call 4/19/2016 notes
Start: 4/19/2016 08:00 PDT
End: 4/19/2016 09:15 PDT
Present:
Jouni Korhonen
Rami Al-obaidi
Bomin Li
Kevin Bross
Ofir Mahazri
Richard Tse
Stuart Whitehead
Xhafer Kraniqi
Gareth Edwards
Steinar Bjornstad
Stefan ??
Mohammad Akhter
Actual agenda:
* Goint through the contributions for next week f2f.
* Plans with the 1904.3 specification
Discussion:
* 1904.3 D0.51 specification should be uploaded into
webpage any time soon.
* The intention is to go to TF review phase after the
f2f meeting. This means producing the D1.0 of the spec.
* Kevin Bross went through documents on:
- Redefining Timestamp Precision
- Time Accuracy of Sequence Numbers
- Control Packets for RoE
- Handling Multiple K-Characters
* Jouni asked about the K-characters being limited to 8/10B?
What about 64/66B.
* Kevin responses main effort has been on 8/10B. The solution
should apply for 64/66B as well.
* Jouni asks whether this could be handled by configuration and
expecting K-charecters to have a known location.
* Kevin points out that is not too agnostic at that point anymore
and also would put restrictions on the payload size.
* Jouni argues the only application for this is OBSAI.
* Jouni encourages to continue the discussion on the list
before the meeting.
* Kevin's notes on control packet discussion:
- We need to clarify that the Control Packets being referred
to are only for setup and management of the flow itself;
this has nothing to do with any control data that may be
embedded in the radio stream or used to modify the data
stream itself.
- Do we need a method to re-synchronize sequence numbers if
systems get out of synch, or should we just Terminate a
flow and then Initialize it again?
- Do we need to specify the starting number for a sequence number,
or can we always start with a sequence number of 0?
* Jouni points out the set of parameters/counters etc discussed here
are good and should be captured even if the control protocol would
not be there. Jouni also points out the actual protocol here, not
the parameters, should be a separate document/amendment in order
to keep the basic 1904.3 spec in schedule and short.
* Jouni asks about ACK presented in the packet subtype. Kevin says
it only applies to real control messages, not those used to carry
data e.g., CPRI control words (data streams).
* Jouni presents quickly the EtherType and RoE common header format.
This is an issue that we have to address.
* R. Tse went through the RoE use cases; this was motivated by the
call two weeks ago on the use of TS+SN in the data packet.
* Jouni encourages discussion on the reflector before the meeting
on all submitted documents.
Next Call:
* 5/3/2016 8AM PDT