| Hi Ming,
I second what Abhi said. Can you indicate accepted spec text which specifies that the actual TSF values are synchronized or that the “TSF offset” in your figure is shared across links/BSS? The text you referred to in 35.3.1 specifies that the drift between APs shall be limited to +-30us and not that the absolute TSF values shall be synchronized.
And even if actual TSF values were synchronized/very close (via some future spec), then using TSF on respective links would equally apply. So what’s the motivation to use a different link’s TSF just because signaling is done on that link.
As Abhi mentioned, having “synchronized iTWTs” is still possible on multiple links by adding multiple TWT elements, each negotiating for the respective link. I think we should avoid this confusion/inconsistency of indicating multiple links in one TWT element just to save small overhead in setup, if any. All other TWT functions like suspension/resumption, termination, tear-down and Flexible TWT operation are link level and should be based on TSF of respective link to be consistent.
A couple of other comments:
1. 21/552 proposes to change that when only one link is indicated in Link ID bitmap, even then the TSF of link on which signaling is done will apply instead of the respective link on which the agreement is being setup. This is inconsistent why a different link’s TSF should apply just because signaling is done on that link. That TWT responding STA may also be setting up TWT agreements via signaling on its own link. So now it (and corresponding TWT requesting STA) need to keep track of on which link setup was performed to find which TSF applies? This is not necessary and causes inconsistency.
2. 21/552 also proposes “ An individual TWT is uniquely identified by the tuple<TWT
flow identifier, MLD MAC address of the MLD
with wich TWT requesting STA is affiliated, MLD MAC
address of the MLD with wich TWT responding STA is affiliated, Link
ID associated with the indicated link in the Link ID bitmap >”
As per baseline, iTWT agreement is identified by <TWT flow identifier, MAC address of TWT requesting STA, MAC address of TWT responding STA> and we should keep the same, the motivation to do otherwise is not clear as TWT agreements are link level.
I do see some inconsistency in existing text in 35.8.2 in D1.5, which can be fixed by modifying the second paragraph of the 35.8.2 as follows:
During the negotiation of individual TWT
agreements, a TWT requesting first STA affiliated with an MLD and a TWT responding second STA affiliated with another MLD may include multiple TWT elements where each of
the Link ID Bitmap subfields in each TWT element indicates different link(s) in
the same TWT Setup frame. The TWT
parameters provided by each TWT element shall be applied and be in reference to
the respective link that is indicated in the TWT element.
This will fix the inconsistency. 21/551 instead adds “TWT requesting STA” and “TWT responding STA” to sub-bullets of first paragraph. I don’t agree with this as just because a STA of an MLD is signaling TWT element for another STA affiliated with that MLD, it doesn’t make the first STA a TWT requesting or responding STA. That is not consistent with baseline definition.
Thanks, Kumail.
Correct the following example Taking the following fig for example, an AP MLD and a non-AP MLD setup two TWT agreements through one TWT element on link 1, the target wake up time is in reference to the TSF of link 1. If the target wake up time for link 1 is T, then target wake up time for link 2 is T-Delta, right? Here clock drift is ignored. <image003.png> Hello Abhi Could you elaborate why the case you mentioned can’t work. For the TSF topic, it is right, we discussed it several times. Fortunately, the group reached consensus on it. Each affiliated AP shall do synchronization on TSF, and can ensure the clock drift to be within +-30us. Moreover TSF offset is also provided. In this case, there is no issue to reference the TSF of any link . Taking the following fig for example, an AP MLD and non-AP MLD setup two TWT agreements through one TWT element on link 1, the target wake up time is in reference to the TSF of link 1. If the target wake up time for link 1 is T, then target wake up time for link 1 is T-Delta, right? Here clock drift is ignored. <image005.png> Best wishes Ming Gan Hi Ming,
As I explained during the TGbe MAC call, the TWT element carries TWT Parameter set which contains fields whose values apply to the link for which the TWT is being setup. The parameter set includes the Target Wake Time field which provides timing information. This timing information is absolute time (and is expressed in TSF). Therefore, unless the TSF on all the links is identical (i.e., offset is 0), the proposal in doc 552 (to allow more than one bit in the Link ID bitmap set to 1) will not work. TGbe group has discussed TSF topic multiple times and has agreed that each AP of an AP MLD can have different TSFs.
<image006.jpg>
If you are looking to setup overlapping individual TWTs, the existing spec (D1.5) allows you to do that by including more than one TWT element in the TWT setup frames. In this case, each TWT element carries a TWT parameter set with only one bit in the Link ID bitmap is set to 1 and the value in the Target Wake Time subfield is with respect to the TSF of the link matching the bit set to 1 in the Link ID bitmap. Relevant text can be found in 35.8.2 (Individual TWT agreements)
<image009.jpg> Hope this helps!
Regards,
Abhi
Hello all This email is for the discussion of Individual TWT issue. If you have comments, please let me know. Best wishes Ming Gan
To unsubscribe from the STDS-802-11-TGBE list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBE&A=1
To unsubscribe from the STDS-802-11-TGBE list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBE&A=1
To unsubscribe from the STDS-802-11-TGBE list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBE&A=1 <oledata.mso>
To unsubscribe from the STDS-802-11-TGBE list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBE&A=1
|