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

Re: [STDS-802-11-TGBE] Regarding CR document DCN 1442

Hi Juseong,

 

Thank for your comments,

 

I do believe that these cases still covered by bullet (3) and paragraph on line 17. AND Note 1: NOTE 1—The backoff counters for each link count down as specified in 10.23.2.4 (Obtaining an EDCA TXOP). Which directly imply that you have more  that one BK that can be decremented simultaneously

 

The mechanism described in 35.3.16.6 does not link or interrelate EDCAF with each other. For example, STA may have AC_BE and AC_VI running in parallel on two links. AC_VI may countdown to zero on link 1 first and STA may decide (this is implementation specific) not to TX from it and wait for AC_VI on link 2 to countdown to zero. Later, AC_BE may countdown to zero on link 1 and with some luck AC_BE also countdown to zero on link 2. STA may decide at this moment to TX from AC_BE on both link 1 and 2. If so, Backoff counter AC_VI on link1 remain ==0 .

And it will remain 0 until STA on link 1 decide not to wait anymore or STA on link 1 detect medium transition from BUSY to IDLE. In both cases STA need to perform new backoff for AC_VI.

 

And STA will do that , as follows from NOTE1 for every EDCAF.

 

Also, please see below in bold

 

 

 

From: 문주성 (Juseong Moon) <jsmoon0211@xxxxxxxx>
Sent: Monday, September 12, 2022 12:19 PM
To: STDS-802-11-TGBE@xxxxxxxxxxxxxxxxx; Akhmetov, Dmitry <Dmitry.Akhmetov@xxxxxxxxx>; ronny1004@xxxxxxxxx
Subject: RE: Regarding CR document DCN 1442

 

Dear Dmitry Akhmetov,

 

Hello, I'm Juseong Moon from KNUT.

I want to further discuss on your resolution of CID 12414.

 

It is to cover the case of a STA with multiple EDCAFs whose backoff counter is 0. Your resolution is to reject the comment because the behavior of other EDCAFs are covered by bullet (3) and paragraph on line 17. However, bullet (3) and paragraph on line 17 is described in terms of STA so in my opinion, it doesn't cover the multiple EDCAFs with backoff counter 0 case.

 

(3) does not cover "other" EDCAFs becasue this case doen't fall into step (3). Because a STA has already transmitted a frame with selected one EDCAF, the STA is no longer "(3) A STA with backoff counter that has already reached zero and that choose not to transmit following condition 1b)."  

 

In the case of transmitting from the selected one EDCAFs among EDCAFs with backoff counter 0, there are two possible cases for the not selected EDCAFs: 1. keeping backoff counter 0 and 2. invoke new backoff counter.  

 

1. Keeping backoff counter 0
A paragraph on line 17 may be applied. However, "A STA with backoff counter that has already reached zero" may have more than two EDCAFs with backoff counter 0. Then a STA needs to choose one EDCAF. We need a description how to select one EDCAF if there are multiple EDCAFs with backoff couter 0.

Not sure what is the issue with that. Say, STA has AC_BE, AC_VI and AC_VO with BK=0 on link1. On link 2 AC_BE counted down to zero. STA on link 1 select ONE EDCAF (say – AC_VO) for transmission and MLD transmit on both links. STA on link 1 still has AC_BE and AC_VI with BK=0. STA may keep each at zero and keep waiting. Or it may perform new BK, this is STAs chose. It may decide to run new contention for AC_VI and keep AC_BE at zero. I do not see why we need to have a description how to select EDCAF. Moreover, as I mention in CR section – STA after transmission from a particular AC will detect transmission from BUSY to IDLE and will be forced to re-do BK for all ACs

 

2. Invoke new backoff counter
This might be the clearer solution. Since (3) does not cover other EDCAFs with backoff counter 0,  "CW[AC] and QSRC[AC] are left unchanged" can not be applied. We need a description what kind of EDCAFs parameters's value should be used.

 

If I am missing your point, please let me know.

 

Best Regards,
Juseong Moon

 

---------- 원본 메일 ----------
보낸사람: "문주성 (Juseong Moon)" <jsmoon0211@xxxxxxxx>
받는사람: Dmitry.akhmetov@xxxxxxxxx
참조: "ronnykim" <ronnykim@xxxxxxxx>, ronny1004@xxxxxxxxx
날짜: 2022-09-12 (월) 15:54:29
제목: Regarding CR document DCN 1442

 

Dear Dmitry Akhmetov,

 

Hello, I'm Juseong Moon from KNUT.

I have read your CR document DCN 1442 r1, on 35.3.16.6.

 

Three comments (12409, 12414, 12426) from me and my colleague are resolved by you.

 

-12409: Three NSTR link case

I tried to say the case where two link pairs with one common link. In this case, there might a case where three links may perform start time sync transmission, etc.

 

-12414: EDCAF selection

I have submitted to resolve this CID. If ACs, which are not selected, keep their counter at zero, they will definitely collide. If they invoke backoff procedure, what kind of EDCAF parameters(unchanged, or doubled) shall be used.

-12426: multi-TID A-MPDU construction

With this CID I am trying to say if the length of the selected AC is short and there is another AC with backoff counter 0, we may construct multi-TID (with two different ACs) A-MPDU.

 

Could you please take out these three comments from you resolution or exclude them from the CID list of straw poll?

 

Best Regards, 

Juseong.

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