T2TRG summary meeting

Time: Monday, December 12th, 15:30-17:30 UTC

NEW Meetecho:
https://meetings.conf.meetecho.com/interim/?short=87d786a7-c3f3-4c8f-abc7-54567d422ed3

Intro, RG status, upcoming meetings and activities

Dan gave update on security draft. Reviews welcome.

MCR: interested to get progress on the doc. Need to step forward since
doc was started; not using bootstrapping but onboarding term. Need
top-to-bottom, assume it's 2022. OK to say we had old terms. Happy to
send PRs. Need RG wide consensus to change the title.

Dan: title was already changed to "initial setup". Bootstrapping was
removed from new title. Try not to favor any specific terminology. Use
terms used by different protocols and standards. If don't mention
bootstrap term, can't understand those specs.

MCR: need to re-write like section 2 about BPP has bootstrapping info --
has changed. Major contribution to nail down those terms.

CB: can recognize industry terms, but these are not always sharply
defined. When ambiguous may want to qualify terms.

(Chat: Michael Richardson: +1 to what Carsten says, but where we find it
ambiguous, then we SHOULD actually mention the ambiguity, and then NAIL
DOWN a proper definition. Carsten Bormann: +1!)

Dan: Thanks! PRs and list discussion welcome

SECCORE update

Göran Selander (GS) presented: looking for drivers / pen holders to
progress the work. Number of topics. Ongoing work on amplification
attacks. Two candidate topics and looking forward to going into details
in follow-up meetings: 1) Software updates using CoAP group
communication and 2) Delegation of rights using ACE-OAuth. Some ideas
already, waiting for people to have time to engage.

Ari: meetings coming?

GS: probably similar to RG summary meetings, one before/after IETFs.

Amplification attacks using CoAP

John Mattsson (JM) presented changes in the latest version (slide 2).
Many new comments since -02, including discussion on implementations.
Important questions for RG: what is long term goal and scope? Own idea
raising awareness in general. Also should talk about mitigations? Should
discuss non-amplification DoS attacks? If material, why not, but usually
involves hacking devices, so quite different attacks. IETF can mitigate
amplification with protocols.

Polls:

• how many have read a version? 6 hands

• Who is in favor? 8 raised (1 not raised)

CB: any comments on why should not be adopted?
(no response)

Update on "A Taxonomy of operational security considerations for manufacturer installed keys and Trust Anchors"

MCR presented briefly background, goals, and status. If people have
ideas for shorter title, comments welcome. Many have been doing such
process for a long time but not widely known. Have presented some 10
times for big groups. Some more private presentations, but details often
under NDA. Got feedback that "if could refer to public source could
simplify this". Links in the slides to various talks.

Goal is to give names on the ways to provision and secure keys (not
evaluation).

Next steps to get doc adopted. Goal to get published in 2023

CB: important contribution to have stable documentation. Phrasing as
terminology instead of evaluation makes it easier to have RG consensus
for publishing.

Polls: Who has read a version of the doc? 4 raised, 6 not

MCR: update just before IETF, that version is pretty stable and I'm
happy with it. Some gaps, but 92% way there.

CB: can underscore request to review by going for RG adoption call.

Poll: who in favor? 6 raised (0 do not raise)

CB: seems we have strong support and can issue call for adoption at the
list with this as input

OneDM and iotschema.org update

MJK presented status of OneDM. Focus shifted from model convergence to
model contributions. Have people build out the first step. Have now SDF
that is almost there. Opening up so people can create their own
repositories in the same space. Technical work on semantic proxy and
mappings.
OMA and OCF models have repositories already, starting work on Bluetooth
models, reaching out to CSA/Matter.
Also in OneDM looking also at digital twin models. Focusing on if we
have everything we need to do the modeling. How to model also relations
that come from ontologies and graphs. Building use case for SDF to
support abstract references in definitions and link data types. Azure
DTDL examples and other related work.

Milan M: the model contributions, for mapping to common model/format?

MJK: expected outcome to have people participating in common modeling
without changing their models. Also for semantic proxy mappings/bindings
would like to try any model to use SDF and common patterns. Layer of
translation and mapping tools to have practical interoperability instead
of going straight to common models. For example, OMA could use OCF
models before going for a common set of models. Meantime can publish
models with common format. SDF provides reasonable alternative for
people to adopt.

MJK gave quick update on iotschema. Started with a similar goal on
collecting models. Created RDF model with actions/events/properties like
WoT TD and SDF use. Have a small, useful set of models that have been
used with WoT for plugfests. Map to/from SDF using simple path model. If
there's RDF framework, don't need many different models to adapt. Can
reason about the models. Interesting point we need to look at. How we
could deliver RDF models in this space -- maybe translating SDF models.
Only ad-hoc governance and meetings paused at the moment. Maybe some
future in the W3C community group. Now was experiment that was
moderately successful but some aspects that people are more likely to do
in OneDM. Not integrating with schemaorg, since that's more on high
level business concepts.

CB: wonder if can get those models in SDF and include in the OneDM
playground?

MJK: interesting idea, will look into that!

SDF update and status

Carsten gave update on SDF. Looking at the digital side of Things and
describe interactions device can do in terms of affordances. Started at
OneDM, continued work in this RG, and resulted in IETF WG (ASDF). Stable
version (-12) of spec exists, small issues remaining. Also lots of work
on tools, e.g., SDF/WoT discussed later today, but also for YANG, OCF,
DTDL, OMA ecosystems and models.

Several documents out now for various extensions or alternate views of
the space. Including human-readable format based on YAML, and SDF
mapping mechanism that can augment models with additional qualities. Two
documents about links: one on model level and another on data type for
interactions with affordances. Former, on finding more information about
models, the latter about pointing to specific instances. But the
distinction is not entirely clear. For example, unit may be model level
link but its value could be provided at runtime via property that is
read. Need to glue models together in various ways so need more than
sdfRef. And also need ways to exchange links. Now need to find out if we
can express what the link contains in good way; for that will need
examples and hope to get feedback from this RG.

Also work on extension points, in particular for new qualities that may
be ecosystem specific. Currently issue in the repo. Less simple:
extending the architecture. Currently, we put protocol bindings into
mapping files, but doesn't answer if those can be described in
cross-ecosystem ways. Not easy part: how to go one meta-level above,
e.g., "mapping file for BT looks like this" or "OMA has IDs here". Can
write in English, but good to have automated process support here. ...
And even further, information like "purpose in life" or "role", location
or links to other things. All we need in the links and relations? Also
need to describe things that go into specific instances.

Brief W3C WoT update, SDF/WoT conversion

Jan Romann (JR) presented quick WoT update. Current charter until end of
Jan 2023 but likely to be extended. Thing Description (TD) and
architecture docs updated. New document on discovery. Restart with new
charter in May, including profile spec, 2.0 of the TD spec, new topics
including stronger focus on protocol bindings.

Jan presented his work on mapping between WoT TD and SDF. Both TD and
SDF work on improving interoperability, coming from different angles. TD
primarily focused on instances. Complement each other quite well. But
currently no canonical mapping between the two. The thesis proposes a
first proposal on this regard and converter implementation to automate
the conversion.

Both SDF and WoT have many similarities (serialization format,
affordances, JSO-inspired qualities). TD has JSON-LD context, security
information, and protocol bindings that are not in SDF models itself.
Raised question how we can map the WoT specific vocabulary to SDF. Here
used mapping files. JSON pointers indicating which parts of the models
are augmented. These can be then converted to TD models.
To map SDF models to WoT using new concept called Thing Models, that are
allowed to omit instance information.
Conversion process using SDF and models was able to cover all three
types (SDF, TD, TM) with just two implementations.
Challenges on different kind of nesting; wot uses linking approach
whereas SDF can use sdfThing. With slight adjustments was possible to
map to sdfThing, but used also experimental TM/TD collections. For
round-tripping, using "sdf" prefix in WoT docs and mapping files for
SDF. But needs to be revisited when making more formal spec on this.
Also, external references need to be resolved before converting between
documents.

Implemented mapping using Python. Published as library and can be
installed with pip. Also created web-app for experimenting. Converter
also featured in collection of converters made by Petri.

Always grateful for feedback and issues raised.

Summary: WoT and SDF can be mapped into each other quite well. TMs
provide good intermediary for the process. Implementation exists. More
standardization work needed on the canonical mapping. Also: how to
implement the linking process, and nested TMs/TDs in a single doc.

AK: when and where can we read more?

JR: thesis being finalized, can share when ready

Milan: key objectives here? Feasibility or to provide tools?

JR: tool was one of the objectives but also proposal for mapping that
could be used for standards process. Also demonstrated feasibility, but
that also input for standards process.

Knowledge Graphs for IoT Platform Digital Twins based on SDF

Alejandro Peñas (AP) presented his work on knowledge graphs for Digital
Twins using SDF. Key results included exploration of knowledge-storage
methods for Digital Twins, IoT-based testbed for knowledge ingestion
using SDF for device classes, and device data integration into knowledge
graph.
Using TypeDB for the knowledge graph, both class and data information.
SDF for defining the semantics of the devices, using UUID attribute to
identify instances of the classes. If detect a device without known
schema, will query SDF description. Based on description generate
automatically TypeDB query and create schema for the knowledge graph.
Using car manufacturing plant as example. Implementation using pub/sub
of data between knowledge graph consistency handler, TypeDB, and
data/class info sources. When new devices discovered, depending on
similarity to existing devices either integrate new devices or
complement/replace existing devices. Similarity implemented using
natural language processing. Can integrate information in SDF classes
using the sdfRelation extension. Also look at behavioral similarity
based on the time series data of properties.
Future work on automatic SDF generation from device specs, optimizing
the computation of the closest device, integrating devices from scratch,
and testing with real data from actual IoT platform.

MJK: great work and close to some work I've done. How much
standardization expect to exist in SDF vocabularites. Do you expect
vocabularies to be harmonized already.

Bin Xiao: good question; open for different kind of vocabularies. Here
calculating distance. Now doing search in the existing knowledge graph.
If close in distance can integrate in the KG.

MJK: great, thanks!

Wrap-up

AK: Seems we have a good way forward with the documents and other work
presented today. Let's continue the work and discussions on the list and
follow-up meetings at OneDM, ASDF, and RG work and summary meetings.