure, the system should



An, et al.                 Expires 6 May 2027                  [Page 33]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   support Over-The-Air (OTA) updates to promptly replace compromised
   keys.  Fourth, if a security breach occurs within the centralized
   management system, exposing encryption keys, the entire system should
   undergo a reset to perform a security initialization.  This process
   will generate and distribute new encryption keys to ensure the
   continued protection of sensitive data.

   Introducing an A2A overlay within the Agentic AI framework adds
   further security-sensitive surfaces that need to be addressed.  Each
   agent MUST have a verifiable identity and a signed Agent Card whose
   contents (e.g., advertised ACFs, supported intents, authorization
   scopes) are authenticated by the Cyber Security System.  A2A messages
   between agents MUST be mutually authenticated, integrity-protected,
   and confidentiality-protected, and SHOULD carry replay protection
   suitable for vehicular environments.  Agents SHOULD operate under a
   least-privilege capability model: an SDV Agent cannot reconfigure RSU
   or MNO ACFs unless explicitly authorized, and a compromised agent's
   credentials MUST be revocable through the same OTA mechanism used for
   cryptographic key rotation.  Finally, because A2A may carry LLM-
   generated or LLM-influenced content, deployments SHOULD apply input
   validation, prompt-injection mitigation, and human-in-the-loop
   oversight for safety-critical decisions, consistent with the agentic-
   AI principles discussed in [I-D.jeong-nmrg-agentic-ai-framework],
   [I-D.jadoon-nmrg-agentic-ai-autonomous-networks], and
   [I-D.hong-nmrg-agenticai-ps].

10.  References

10.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,
              <https://www.rfc-editor.org/info/rfc2119>.

   [RFC6241]  Enns, R., Ed., Bjorklund, M., Ed., Schoenwaelder, J., Ed.,
              and A. Bierman, Ed., "Network Configuration Protocol
              (NETCONF)", RFC 6241, DOI 10.17487/RFC6241, June 2011,
              <https://www.rfc-editor.org/info/rfc6241>.

   [RFC7950]  Bjorklund, M., Ed., "The YANG 1.1 Data Modeling Language",
              RFC 7950, DOI 10.17487/RFC7950, August 2016,
              <https://www.rfc-editor.org/info/rfc7950>.

   [RFC7951]  Lhotka, L., "JSON Encoding of Data Modeled with YANG",
              RFC 7951, DOI 10.17487/RFC7951, August 2016,
              <https://www.rfc-editor.org/info/rfc7951>.




An, et al.                 Expires 6 May 2027                  [Page 34]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   [RFC8040]  Bierman, A., Bjorklund, M., and K. Watsen, "RESTCONF
              Protocol", RFC 8040, DOI 10.17487/RFC8040, January 2017,
              <https://www.rfc-editor.org/info/rfc8040>.

   [RFC8329]  Lopez, D., Lopez, E., Dunbar, L., Strassner, J., and R.
              Kumar, "Framework for Interface to Network Security
              Functions", RFC 8329, DOI 10.17487/RFC8329, February 2018,
              <https://www.rfc-editor.org/info/rfc8329>.

   [RFC8799]  Carpenter, B. and B. Liu, "Limited Domains and Internet
              Protocols", RFC 8799, DOI 10.17487/RFC8799, July 2020,
              <https://www.rfc-editor.org/info/rfc8799>.

10.2.  Informative References

   [I-D.bernardos-nmrg-agentic-network-optimization]
              Bernardos, C. J., Mourad, A., and M. A. Jadoon, "Solutions
              for enabling agentic sensing with network optimization",
              Work in Progress, Internet-Draft, draft-bernardos-nmrg-
              agentic-network-optimization-02, 2 October 2026,
              <https://datatracker.ietf.org/doc/html/draft-bernardos-
              nmrg-agentic-network-optimization-02>.

   [I-D.chuyi-nmrg-agentic-network-inference]
              Guo, C., "Agentic Network Architecture and Protocol for
              Supporting Agent Interconnection Communication and Multi-
              level Inference", Work in Progress, Internet-Draft, draft-
              chuyi-nmrg-agentic-network-inference-01, 6 July 2026,
              <https://datatracker.ietf.org/doc/html/draft-chuyi-nmrg-
              agentic-network-inference-01>.

   [I-D.hong-nmrg-agenticai-ps]
              Hong, Y., Youn, J., Wu, Q., and B. Claise, "Motivations
              and Problem Statement of Agentic AI for network
              management", Work in Progress, Internet-Draft, draft-hong-
              nmrg-agenticai-ps-02, 5 July 2026,
              <https://datatracker.ietf.org/doc/html/draft-hong-nmrg-
              agenticai-ps-02>.

   [I-D.ietf-nmop-network-incident-yang]
              Hu, T., Contreras, L. M., Wu, Q., Davis, N., and C. Feng,
              "A YANG Data Model for Network Incident Management", Work
              in Progress, Internet-Draft, draft-ietf-nmop-network-
              incident-yang-20, 16 October 2026,
              <https://datatracker.ietf.org/doc/html/draft-ietf-nmop-
              network-incident-yang-20>.





An, et al.                 Expires 6 May 2027                  [Page 35]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   [I-D.irtf-coinrg-use-cases]
              Kunze, I., Wehrle, K., Trossen, D., Montpetit, M., de Foy,
              X., Griffin, D., and M. Rio, "Use Cases for In-Network
              Computing", Work in Progress, Internet-Draft, draft-irtf-
              coinrg-use-cases-07, 4 December 2024,
              <https://datatracker.ietf.org/doc/html/draft-irtf-coinrg-
              use-cases-07>.

   [I-D.jadoon-nmrg-agentic-ai-autonomous-networks]
              Jadoon, M. A., Robitzsch, S., and C. J. Bernardos,
              "Agentic AI Architectural Principles for Autonomous
              Computer Networks", Work in Progress, Internet-Draft,
              draft-jadoon-nmrg-agentic-ai-autonomous-networks-00, 2
              March 2026, <https://datatracker.ietf.org/doc/html/draft-
              jadoon-nmrg-agentic-ai-autonomous-networks-00>.

   [I-D.jeong-nmrg-agentic-ai-framework]
              Jeong, J. P., Gu, M., Ahn, Y., Park, J., and B. R. An, "An
              Agentic AI Framework for the Management of Networks,
              Security, and AI Services", Work in Progress, Internet-
              Draft, draft-jeong-nmrg-agentic-ai-framework-00, 2
              November 2026, <https://datatracker.ietf.org/doc/html/
              draft-jeong-nmrg-agentic-ai-framework-00>.

   [I-D.rosenberg-ai-protocols]
              Rosenberg, J. and C. F. Jennings, "Framework, Use Cases
              and Requirements for AI Agent Protocols", Work in
              Progress, Internet-Draft, draft-rosenberg-ai-protocols-00,
              5 May 2025, <https://datatracker.ietf.org/doc/html/draft-
              rosenberg-ai-protocols-00>.

   [I-D.wmz-nmrg-agent-ndt-arch]
              Wu, Q., Zhou, C., Contreras, L. M., Han, S., and Y. Hong,
              "Network Digital Twin and Agentic AI based Architecture
              for AI-driven Network Operations", Work in Progress,
              Internet-Draft, draft-wmz-nmrg-agent-ndt-arch-06, 28
              August 2026, <https://datatracker.ietf.org/doc/html/draft-
              wmz-nmrg-agent-ndt-arch-06>.

   [I-D.yang-nmrg-a2a-nm]
              Lopez, D., Moreno, N. R., Tailhardat, L., Ma, Q., Wu, Q.,
              Yang, Y., and S. Prabhu, "Applicability of A2A to the
              Network Management", Work in Progress, Internet-Draft,
              draft-yang-nmrg-a2a-nm-04, 2 November 2026,
              <https://datatracker.ietf.org/doc/html/draft-yang-nmrg-
              a2a-nm-04>.





An, et al.                 Expires 6 May 2027                  [Page 36]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   [I-D.yang-nmrg-mcp-nm]
              Yang, Y., Wu, Q., Lopez, D., Moreno, N. R., Tailhardat,
              L., and S. Prabhu, "Applicability of MCP for the Network
              Management", Work in Progress, Internet-Draft, draft-yang-
              nmrg-mcp-nm-04, 2 November 2026,
              <https://datatracker.ietf.org/doc/html/draft-yang-nmrg-
              mcp-nm-04>.

   [I-D.yao-tsvwg-cco-problem-statement-and-usecases]
              Yao, K., Shiping, X., Li, Y., Huang, H., and D. KUTSCHER,
              "Collective Communication Optimization: Problem Statement
              and Use cases", Work in Progress, Internet-Draft, draft-
              yao-tsvwg-cco-problem-statement-and-usecases-00, 23
              October 2023, <https://datatracker.ietf.org/doc/html/
              draft-yao-tsvwg-cco-problem-statement-and-usecases-00>.

Acknowledgments

   This work was supported by the Institute of Information &
   Communications Technology Planning & Evaluation (IITP) grant funded
   by the Korea Ministry of Science and ICT (MSIT) (No.RS-2024-00397615,
   Development of an automotive software platform for Software-Defined-
   Vehicle (SDV) integrated with an AI framework required for
   intelligent vehicles).

   This work was in part supported by the Institute of Information &
   Communications Technology Planning & Evaluation (IITP) grant funded
   by the Korea MSIT (No.  RS-2024-00398199, Standard Development of SDV
   Software Framework for Intelligent Convergence Services).

Authors' Addresses

   Byoungman Robert An (editor)
   Intelligent Information R and D Division Mobility Platform Research Center
   Korea Electronics Technology Institute
   Global R and D Center 6th floor
   #22, Daewangpangyo-ro 712beon-gil
   Seongnam
   Gyeonggi-Do
   13488
   Republic of Korea
   Phone: +82 31 739 7463
   Email: bman@keti.re.kr
   URI:   https://www.keti.re.kr/eng/main/main.php







An, et al.                 Expires 6 May 2027                  [Page 37]

Internet-Draft   Agentic AI Framework for SDVs in C-ITS    November 2026


   Jaehoon Paul Jeong (editor)
   Department of Computer Science & Engineering
   Sungkyunkwan University
   2066 Seobu-Ro, Jangan-Gu
   Suwon
   Gyeonggi-Do
   16419
   Republic of Korea
   Phone: +82 31 299 4957
   Email: pauljeong@skku.edu
   URI:   http://iotlab.skku.edu/people-jaehoon-jeong.php


   Pusik Park
   Intelligent Information R and D Division Mobility Platform Research Center
   Korea Electronics Technology Institute
   Global R and D Center 6th floor
   #22, Daewangpangyo-ro 712beon-gil
   Seongnam
   Gyeonggi-Do
   13488
   Republic of Korea
   Phone: +82 31 739 7507
   Email: pusik.park@keti.re.kr
   URI:   https://www.keti.re.kr/eng/main/main.php


   Soohyun Jang
   Intelligent Information R and D Division Mobility Platform Research Center
   Korea Electronics Technology Institute
   Global R and D Center 6th floor
   #22, Daewangpangyo-ro 712beon-gil
   Seongnam
   Gyeonggi-Do
   13488
   Republic of Korea
   Phone: +82 31 739 7436
   Email: shjang@keti.re.kr
   URI:   https://www.keti.re.kr/eng/main/main.php












An, et al.                 Expires 6 May 2027                  [Page 38]