Session Date/Time: 29 Oct 2025 07:00

ASDF

Summary

The ASDF interim meeting covered updates on several drafts, including SDF Context, SDF for API Translation, and Digital Twin modeling. Key discussions revolved around the evolving syntax for context information, a proposed extension for API translation, and structural updates to the Digital Twin specification. A primary decision was made to prioritize the advancement of the NIPSY draft and its dependent SDF Protocol Mapping draft for the upcoming IETF 124 meeting.

Key Discussion Points

SDF Context and Non-Affordance Information (Carsten Bormann):
- Presented an update on the "non-affordance information" (now called SDF Context), which is structurally similar to SDF Properties. A syntactic question remains whether to use a new sdf-context keyword or extend sdf-property with a qualifier (e.g., "off-device"). The current drafts use sdf-context syntax.
- Introduced the principle of "messages as instances," which are data structures described by but distinct from the interaction model, allowing for interchange without modifying the model.
- Illustrated examples of context snapshots (static/immutable information) and instance messages as deltas using sdf-patch and mergePatch based on a message-ID reference.
- A concept of "model patches" was introduced (a data message potentially including a patch to the model itself). Concerns were raised regarding security implications and the need for a meta-model to define allowed model patches.
- Next Steps: More examples, implementation experience, deeper understanding of meta-information (e.g., thing-ID, timestamp).
SDF for API Translation (Lorenzo Vangelista on behalf of Edgar Ramos):
- Presented a new draft proposing an SDF extension to facilitate API translation between devices and AI agents. The use case highlighted was a car's device needing to translate API calls to different physical devices (e.g., traffic light camera) performing the same function due to API variations over time or across vendors.
- The proposed solution introduces qualities like template-hrep (for URL patterns following RFC6570) and template-value (for mapping values to target API parameters), and uses sdf-relation.
- Discussion:
  - Carsten Bormann suggested considering an "N by M" problem with an intermediate "red star" abstraction layer for translation, distinguishing between abstraction (vendor API to red star) and concretization (red star to application-specific API).
  - A participant noted the potential need for a dictionary of terms to handle homophones and synonyms in translations.
  - Bart van der Poel questioned the scope, suggesting the mechanism might primarily address syntactic differences rather than entirely different API approaches, and emphasized the need for validation with real API implementations. He also advised against integrating this work directly into sdf-protocol-mapping to avoid delaying existing normative work for NIPC.
  - Carsten Bormann viewed this as a research-oriented exploration that demonstrates SDF's extensibility rather than an immediate normative requirement for existing drafts.
SDF Modeling for Digital Twin Updates (Yunjong Kim):
- Provided an update on version 2 of the Digital Twin draft, aiming for a standardized SDF approach to represent digital twins. This involves describing physical/logical entities as sdf-thing and sdf-object, capturing behaviors via sdf-property, sdf-action, sdf-event, and integrating context/relationships using sdf-context and sdf-relation.
- Key Changes: Added deep-sea and product mapping references, included tables for components and SDF elements in examples, added a healthcare system as a new domain example, and restructured chapters for clarity and to serve as both a guideline and a reference implementation. sdf-protocol-map was integrated into asset relations examples (e.g., Bluetooth for device communication).
- Domain Examples: Illustrated modeling of a marine vessel (Boat #7) with components as sdf-object and internal relationships using sdf-relation, enabling predictive maintenance and real-time monitoring. A healthcare scenario demonstrated patient ID as sdf-context and alerts as sdf-event and sdf-relation.
- Comment: Bart van der Poel appreciated the domain-specific examples and encouraged aligning the work with external SDO engagements (e.g., IEEE) to validate the SDF design for digital twin information.

Decisions and Action Items

Decision: The primary focus for the ASDF working group at IETF 124 will be to advance the NIPSY draft to completion and the associated SDF Protocol Mapping draft.
Action Item (Bart van der Poel / NIPSY and SDF Protocol Mapping authors): Prepare an overview of recently closed issues and highlight any significant or complex changes for discussion during the IETF 124 sessions. Consider submitting the drafts by Sunday (when the submission window opens) to allow more time for working group review.
Action Item (Yunjong Kim): Continue to align the Digital Twin draft with other core ASDF working group documents (e.g., main SDF, non-affordance, instance input), incorporate feedback from the interim, and extend modeling to additional domains.
Action Item (Edgar Ramos / Lorenzo Vangelista): Continue exploration and validation of the SDF for API Translation concept, focusing on proofs of concept and considering abstraction layers.

Next Steps

IETF 124 Meeting: Focus on completing the NIPSY and SDF Protocol Mapping drafts.
SDF Context: Further develop examples, gather implementation experience, and refine the understanding of associated meta-information.
SDF for API Translation: Continue as a research effort, validate the proposed mechanisms with real-world API examples, and explore the concept of "red star" abstraction.
SDF Digital Twin: Expand modeling to cover other domains (smart building, energy, factory systems) and ensure continued alignment with foundational ASDF drafts.
DINRG Interim Meeting: An upcoming DINRG interim meeting (late this year/early next) will discuss AI agents' use of IETF IoT technologies, with SDF identified as a useful mechanism for describing AI agents' interactions. Interested parties are encouraged to join the DINRG mailing list.