Ming,
My disagreement (or at least clarification), is that, in my understanding:
- The MLD layer (upper MAC sublayer) maintains the entire “master” block ack status, including the ack state for all the links. That is, when any frame is received and/or BA’d at an individual link, the MLD level block ack status must be updated to know this. (There might be some delay, etc., here, as the implementation may not need that information at the MLD level immediately – but in the standard we generally don’t get into such discussions.)
- Each link can be implemented to either also have a copy of this completely/master state, or it could have only a subset of that state which includes at least all the frame received/ack’d over that link, and optionally some/all of the state from other links. The latter part is not important, generally. If the links only have a subset of the master state, and a frame is retried over a different link, the receiving links need to (quickly?) coordinate to make sure the BA state is consistent. I think in practice this ‘just works’, but I’d have to think about it.
So, where I got hung up was that I read your email to imply that the MLD is _only_ handling the delivery of BA state between/among the links. But, I think it is important that the MLD also has the master BA state. That latter part seemed to be missing in your description.
Mark
Hello Mark
I am not sure I get your point. Which part do you disagree?
From this, it can be seen that the MLD Upper MAC sublayer manages the scoreboarding for all MPDUs in this BA session, on all the links. The scoreboarding in the MLD Lower MAC sublayer is only the scoreboard for MPDUs that happen to be received over that link
[Ming] agree
This is just a local link optimization. HT-immediate block ack can be implemented by responding based on the Lower MAC sublayer’s (partial) scoreboard, or optionally with additional information about other links if the implementation desires to do so.
[Ming] I am not sure what do you mean by “This is just a local link optimization”. My question which scoreboard does an affiliated STA use when it responds with BA. I assume it is local scoreboard, is this correct? My follow up question is how does each affiliated STA maintain its own scoreboard by following HT-immediate block ack architecture.
Best wishes
Ming Gan
Ming,
I disagree with how you describe scoreboarding based on 21/577r5. I reference this text:
When MLO is being used, the “Block Ack Scoreboarding” block in the MLD Upper MAC sublayer manages the BA status of the MPDUs (of this BA session) that are received on any setup link. The “Block Ack Scoreboarding” block in the MLD Lower MAC sublayer manages the BA status of the MPDUs (of this BA session) that are received on this link. It may convey BA status of the MPDUs received on another link if it obtained such info from the other link via the MLD Upper MAC sublayer.
From this, it can be seen that the MLD Upper MAC sublayer manages the scoreboarding for all MPDUs in this BA session, on all the links. The scoreboarding in the MLD Lower MAC sublayer is only the scoreboard for MPDUs that happen to be received over that link, and optionally for some/all other MPDUs on other links. This is just a local link optimization. HT-immediate block ack can be implemented by responding based on the Lower MAC sublayer’s (partial) scoreboard, or optionally with additional information about other links if the implementation desires to do so.
Mark
Hello Mark
Sorry for missing your email. Based on 21/577r5, each link maintains its own scoreboard and the MLD Upper MAC sublayer optionally delivers the BA record on one link to the MLD Lower MAC sublayer of other links. Now it seems the size of the scoreboard of each link is quite important since each STA will use this to provide reception status. However, it is still confusing how to implement the HT-immediate block ack architecture, including WinSizeO, WinSizeR. What do you think about it?
Best wishes
Ming Gan
Hi Mark,
Yes, adding it to your 21/1111 sounds good to me. I think that would be a good clarification.
Thanks,
Duncan
CAUTION: This email originated from outside of the organization.
By the way, we could add this to document 11-21/1111, if you are worried about touching 11-21/0577 since it has already passed straw poll.
Mark
Ming/Duncan,
Should we add a statement that the Block Ack agreement is between the peer MLDs, the scoreboarding function that is in the Lower MAC sublayer is just a “link local” optimization that may provide partial information about block ack status? That is, it is not where the overall block ack agreement is negotiated or maintained.
Mark
Hi Ming,
The diagram and associated text made no such requirement/assumption since they are meant for describing the functions and not how the detailed parameters are set or negotiated, etc.
Thanks,
Duncan
From: Ganming(Ming Gan) <