Clone wiki

meetings / 130823_webex

Minutes Webex 23 August 2013, 6TSCH group

Note: timestamps in PDT.

Taking notes (using Etherpad)

  1. Xavi Vilajosana
  2. Dominique Barthel
  3. Thomas Watteyne

Present (alphabetically)

  1. Carsten Bormann
  2. Diego Dujovne
  3. Dominique Barthel
  4. Elvis Vogli
  5. Kuor Hsin Chang
  6. Maria Rita Palattella
  7. Pascal Thubert
  8. Pouria Zand
  9. Qin Wang
  10. Raghuram Sudhaakar
  11. Thomas Watteyne
  12. Tom Phinney
  13. Xavi Vilajosana

Recording

Slides

Agenda

  • Administrivia [2min]
  • CBOR [10min]
  • Identifying the flows [10min]
  • Update "Basic" Draft [10min]
  • Time parent selection and EB priority [20min]
  • renaming 6TSCH [5min]
  • This Coming Week [2min]
  • AOB [1min]

Minutes

  • [08.02] Meeting starts
  • [08.02] Administrivia
    • Carsten is joining us to give an overview of CBOR. Thanks!
    • Overview of the agenda
    • Approval of the BoF Minutes

      No issues raised.
      Action item: Thomas to send final call for approval on ML.
      Action item: Thomas to upload final minutes onto meeting materials website.

    • Approval minutes webex 08/09

      No issues raised. Minutes approved.

    • Slides during the call can be accessed on Bitbucket before the call, or by flipping through slides shared in webex.
    • IETF 88 Vancouver:
      • Registration opens today
      • Don't wait too long to book hotel
      • IETF 88 is in 71 days
  • [08.08] CBOR [Carsten Bormann]
    • CBOR: Concise Binary Object Representation - Design and Overview.
    • Available at http://tools.ietf.org/html/draft-bormann-cbor-06
    • Why a new binary format? --> different design, not all formats are useful for all protocols:
      • Simple format, can be implemented in small amount of code
      • Encoded data is compact
      • No compression or even fiddling. Not as compact as using compression, but the amount of code used to produce the format is very small.
    • JSON-like format.
      • no schema required. -- schema-less
      • compact implementation
      • applicable to constrained nodes and high volume applications
      • supports all JSON data types.
      • extensible.
    • What does it look like?
      • Initial byte:
        • major type (3bits) and additional information (5bytes)
        • 8 major types: bool, int, etc.
    • CBOR vs FOO (e.g. Binary JSON)
    • Questions
      • [Pascal] Usage of CBOR? Are there any standards already using it?
      • [Carsten] Several standard are looking for applicability. Early stage though
      • [Thomas] Can CBOR encode all JSON type? I.e. can we design using JSON, knowing tha CBOR could encode it?
      • [Carsten] Yes. Do not talk about compression, rather encoding.
      • [Thomas] Is there a aeb based application which can be used to test encoding of different JSON strings for initial testing?
      • [Carsten] People have implemented it in an afternoon. I have a robust implementation I can push out if needed.
      • [Thomas] Status?
      • [Carsten] In final steps. At the IESG now.
  • [08.22] Identifying the flows [Pouria Zand, Qin Wang]
    • Discussion about identifying the flows:
      • Centralized vs distributed system manager architecture concept
      • "ME" is the management entity. Can be an upper layer (distributed scheduling) or a PCE (centralized scheduling).
      • 3 control flows:
        • ME <-> 6top: ME can be in PCE or in upper layer
        • 6top <-> TSCH
        • 6top <-> 6top: message flows between 6top of node A and 6top of node B
      • Control flows:
        • action flows: Action request - confirmation..
        • query flows:
        • Report flows:
      • example of interaction between node A and B to schedule a soft cell between them, at the initiative of node A's higher layer.
      • example of interaction for creating a hard cell:
        • Create a link between 2 nodes managed by the PCE.
        • safer to have PCE send two separate request to A and B to create a hard cell (as opposed to having A pass on the request to B)
        • [Pascal] in the future (after rechartering), we could have a train of requests in RSVP-like fashion.
        • [Thomas] we have to make sure the format agreed upon now can be reused in the future.
        • [Maria Rita] The PCE should know that a link between A and B exists when it instructs to allocate a cell from A to B.
        • [Qin] three message interfaces: ME-6top, 6top-TSCH, 6top-6top, this work focus on ME-6top first.
        • [Thomas] Let's come up with an actual description of the contents of the packet, and come again to this question in the future.
      • How to figure out the Contents of the Control Flow:
        • Select what format and content is necessary.
        • Bottom up approach: from 6top to upper layers
        • Top down approach: from ME to 6top
      • Bottom up investigation:
        • Content of Action Flow (see slides)
        • Content Query Flow:
          • not define, needs to work from top down approach.
        • [Qin] It is unclear at this stage which slots to use to transport the report flow.
        • [Thomas] In the context of the basic draft, this question is already answered.
        • [Pascal] Add more information about tracks? More commands specific for managing tracks?
        • [Thomas] Installing a mapping between cells and tracks used to be part of the list of type?
        • [Qin] List on slides are truncated

          Action item: Qin to send the document with full list to the ML.

        • [Thomas] When a mote wants to send more data, it has to contact the ME and request more links. Is that mapped in that commands?
        • [Qin] Centralized case: The PCE configures the motes on how to report alarms. The PCE can configure the track that should be used to request more bandwidth.
        • [Thomas] How does a mote let the PCE know that it is going to send data to a new destination which is not in an existing track?
        • [Qin] We need to define how a request for a new track or more bandwidth flows to the PCE.
        • [Qin] Application needs to talk to ME.

          Action Item: Take this discussion in the ML.

      • Top down investigation:
        • RPL rank is needed, rank should be an attribute of the neighbour table.
        • Need input from someone familiar with PCE.
        • [Thomas] Does this impact the format of the packets?
        • [Qin] We need to know the message requirements from the PCE.
        • [Pascal] RPL and PCE can install routes. The management entity has requirements that we need to identify, the top down approach has to show that. The flows of information in the forwarding plane.

          Action Item: Qin request people to join her discussion and define what the information is needed form a top down approach point.

  • [08.50] Update "Basic" Draft [Xavi Vilajosana]
    • Build gradient on join priorities, by adversing +1 on best join priority heard.
    • TO be used before we have a RPL DODAG in place.
    • Proposes to use the "zero" RPL Objective Function.
    • One preferred parent and one alternate.
    • Non-storing mode as a MUST implement.
    • Explains rank computation as defined in "zero" OF.
    • Use of ETX metric to compute rank increase.
    • [Pascal] normalized to predefined limits. 1 to 9. For example, assume ETX of a fairly good radio link being 1.2, the rank increment can be set to 0.4 such that this ETX is represented by 3.
  • [09.02] Time parent selection and EB priority
    > We are running out of time. This discussed is moved to next week.
  • [09.02] Renaming 6TSCH
    • Thomas announces current ranking of votes on the mailing list.
    • Will publish on the mailing list.
    • Of the proposed new names, none raised concerned with IESG.
  • [09.04] This Coming Week
    • ETX implementation according to RFC6551 and RFC6552
      • Private discussion, OpenWSN implementation
      • Lead: Xavi Vilajosana
    • Synchronization and Clock accuracy
      • See ML thread: “RPL on Basic”
      • Lead: Pouria Zand
    • Fast join
      • See ML thread: “Synchronization Issues”
      • Lead: Alfredo Grieco
    • JSON, BSON, CBOR comparison
      • Private discussion
      • Lead: Qin Wang
    • Take a stab at packet format
      • Lead: ?
    • Discussion Time parent selection and EB priority
      • Lead: ?
  • [09.07] AOB

    No other business from the attendance.

  • [09.08] Meeting ends

Updated