Issue with RTT measurement

[soora-jp]

Hello.

I'm measuring the end-to-end round trip time (RTT) between UE and MEC host in MEC environment. Most UEs' RTTs are between 13-14ms, but specific UEs' RTTs are between 19-20ms. It looks like it takes 2-times transmissions to send the message because of not enough sduSize. I have no idea why this happens. Could you give me some insights about it?

Application specification

• the size of request and response messages is fixed with 100Byte
• intervals between sending messages are random within 1s
• transport layer is TCP

Thank you and Best regards.

[giovanninardini]

Hello,

I guess the reason might be the channel quality (CQI) experienced by those UEs. When a UE asks for an UL scheduling grant to transmit data, the gNB first grants ONE Resource Block, to let the UE sends its Buffer Status Report. That RB can also be used to start transmit data, though. If the data you want to transmit is small enough to fit that one RB, then all the data is transmitted in "one shot". Otherwise, if the data is too large, or the CQI is low (i.e., fewer bits can be allocated in one RB), then the UE will need to wait for at least another scheduling grant from the gNB, hence the increased delay.

In your scenario, maybe the UEs with higher RTT are located far away from the gNB and have lower CQI than the other UEs.

Hope this helps.

Best regards. Giovanni

[soora-jp]

Thank you for your response.

As you said, averageCqiUl of UEs with higher RTT were 14 while averageCqiUl of UEs with lower RTT were 15. However, in my scenario, all UEs are in exactly the same position. I tried to find the reason for the low CQI by finding where the CQI was calculated, but I couldn't.

So, I would like to know the following

1. Where is the CQI calculated?
2. What caused the low CQI other than the UEs’ position?

Best regards. Sora

[giovanninardini]

CQI is calculated by the function requestFeedback(), in classes LtePhyEnb and LtePhyEnbD2D. One possible cause for different CQIs for UEs in the same position may be due to, for example, fading and/or shadowing effects (if enabled in the channel model).

[Ali-K-Abbas]

@soora-jp Dear, Good day Can You help me with how you exactly measuring the end-to-end round trip time (RTT) between UE and MEC host in MEC environment?

[soora-jp]

@giovanninardini Hello. I sincerely apologize for not replying. I will learn wireless communication and 5G protocols from scratch again. Thank you.

[soora-jp]

@Ali-K-Abbas Hello. I’m sorry for the late response.

When I started my current work, there was no MEC application to measure e2e RTT, so I implemented it myself. However, in version 1.2.1 of simu5g, there is mecRequestResponseApp, and it looks enabled to measure e2e RTT. So I recommend using it. For reference, I used tag function of INET to embed request time in packets.