Hi John,
Thanks for comment.
Yes, I agree with your point. I think EIFS in Motion #72 is a concept of the channel sensing duration. My understanding is that EIFS should be modified to EIFS-DIFS+AIFS[AC], because it follows current EDCA operation.
In 21/0070r2, the first paragraph of Motion #72 is captured as a slot boundary condition f) as follows:
f) Following EIFS – DIFS + AIFSN[AC] × aSlotTime + aSIFSTime – aRxTxTurnaroundTime of idle medium on the 20 MHz channel after the last indicated busy medium as indicated by the CCA mechanism on the OCB secondary channel that is not covered by a) to d), when the duration of channel busy is not known.
In 10.23.2.4, AIFS[AC] is defined as: AIFS[AC]= AIFSN[AC] × aSlotTime + aSIFSTime. Therefore, the underlined text shows EIFS-DIFS+AIFS[AC], and aRxTxTurnaroundTime is a switching delay of PHY to change from receiving to transmitting symbol.
If you have more comment on this, please let me know. Thanks.
Best Regards,
Hanseul Hong
보낸 사람: John Kenney
보낸 날짜: 2021년 2월 2일 화요일 오후 11:27
받는 사람: Hanseul Hong
참조: STDS-802-11-TGBD@xxxxxxxxxxxxxxxxx
제목: Re: [STDS-802-11-TGBD] Discussion on Doc 11-21/0070
Hi Hanseul,
Sorry not to reply sooner.
I have a comment about Discussion 3, specifically about using EIFS when the secondary channel busy-duration is not known. I think we should be using EIFS - DIFS + AIFS[AC] instead of EIFS. This would be the correct value for EDCA. We are using EDCA when we specify AIFS instead of DIFS. In my opinion we should use EDCA consistently. I copy below what I believe is the relevant text on EIFS (802.11-2016):
10.3.2.3.7 EIFS
A DCF shall use EIFS before transmission, when it determines that the medium is idle following reception of a
frame for which the PHY-RXEND.indication primitive contained an error or a frame for which the FCS value
was not correct. Similarly, a STA’s EDCA mechanism under HCF shall use the EIFS–DIFS+AIFS[AC]
interval.
In other words, DCF uses EIFS and EDCA uses EIFS - DIFS + AIFS[AC]. Shouldn't we use the latter? Motion #72 requires us to use EDCA when the busy duration is known and DCF when the busy duration is unknown. I believe that was a mistake.
What do you think?
Best Regards,
John
On Fri, Jan 29, 2021 at 8:04 PM Hanseul Hong <hanseul.hong@xxxxxxxxxxxxxx> wrote:
Hi, Joe, John, and all,
Thanks for comments and discussion on 21/0070r1. I have uploaded 21/0070r2.
This e-mail is to invite more comments and have discussion about the resolutions of several CIDs. Let me explain some discussion points during the call.
Discussion 1: related to CID 1754
There was a question whether we need to add a note in clause 11.15.9 (e.g. A STA does not follow rules in this subclause when dot11OCBActivated is true). In my understanding, because that subclause is description only for HT STA, I’m not sure if we need to add a note that subclause. Therefore, the note is not added in this version.
Discussion 2: related to CID 1021
There was a question whether we need to add a MIB attribute(e.g. dot11VirtualCSActivatedOnOCBSecondary) about virtual CS in the OCB secondary channel. We may need the MIB attribute because an NGV STA is not required to perform virtual CS in the OCB secondary channel(optional). I want to invite experts to collect more opinions from them.
Discussion 3: related to CID 1022, CID 1227, CID 1486, CID 1790
Basically, an NGV STA transmitting a 20 MHz NGV PPDU or 20 MHz non-NGV duplicate PPDUs perform channel access mechanism based on the current EDCA rules in baseline.
However, several points are different:
1.The channel access is performed based on the medium status of the entire 20 MHz channel.
2.When either the medium of the OCB primary channel or the OCB secondary channel is determined to be busy, the 20 MHz medium is considered as busy
3.When the OCB secondary channel is busy and the duration of channel usage is not known, EIFS period sensing is perform before resuming the channel access.
Based on that, I have modified text based on the current EDCA rule, modifying the definition of the slot boundary.
The definitions of slot boundary a)-d) and h) in 21/0070 are same as condition a)-d) and f) in clause 10.23.2.4 respectively, with adding “on the 20 MHz channel” to capture the point above-mentioned point 1). (highlighted in blue)
The definitions of slot boundary e)-g) are added in this subclause to capture the following motions.
[ Motion #71]
· When the OCB primary channel is sensed as channel busy, the backoff procedure on the OCB primary channel shall be same as the backoff procedure of 10 MHz transmission
[ Motion #72]
When OCB secondary channel is sensed busy and the duration of channel busy is not known, after the channel state transitions from busy to idle, EIFS interval shall be used to detect if the channel remains idle before resuming the backoff procedure.
When OCB secondary channel is sensed busy and the duration of channel busy is known, after the channel state transitions from busy to idle, AIFS interval shall be used to detect if the channel remains idle before resuming the backoff procedure.
Note: STA is not required to decode the duration of channel busy on the OCB secondary channel.
Discussion 4: related to CID 1487
As the PHY-CCA.indication primitive is discrete value, there was a suggestion to clarify “the most recent PHY-CCA.indication”. I have checked the current baseline text and found both ways(based on recent indication/based on indication). Either way is fine for me. Please let me know if you have prefer option.
Please let me know if you have any questions or comments on each discussion point. I will try to resolve them.
Thank you
Best Regards,
Hanseul Hong
To unsubscribe from the STDS-802-11-TGBD list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBD&A=1
--
John Kenney
Director and Sr. Principal Researcher
Toyota InfoTech Labs
465 Bernardo Avenue
Mountain View, CA 94043
Tel: 650-694-4160. Mobile: 650-224-6644
To unsubscribe from the STDS-802-11-TGBD list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBD&A=1