Wiki
Clone wikimeetings / 131220_webex
Minutes Webex 20 December 2013, 6TiSCH WG
Note: timestamps in PST.
Taking notes (using Etherpad)
- Xavi Vilajosana
- Pascal Thubert
- Thomas Watteyne
Present (alphabetically)
- Alaeddine Weslati
- Cedric Adjih
- Diego Dujovne
- Giuseppe Piro
- Nicola Accettura
- Pascal Thubert
- Pat Kinney
- Pouria Zand
- Qin Wang
- Raghuram Sudhaakar
- Sedat Gormus
- Thomas Watteyne
- Tina Tsou
- Xavi Vilajosana
Recording
- Webex recording (audio+slides,streaming)
- https://cisco.webex.com/ciscosales/lsr.php?AT=pb&SP=MC&rID=73985697&rKey=fef21ec3b4c68ddf [65min]
Slides
- slides_131220_webex.ppt: slides shared during the call
Action Items
- Thomas to send an email to the ML during the 2 next weeks to make clear that there is no meeting.
- Pascal take "schedule" vs "slotframe" discussion to ML.
- Pascal to detail use of chunks in architecture draft
Agenda
- Administrivia [2min]
- Approval agenda
- Approval minutes last call
- Definitions [15min]
- cell
- unscheduled cell
- scheduled cell
- to schedule a cell
- broadcast cell
- hard cell
- soft cell
- to reallocate a cell
- chunk
- chunk ownership delegation
- chunk ownership appropriation
- quick status update WIP drafts [5min]
- draft-dujovne-6tisch-on-the-fly
- draft-wang-6tisch-6top-interface
- draft-piro-6tisch-security-issues-01.txt [10min]
- AOB [1min]
Minutes
- [08.05] Meeting starts
- Pascal starts recording
- [Thomas] This is the last call of the year. NO call in the next 2 weeks (12/27 and 01/03).
- Administrivia [2min]
- Approval agenda
No issues raised. Agenda approved.
- Approval minutes last call
No issues raised. minutes approved.
- Last call of the year. No more calls during next two weeks.
Action item: Thomas to send an email to the ML during the 2 next weeks to make clear that there is no meeting.
- Maria Rita travelling, progressing terminology draft on her behalf.
- list of terms to talk about are presented.
- Guissepe will update us about the status of the security draft that was submitted earlier before IETF88 meeting in Vancouver
- Approval agenda
- Definitions [15min]
- cell
A single element in the TSCH schedule, identified by a slotOffset, a channelOffset, a slotframeHandle. A cell can be scheduled or unscheduled.
- updated to make it shorter and clearer.
- [Pascal] The "schedule" is an abstraction owned by 6top. Should we replace "schedule" by "slotframe"? We refer to the matrix of cells.
Action item: Pascal take "schedule" vs "slotframe" discussion to ML.
- unscheduled cell
A cell which is not used by the IEEE802.15.4e TSCH implementation.
- [Pascal] Term "cell" is used for an entry in the matrix and at some point is used to identify the unit of transmission.
- [Pascal] Unit of transmission, we should agree either we use this abuse of notation or not.
- scheduled cell
A cell which is assigned a neighbor MAC address (possible the broadcast address), and one or more of the following flags: TX, RX, shared, timeskeeping. A scheduled cell can be used by the IEEE802.15.4e TSCH implementation to communicate. A scheduled cell can be a hard cell or a soft cell.
- removed"hard" flag as this is part of 6top and not part of the cell. What is hard or soft is the unit of transmission, not the cell itself.
- Unit of transmission seems a good candidate to describe the use of a cell and separate from cell (container). The term "unit of transmission" can be changed in a future.
- to schedule a cell
The action of turning an unscheduled cell into a scheduled cell.
- broadcast cell
a scheduled cell with neighbor MAC address the broadcast address.
- hard cell
A scheduled cell which the 6top sublayer cannot reallocate.
- soft cell
A scheduled cell which the 6top sublayer can reallocate.
- to reallocate a cell
the action by the 6top sublayer of changing the slotOffset and/or channelOffset of a soft cell.
- [Thomas] is relocate (as now it is unit of transmission)?
- chunk:
a well-known list of non-contiguous (i.e., well-distributed in time and frequency) cells within a slotframe; a chunk represents a portion of a slotframe that can be managed separately by an individual node. Chunks can be pre-programmed, or can be computed by the PCE when the network bootstraps.
- [Thomas] Non-contiguous seem not correct. They can be contiguous. So recommends removing from definition.
- [Thomas] State also that the way chunks are allocated is out of the scope.
- [Thomas] Missing the concept of identifier of the chunk.
- [Pascal] Remove the last sentence and reword so that the meaning of the sentence indicates that the chunk is global and known by everyone.
- [Diego] Do we allow overlap of different chunks?
- [Pascal] Yes.
- [Qin] How do nodes that own different chunks communicate?
- [Pascal] A node can have multiple chunks. And can use them according to a policy. Important to know if chunks are overlapping or not.
Action item: Pascal to detail use of chunks in architecture draft
- chunk ownership delegation
The process by which an individual node obtains a chunk to manage based on point-to-point interaction with a PCE.
- [Thomas] Not use the PCE as an example, generalize to external entity.
- chunk ownership appropriation:
The process by which an individual node obtains a chunk to manage based on peer-to-peer interaction with its neighbors
- [Thomas] we need to skip detailing in the description specific methods (e.g PCE)
- [Pascal] delegation multiple hops away a cell is allocated, appropriation is one hop, no owner but by means of negotiation
- [Qin] it seems very similar to 6top negotiation.
- [Pascal] unit of tx between A and B and 6top in A negotiates with 6top in B (schedule time frame).
- [Pascal] A chunk delegation or appropriation is a boot-time thing (or a ramp-up thing). The idea is to allocate some chunks (set of cells) to be used to allocate particular units of transmission.
- [Thomas] the mechanism itself (only mechanism) seems very close to 6top negotiation mechanism.
- [Qin] 6top-to-6top should support ownership appropriation?
- [Thomas] the 6top interface should have the methods to manipulate chunks as it has now methods to manipulate soft cells.
- [Qin] we need to define some specific mechanism to support that and this is almost identical to soft cell negotiation.
- [Pascal] the difference is that the chunk negotiation is broadcast, one node to all neighbors to indicate the presence of a chunk
- cell
- [08:51] Quick status update WIP drafts [5min]
- draft-dujovne-6tisch-on-the-fly
- Diego speaks about 10 issues logged in issue tracker
- other issues not update in the slide.
- https://bitbucket.org/6tisch/draft-dujovne-6tisch-on-the-fly/issues?status=new&status=open has the issues tracked in bitbucket since this is a personal submission.
- Thomas suggest to group issues a bit more to limit the overhead
- [Diego] Should be CC ML on all issue updates?
- [Thomas] Maybe. Thoughts?
- [Pascal] better to not have issue notification to the ML for non-working group documents.
- draft-wang-6tisch-6top-interface
- [Qin] we need to define the MIB tables defining the content of the 6top layer. This is an intermediate step to achieve a useful yang model. The real format is not important for that tables as they will not appear in the draft.
- draft-dujovne-6tisch-on-the-fly
- draft-piro-6tisch-security-issues-01.txt [10min]
- new version submitted.
- the new document takes the feedback received from the ML into account.
- use 6top layer to build the secure context. 6top and MAC interact using the commands defined in 6top draft.
- [Thomas] Call for reviewers for this security drafts.
Action item: Thomas to call for reviewers of draft-piro-6tisch-security-issues-01.txt on ML.
- draft-richardson-6tisch-security-architecture-00
- Michael Richardson submitted a new draft on security, the draft will evolve as discussions evolve.
- goal is not a solution draft but a high level architecture. *9:07 AOB [1min]
- next meeting will be the 10th of January.
- happy holidays!
- [09.10] Meeting ends
Updated