Sorry for the late response to the questions for 11-19/0951r0 because my e-mail hasn’t joined IEEE 802.11be list before and I did not receive the questions timely. Thanks for the questions of the experts. My answers to the questions are showed in the following. If you have any question please don’t hesitate to send me.
Q1. On slide #6, could you please elaborate on how you differentiate between RTA and non-RTA traffic? As you may know traffic class information is usually lost when a packet traverses thru a legacy routers along the way.
I think the identification of traffic flow, which corresponds to its traffic class, can be used to differentiate between RTA and non-RTA traffic in PHY & MAC.
Q2. On slide #7, the CAP period are set aside to aid RTA transmissions. However, your slide mentions that non-RTA traffic is allowed during these times along with non-CAP periods. Seems like non-RTA traffic will benefit in your scheme. Shouldn’t you disallow non-real-time traffic during CAP periods? Also how do you prevent OBSS STAs from occupying the medium during CAP periods?
Q3. The proposal focuses on real-time applications with high throughput and “worst case latency” requirements. Based on Slide 7, non-RTA traffic has channel access opportunities in CSMA TXOPs and CAPs. And, RTA traffic use scheduled access in CAPs. It is a bit unclear to me how the proposed solution is improving the worst case latency performance.
This contribution mainly focuses on the transmission of RTA data scheduled during the CAPs in different channels in order to guarantee the latency requirement as the contention-based channel access is best-effort depending on how busy with the channel. And if there is redundant TXOPs besides the TXOPs occupied for RTA transmission it is not excluded that the non-RTA data uses the redundant TXOPs.
This contribution doesn’t say that there is no OBSS STAs occupying the medium during CAP periods. It focuses on the flexible usage of multiple channels to reduce the inferences of OBBS STAs and to increase the robustness of the scheduled transmission of RTA data. For example, if there exists strong inference from OBSS STAs in one channel, the CAP periods of the other channels, which may suffer from no or weak inference, can be used for the transmission of RTA data. If both channels suffer from strong inference, the redundant transmission in different channels can be considered.
Q4. In the below figure, when this AP setup CAP in band 1, the OBSS will not able to setup RTA application anymore, because the duration of CAP may larger than the latency requirement of RTA. Am I understand correctly?
This figure just gives an example to show the scheduled TXOPs. I think whether the duration of CAP may be larger than the latency requirement of RTA depends on the scheduling algorithm.
Q5. When there are legacy OBSS in band 1, how AP can setup CAP?
Please see the above answer to Question 2 & 3.
Best Regards,
Liuming Lu