Thread Links | Date Links | ||||
---|---|---|---|---|---|
Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Fair. -B
From: Das, Dibakar <dibakar.das@xxxxxxxxx>
Sent: Thursday, July 31, 2025 2:19 AM
To: Brian Hart (brianh) <brianh@xxxxxxxxx>; STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: RE: [STDS-802-11-TGBN] SP list
It sounds like you view the timing info in ICR to be the main signaling and that in ICF to be of secondary value.
From: Brian Hart (brianh) <brianh@xxxxxxxxx>
Sent: Thursday, July 31, 2025 1:48 AM
To: Das, Dibakar <dibakar.das@xxxxxxxxx>; STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: RE: [STDS-802-11-TGBN] SP list
My comments below principally relate to the ICR. Having said that, time signaling in ICF:
- Allows a peer AP who only has a long low-MCS buffered to pass on the current sharing offer.
- Perhaps for other secondary purposes (tune its ICR response to make it more realistic for the available time, make preliminary scheduling plans if an allocation occurs etc)
Best wishes
Brian
From: Das, Dibakar <dibakar.das@xxxxxxxxx>
Sent: Thursday, July 31, 2025 1:41 AM
To: Brian Hart (brianh) <brianh@xxxxxxxxx>; STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: RE: [STDS-802-11-TGBN] SP list
What is the value of time signaling in ICF if ICR anyway indicates how much time shared AP needs ?
From: Brian Hart (brianh) <00000c7561051aea-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Thursday, July 31, 2025 1:38 AM
To: STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-11-TGBN] SP list
Hi all
Let’s back up for a second.
- The fundamental use case for Co-TDMA is an AP polling its peers, and trying to figure out who can use medium time most efficiently – itself, this peer, that peer, or this combinations of peers. Specifically, we need to avoid the scenario of a peer AP with lower priority traffic being granted time first (for just video), and that peer AP takes up all the time, such that another peer AP with higher priority traffic (e.g., voice) is thereby denied access.
- We recognized in 11ax that BSR was the minimum information needed for APs to scheduling UL traffic. I don’t know why we would expect that anything less is sufficient for Co-TDMA.
- We recognized that BSR could be constructed with acceptable complexity (and that was two generations ago). APs support more clients, but equally have higher capability to support those clients. So BSR can equally be constructed with acceptable complexity
- Acknowledging that a peer AP’s knowledge of its BSS’ UL requirements may be imperfect – which can be solved by a ICR response that errs on the side of conservatism.
- If mobile APs etc foresee implementation complexity problems, we can have special values reporting “don’t really know” for each of these fields.
Best wishes
Brian
From: Das, Dibakar <dibakar.das@xxxxxxxxx>
Sent: Wednesday, July 30, 2025 9:08 AM
To: STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-11-TGBN] SP list
Hi Klaus,
Just repeating again. Having more info cant hurt theoretically. Whether its significantly useful is not clear. I think while focusing on this ICF/ICR part, we are overlooking quite some stuff about the overall protocol.
- When the shared AP doesn’t support TB PPDU (which I expect to be many APs in the field), this SP2 proposal is not useful since whatever time ICF signals will anyway be the allocated time. Medium efficiency for this can only be done through TXOP return. Similarly, for cases where there is only one shared AP active at the time.
- The ability for a sharing AP to do back-to-back TXOP grants to multiple shared APs is not very high if the medium is congested and sharing AP loses chance to regain TXOP to another STA during the 1st allocated duration. Based on our understanding that typically trigger based mechanisms are enabled when medium is congested, I would think there will be quite many environments where the sharing AP will do at most one allocation.
Regarding your comment below about TXOP return being optional, I don’t get why mention that. We don’t need two mechanisms for the same purpose because one is optional. If one cares about the problem, they would implement that feature.
Regards,
Dibakar
From: Klaus Doppler (Nokia) <0000320c1b22a542-dmarc-request@xxxxxxxxxxxxxxxxx>
Sent: Wednesday, July 30, 2025 8:50 AM
To: STDS-802-11-TGBN@xxxxxxxxxxxxxxxxx
Subject: Re: [STDS-802-11-TGBN] SP list
Samat: We prefer a simple solution based on the TXOP return mechanism that can achieve the same goal .
[Klaus: TXOP return is an optional feature right now. Sending additional frames to return TXOPs increases the overhead and does not make things simpler. From a coordinating AP perspective, it is much simpler to allocate TXOP durations to other APs when it knows approximately how much they need.
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
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
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
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
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