Thread Links | Date Links | ||||
---|---|---|---|---|---|
Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Hi Mengshi, Thank you for the response.
My point is very simple: before we talk about
estimation of CSI variation, we need to know what CSI variation is, i.e., how it is defined. Otherwise, how can we tell if the rule you proposed is reasonable or not.
This logic is no difference from channel estimation: we first define the CSI is in frequency domain. Then, we can talk about the rule for CSI estimation and its feedback. Agree? Best regards, Ray From: humengshi <humengshi@xxxxxxxxxx>
Hi Rui,
Thanks for your comments. Actually I’m a little bit confused about the latest revised one. In the motion text, you can see the first sentence is saying
“the CSI variation is implementation specific”, but in the subbullets you are saying
“If the CSI variation function should be specified is TBD”. These two seem to be contradictory. The main intention of this motion is to ask
“do you agree that the estimation of CSI variation is implementation specific, but some rules shall be followed”, instead of adding a note saying the CSI estimation (having a function or implementation specific)
is TBD. Like I said before, the CSI variation is suggested to be implementation specific according to most people’s opinions obtained from many offline and online discussions, and it also can be seen from the SP result (Most people
agree with it). If I change it into “whether or not to define a function for CSI estimation”, it is not aligned with the SP’s intention. At this stage, I’d
like to run the implementation specific one. Hope you don’t mind. Hi Tony,
I would like to update the 1st motion text corresponding to SP 1 in contribution 21/1364r3. Motion 1 Move to add the following to 11bf SFD:
Corresponding document 21/1364r3, SP results: 14Yes, 5 No, 6 Abstain Best regards, Mengshi Hu 发件人: Rui
Yang <Rui.Yang@xxxxxxxxxxxxxxxx> Hi Mengshi, Thank you for the suggestion.
The sentence in the Note is somewhat confusing (CSI variation corresponds to 0 or 1?). However, this sentence and your mention of correlation coefficient give an idea
how to make the motion text more reasonable. Here is a suggestion for the motion text:
Motion 1 Move to add the following to 11bf SFD:
I believe the text above makes the mapping to [0, 1] more meaningful.
Again, without clear definition of the CSI variation function, I am still concerned about the consistency of the feedback from different devices – e.g., given the same
movement in the environment and the same threshold, different devices may feedback different results (below or above a given threshold). In the previous meeting, you mentioned that this may be managed by higher layer and using different threshold for different
devices. Even if this is the case, I think it could lead to inconsistent performance using those devices for the same sensing application in different networks which may also be implementation dependent. Anyway, I suggest the group discuss this further.
Best regards, Ray From: humengshi <humengshi@xxxxxxxxxx>
Hi Rui, Thanks for sharing your thoughts.
To some extent I agree with you that defining some formulas to make the CSI variation clear has its benefits, but also there are some drawbacks (See 21/351,
21/1069 for details, especially in Slide 7 of 21/1069, I listed Pros and Cons there). After the discussions, most people seem to prefer the implementation specific way.
How about I put a “TBD” in the motion text shown below? TBD indicates that the degree 0 and 1 may be implementation specific or determined by some formulas, and it can be discussed later. For example,
degree 0 indicates the two CSIs are almost the same, and degree 1 indicates that the correlation coefficient of the two CSI vectors indicates orthogonal. Your further thoughts on degree 0 and 1 are also welcome.
Motion 1 Move to add the following to 11bf SFD:
Best regards, Mengshi Hu 发件人:
stds-802-11-tgbf@xxxxxxxx <stds-802-11-tgbf@xxxxxxxx>
代表 Rui Yang Hi Mengshi, Thank you for the detailed clarifications. Yes, my first question was about the definition of CSI variation.
My concern is that, if the definition of “CSI variation” is implementation specific, the meaning of “degree difference” and mapping to the internal [0, 1] will all
be implementation specific. This could make the threshold-based method less useful, especially when we need to consider the interoperability of devices with different implementations. In addition, without clear definition of CSI variation, I don’t know what “the largest (or the smallest) degree of the CSI variation” is. To make the threshold-based
method meaningful, I believe it should be the theoretically largest value, rather than some value based on measurements. Otherwise, the feedback about higher or lower a threshold will not be consistent over time for each device. To deal with this issue, the
definition and/or the characteristics of CSI variation may need to be specified.
Best regards, Ray From: humengshi <humengshi@xxxxxxxxxx>
Hi Rui, Thanks for your comments. Please see the replies shown here: Q1:
Do you mean “the definition
and the estimation of CSI are implementation specific”? A1: I think here what you wanted to ask is the
“CSI variation”, instead of the CSI, because we do not touch the definition and estimation method of CSI. The CSI is just the traditional one. If your question is on the CSI variation, my answer is: 1. The
definition of CSI variation is not implementation specific (See A2 below). 2. The estimation or calculation of the CSI variation is implementation specific, but the final estimation result shall be in a closed interval [0,1]. This will provide benefits for
the metric unification and the threshold adaptation. Q2:
What is the definition of “CSI variation”? From SFD, I see “The difference between the current measured CSI and the previous measured CSI is quantified. The difference is referred
to as CSI variation.” Is this the definition? If so, what does “difference” mean here? A2: This is not the exact definition. According to some discussions in the call, people think more details can be added to the above sentence. For example, one
type of the CSI variation can be: the degree of the difference between the current measured CSI and the previous latest measured CSI. Another type can be: the degree of the difference between the current measured CSI and a marked CSI or the latest reported
CSI, etc. Maybe the initiator can choose which CSI variation the responder should report. This can be further discussed. Regarding the word
“difference”, I agree with you that this is implementation specific. If you think the definition for CSI variation shall clarify how to calculate the degree of the difference, my thought is that you can say
it is implementation specific. Q3: Is “CSI” defined in frequency domain or in time domain? A3: Frequency domain. Your further comments and questions are welcome. In addition, according to some discussions (people think it is a little bit confusing to say
“indicates a minimum CSI variation”
and “indicates a maximum CSI variation”),
I update the 1st motion text (corresponding to the SP 1 in 1364/r3) to make it clear: Motion 1 Move to add the following to 11bf SFD:
Best regards, Mengshi Hu 发件人: Rui
Yang <Rui.Yang@xxxxxxxxxxxxxxxx>
Hi Mengshi, Thank you for posting the motion texts.
I have a few questions about Motion 1: Do you mean “the definition
and the estimation of CSI are implementation specific”? If not, what is the definition of “CSI variation”? From SFD, I see “The difference between the current measured CSI and the previous measured CSI is quantified. The difference is referred to as
CSI variation.” Is this the definition? If so, what does “difference” mean here? And, is “CSI” defined in frequency domain or in time domain?
Best regards, Ray From: humengshi <humengshi@xxxxxxxxxx>
Hi Tony, I also would like to submit 2 motion requests corresponding to the contribution 21/1364r3. Motion 1 Move to add the following to 11bf SFD:
Corresponding document 21/1364r3, SP results: 14Yes, 5 No, 6 Abstain Motion 2 Move to add the following to 11bf SFD:
Corresponding document 21/1364r3, SP results: 16 Yes, 1 No, 6 Abstain Best wishes, Mengshi Hu 发件人:
stds-802-11-tgbf@xxxxxxxx <stds-802-11-tgbf@xxxxxxxx>
代表 durui (D) Hi Tony, I would like to submit 2 motion requests corresponding to the contributions 21/0876r3 and 21/1288r2. Motion 1 Move to adopt the document (21/0876r3) as the official Evaluation Methodology and Simulation Scenarios document for IEEE 802.11 bf ? Simulation is not mandatory for any contributions. Corresponding document 21/0876r3, SP results: 20 Yes, 0 No, 6 Abstain Motion 2
l
Move to adopt Truncated Channel Impulse Response(TCIR) described as follows as one optional type of the sensing measurement results ?
Ø
Calculating the CIR (time domain) from CSI/CFR (frequency domain) through IFT(usually, IFFT) .
Ø
Reporting the subset of complex samples corresponding to the range of interest of the entire CIR . Note: the size of the subset is TBD. Corresponding document 21/1288r2, SP results: 24 Yes, 6 No, 16 Abstain Best wishes, Rui Du 发件人:
Hanxiao (Tony, WT Lab) [mailto:tony.hanxiao@xxxxxxxxxx]
Dear all, This is a reminder and
call for submissions for the upcoming
meeting
on
September 28 (Tuesday),
10:00am – 12:00pm ET. Please note:
•
DCN, Title, Author (affiliation), Time duration
a.
If there is not enough time during the meeting, the presentation will be delayed to the next meeting.
b.
If there is enough time during the meeting, the presentation will be moved to the end of the queue and be presented in the end. However, any related SP should
be delayed to the next meeting.
Meeting number: 234 154 47759
Meeting password: wireless (94735377 from phones and video systems)
Join by phone: Tap to call in from a mobile device (attendees only)
+1-408-418-9388 USA Toll Access code: 234 154 47759
Join from a video system or application Dial
23415447759@xxxxxxxxxxxxxxxx
You can also dial 173.243.2.68 and enter your meeting number. Join using Microsoft Lync or Microsoft Skype for Business Dial
23415447759.ieeesa@xxxxxxxxxxxxxx
Need help? Go to
http://help.webex.com Best Regards : ) Tony Xiao
Han To unsubscribe from the STDS-802-11-TGBF list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBF&A=1
To unsubscribe from the STDS-802-11-TGBF list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBF&A=1
To unsubscribe from the STDS-802-11-TGBF list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBF&A=1
To unsubscribe from the STDS-802-11-TGBF list, click the following link:
https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBF&A=1
To unsubscribe from the STDS-802-11-TGBF list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-11-TGBF&A=1 |