Hi Behnam,
Thank you for the presentation of 11-25/0110r4. Sorry I did not have a chance to ask questions during the meeting.
Some questions and comments as follows:
1.
Slide 4, is it assumed that the issue of long tail latency is caused by lower RSSI received by the AP in the field ?
BD: That
is correct. Discrepancies in RSSI values may be caused by factors such as obstacles and distance, which in turn affect the channel access probability of STAs and reduce the effectiveness of P-EDCA for STAs with weaker RSSI..
2.
Slide 5, Agree with your analysis. I think we should consider multiple ways to resolve the fairness concerns
a.
if STAa and STAb have used P-EDCA and acquired TXOP, then the P-EDCA mechanism should suspend STAa and STAb to continue using P-EDCA so as to provide the fairness to other
STAs. This can provide a chance for STAc with lower RSSI to acquire TXOP in the next run. We need to define a fairness rule for the STA that has used P-EDCA for TXOP acquisition.
b.
In addition, if we consider “defer time due based condition” for the STA with low RSSI to use P-EDCA, it could prevent a STA with high RSSI from acquiring a TXOP first all
the time, like in the slide 13, the STA closer to AP has higher chance to acquire TXOP.
BD: Yes,
the goal is to allow all STAs to benefit from P-EDCA regardless of their RSSI. This can be achieved through various means, such as fine-tuning the P-EDCA parameters, introducing a delay after the last usage of P-EDCA, and so on.
3.
Slide 9, do you suggest to limit STAd to transmit RTS within 3 slots instead of 7 defined in D0.3?
BD: This
slide does not propose any new contention window sizes. Instead, it aims to demonstrate that when a STA with a poorer RSSI value becomes eligible to use P-EDCA before other STAs, it gains priority access to the channel. Even if its DS frame collides with RTS
frames from other STAs, it still has an advantage: while the other STAs must wait for an ACK/CTS timeout, the STA sending the DS frame can contend for the channel again immediately after the AIFS. In other words, we are leveraging P-EDCA to solve the unfair
channel access problem in time domain instead of using other complicated mechanisms such as transmission power control.
4.
Slide 11, as a STA does not know whether other STAs would use P-EDCA, how does it decide whether to use P-EDCA for LL transmissions? For example, 5 STAs in RSSI range (-70
to -79). Also how does the AP control the number of STAs using P-EDCA not exceed the limit?
BD: If
the method presented in Slide 11 is used, the assumption is that the AP is keeping track of the number of STAs intending to use P-EDCA, as specified in the latest draft. If each LL STA is aware of the number and RSSI of other LL STAs, it can then select P-EDCA
parameters that enable fair competition among all LL STAs. For both methods presented in Slides 10 and 11, SCS updates (or profiles) can be used to announce new distribution values or P-EDCA parameters to maintain fairness over time.
Other methods, such as reporting to the AP the duration of channel access and number of DS transmissions, can be used to further
enhance the proposed mechanism.
Controlling the number of STAs that are allowed to use P-EDCA is beyond the scope of this work. However, regardless of how many STAs use P-EDCA, our goal is
to ensure that all of them receive a fair share of channel access.
Best Regards
Yonggang