Minutes of the meeting of Thing-to-Thing Research Group (T2TRG) April 7th, 2016. IETF #95, Buenos Aires, Argentina Chairs: Carsten Bormann & Ari Keränen 16:20 (Carsten) RG overview, status This is a summary meeting. Research group has had multiple meetings before official chartering and also earlier this year. Three RG documents in progress and outreach activities. Plans for more co-located meetings with academic and open source communities. Juan Carlos Zamora: Co-location plans for IETF in Berlin? Carsten: RIOT summit not in the same hotel, but in Berlin right before IETF. 16:30 (Ari) RESTful design draft-keranen-t2trg-rest-iot Guidance document for designing IoT systems, relevant terminology, pointers to relevant references. In the latest version idempotency and application state clarified. More terminology. Also clarified what's different with IoT for RESTful design. Remains to be done: media types, hypermedia-driven application, design patterns. Bob Moskowitz: REST definition. Sometimes servers keep state, when it does not apply to REST. Do we call these systems RESTful? Matthias: Resource state is on the server side. Problem is with Application state definition. Should we stick to Roy F thesis or should we reflect the current language by people? Bob nods to more reflect current language and usage rather than thesis. Hannes: +1 to more discussion. Are events already discussed in the group? Ari: Is on the agenda, not much there yet. Hannes: Pub/Sub broker and sleepiness requirements on EPs, how do they map to REST? Good to discuss in the draft. Ari: Design patterns part should reflect that. Carsten: Target is not to define "REST" but to see what does it bring to IoT. Might even drop a specific REST constraint if it is not helpful. 16:35 (Mohit) secure bootstrapping survey draft-sarikaya-t2trg-sbootstrapping Definition of Secure bootstrapping Broader understanding of IoT devices taken into account. Hannes: definition there is not OCFs nor OMAs, might be confusing. Mohit: It is from the security considerations document Hannes: It is confusing when it uses a different definition under the same term Mohit: Happy to incorporate definition; please send to the list Managed Methods; AAA, GBA, LWM2M, Kerberos P2P ad hoc Methods; diffie-hellman + MitM Oportunistic / leap of faith Hybrid Methods End of life and re-bootstrapping: draft-farrell-iotsi Carsten: Terminology seems to be a bigger necessity. 16:45 Chairs Outreach, cooperation Objective: Mutual Education with other IoT SDOs. Ongoing: - Close cooperation with W3C IG Web of Things (WoT) - OCF (once so far, more to come) - Bluetooth SIG (starting) Carsten: Someone in the room who works with another interesting SDO? Please contact us. Georgios Karagiannis: IoT Alliance for Innovation (IOTA) EU commission, presentation there would be welcomed. OneM2M could be another one. Also OCF. Can help with contacts and cooperation. Dave Robin: BACnet and ASHRAE big in building space. Also KNX protocol. And maybe Fairhair that is interoperable with KNX and BACnet. Christian Groves: IoT Semantic Interoperability Workshop (IoTSI) showed some other organizations. One paper was survey that referenced many organizations and activities. Report coming out of the IoTSI Workshop might be interesting. Pat Thaler: Vice chair of IEEE 802. Active work together with IETF. At IEEE 802 several activities, some related to QoS and addressing. Layer 2. Quality of Services for time-sensitive networks, related to DetNet work. Good to coordinate on the things we do. Becoming apparent that many small things will not have global MAC addresses since that would run the space too fast and also manufacturing headache. Carsten: We have already a pretty good relationship with IETF, should work on the gaps. Pat: We have very active coordination effort with IETF but not so yet with IRTF. Subir Das: What is specifically going on in the W3C WoT IG collaboration and what is the goal? Carsten: On W3C Meetings we have found common areas, particular on Information Models (IMs) and Interaction Models (Thing Descriptions). Mutual input has already had significant impact in both Subir Das: ZigBee Alliance also working on semantic Data Models. Might be worth to check for meeting with IRTF. Members meeting coming in June. IRTF activities could be presented there. Can check and get back for more information. Christian Groves: For outreach, have chairs considered contacting influential IoT blogs; that could be worth while. Samita Chakrabarti: 1) Is the outreach part of the charter? 2) Can IETF WGs benefit from that? How would process work? Carsten: 1) Yes. 2) We don't want to replace IAB when it comes to liaisons. We want to have more high-level discussions more at architectural level. Ari: Comment on the relationship between WGs, and the RG: as the RG charter says, in RG scope are the topics that have standardization potential at IETF. If topic is already worked on by a WG, that would be natural place for discussion. If another SDO is working on topics that have no IETF WG, RG would be a natural place to discuss them. We will be forwarding people to right places where needed. Will Ivancic?: May want to check out Open Geospatial Consortium (http://www.opengeospatial.org/). Has also a lot of data on sensor webs etc. Carsten: At the RG we will probably not do application specific data models. There are domain-specific organizations that do that very well. Our job is probably more finding out where the orgs are and collect information instead of working on those. But important to know about the orgs. 16:51 (Matthias) Interaction models, hypermedia controls Information Model for interoperability. - About the "how" machines can access the information. - Server-side is straight-forward. Consumption of descriptions on the client side is harder. Need this to handle change - Controlling alternative similar things - Adding new things and still control old things with new things Needs for Semantic Interoperability - Information Model - Vocabulary - Interaction Model: hyper-media driven applications (HATEOAS) Interaction model with hyper-media: resource composition, atomic interactions, link relations, relation vocabulary. How it works - Client gets one entry point and follows links discovering new resources on the same thing or others. - When forms are found, they can be filled and submitted (actions). - You can bookmark resources you find to avoid having to discover every time. - Since it is done during runtime, new features can be added if needed (e.g., Access Control) CoRAL: how we convey all this information and concise representations Also more generic work ongoing on what kind of links and forms needed for IoT Summary: Semantic interop with information and interaction models. Reference scenarios for requirements and challenges; and PlugREST scenarios prototyping. Building blocks that can be used like machine-understandable links and forms. Looking for more people to join PlugRESTS. Start implementing, discuss at the mailing list, and come to PlugREST events and shape how we try to solve this. More information at: https://github.com/t2trg/ 17:03 Chairs Future work on schema interoperability: call for contributions Hypothesis from IoTSI workshop: There won't be one schema that describes the world. There won't even be one schema language. Need for translation "hub/s". Problem of data loss due to translation between schemas. Translation of data between data models vs. translation of the data model itself. IMs, DMs, but also Interaction Models. Lots of existing work on IMs, DMs, and serializations. How far can we get? What are the limits to translation? Schema Interoperability: What kind of research to be encouraged? What areas should RG take up. Hannes: Need for someone to do mapping between the popular data models. AllSeen alliance has SDK plugin for information and communication model that allows code generation. Once first steps are done, we could use code generation + translation, even for gateways. For example LwM2M to OCF model would be interesting. Dave Thaler: At IoTSI someone commented that you can generate code for each side but in general you need to write code for the mapping. Hannes: Remember the quote but said in context of Bluetooth LE where non-RESTful was used but RPC and no formal description how it works. Some inherent restrictions. Carsten: A lot of this information is buried on tons of pages of specification. Christian Groves: IoTSI workshop discussed "atoms" that if are compatible then can facilitate translation. Encourage people to look at that discussion. Matthias K: Generating code is not new and pointless to some extent (reprogramming my thing to become interoperable with one other thing does not help). Mike Amundsen: We should focus on interoperability on the message, not worry at what happens behind the interface. We should not need to know about data models etc. Hannes: There are different goals. I agree with Matthias, HATEOAS for example has more powerful concepts, but there is a reality of what people are using now and that's why we need it. Question is how do we deal with that part. Dave Robin: Some translation cannot be done without new interactions or a new way to express things. A good research topic would be to look for the design patterns and mappings between design topics. Proper semantic tagging can be used for translation. Bob M: IEEE 11073 has good examples of data modeling. That's mostly for medical devices, but good example how challenging this can be. Carsten: Everyone has their favorite; for example IETF has YANG. Please send further input to the list. 17:15 Chairs Meeting plan Encouraging professors to come and discuss ways to get academic research more involved in this. Future meetings keep pulse with IoT activities. Start with RIOT Summit: 2016-07-15/-16 (Fri Sat) Before Berlin IETF96. September at Lisbon with W3C. If folks know activities at Korea before Seoul IETF meeting, would be interested to plan something around that. Workshop about Software Updates of IoT devices coming June 13-14th at Dublin. Submission deadline May 20th.