Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [STDS-802-11-TGBN] PDT MAC NPCA document 11-25-936 updated to r7



On Mon, Jul 28, 2025 at 4:05 AM Salvatore Talarico (Nokia) <salvatore.talarico@xxxxxxxxx> wrote:

Matt,

 

  Please find inline some clarifications.

 

 From: Matthew Fischer <00000959766b2ff5-dmarc-request@xxxxxxxxxxxxxxxxx>

Sent: Monday, July 28, 2025 12:29 PM

 

Salvatore,

 

On Mon, Jul 28, 2025 at 3:03 AM Salvatore Talarico (Nokia) <0000391239bb5db0-dmarc-request@xxxxxxxxxxxxxxxxx> wrote:

Hi Matt,

 

Please find inline some further comments from my side.


Salvatore,

 

1) 

what is the technical reason to substract the largest between switch back delays of the STA and its intended recipients. 

This would also lead to effectively have STAs returning to primary channel at different times if bullet 1 of section 37.10.3 is kept, which may cause medium sync issues.

Several people have suggested that differing switch back times will cause medium sync problems on the BSS primary channel.

This should not be the case:

 

Regardless of when any STA switches back, there should never be a channel sync problem because every STA that switched did so based on received OBSS information

which is nominally the same information. In such a case, each STA can switch back at different times so long as all STAs switch back BEFORE the end of the OBSS occupancy.

So long as this is true, all returnees will only be able to sit and wait for the end of the OBSS busy time.

If you return earlier, it does not matter, because you are just going to sit and wait.

 

[ST] I agree, but the assumption here is that STA will receive same OBSS information, which is not always the case. Also on top of medium sync issue, my understanding is that by letting a STA switch earlier there would be a less efficient use of the resources as STAs that switch earlier would be able to neither transmit or receive anything on the primary, while they could have stayed longer on the NPCA primary channel and perhaps continue communication.

 

Assuming that two STAs receive different OBSS information to invoke NPCA, then the switch back delay is not the thing that is causing them to return to the main channel out of sync.

It is the OBSS information that they received which causes this mis-sync.

But I could argue that this is not a mis-sync anyway, because, for a STA that has different OBSS information, that information is still valid and still prevents the STA from using the main channel.

 

[ST] Understood – One question: the STAs that switching back to the primary channel while there is still an OBSS will maintain the NAV information, and this won’t be reset even if they switch earlier?   


The intent is that when a STA switches to NPCA, it does this for a duration equal to the OBSS PPDU or the OBSS TXOP.
In that case, either:
a) the NAV on the BSS pch remains in place and continues to count down while the STA is on the NPCA pch, the BSS pch NAV will be at or near zero when the STA switches back
b) the NAV on the BSS pch does not matter because the STA remains on the NPCA pch for exactly the duration of the OBSS activity back on the BSS pch
The text is not actually clear as to which is the method, but these are logically equivalent implementation choices.
The text should eventually clarify this as it is just implicit right now.
If a STA switches back early, then in order to conform to this expectation, the STA would need to continue to obey the NAV on the BSS pch that had been set previously by the OBSS activity
Again, this exactness does not currently exist in the text, but for fairness and channel state sync reasons, I do not think that anyone would object to such a description.

 

 One might propose that if the AP switches, then all STAs should switch, since for STAs that remain on the main channel, communication with the AP will not be possible.

The opposite is not true.

That is, if a STA switches but the AP does not, then there is little that can be done or that should be done.

This is more of a switch or not switch, not a difference in detected OBSS durations.

Such a proposal is problematic because the AP has detected an OBSS BUSY and is not supposed to transmit, so how can it communicate the switch information to the STAs?

 

[ST] Agree that if AP switches back, then all STAs should also do so. However, my comment was for the case when the AP does not switch back because its switch back is shorter. In this case, the AP could still serve those STAs with similar switch back delay, while the STAs with higher switch back can start return to the primary channel.  


The statement was actually referring to the switch from BSS pch to NPCA pch, but is similarly applicable to the reverse case.

Regarding different switch back times:
As noted below, the subtraction in the next revision will be only for that STA's switch back delay time.

As to your rationale to support this change, sure, there could be some additional time for all of the late switchers to exchange data - I do not believe that it is very significant.
If the NPCA duration is 2 msec and the max difference in the sb delays is 128 usec, then the additional time is 5%.
If the NPCA duration is longer, then the savings is less.
In any case, with the change, you have what you are interested in.
 


To unsubscribe from the STDS-802-11-TGBN list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBN&A=1