IMTAphy

MIESM

Asked by Chiu

Dear Jan,

I noticed that you used the MMIB functions from 16m to implement the MIESM of PHY abstraction, but with a difference that you still mapped back to effective SINR instead of directly maping from MMIB values to BLERs as suggested in EMD of 16m.

And you mentioned that you use the Vienna's link level simulator to obtain the BLER-SNR curves.

I have two questions regarding PHY abstraction:

1) Why some of the BLER-SNR curves, which I plot from src\link2System\linkLevelResults, look strange. I mean, in some of the curves, BLER increases with SINR (looks like dent), though very few and typically in high SNR regime. Does that mean 5000 subframes are not enough?

2) In your code, you use three offsets: sinrOffsetQPSK, sinrOffsetQAM16, and sinrOffsetQAM64 to directly offset the BLER-SNR curves. And they all equal to 0.2 dB. I don't understand why we need an offset here, and especially same offset to all curves. Are you suggesting that the results from Vienna would have a 0.2 dB shift w.r.t. 3GPP's result?

Thanks in advance for your help.

BRs,

Sung-En Chiu

Question information

Language:
English Edit question
Status:
Solved
For:
IMTAphy Edit question
Assignee:
No assignee Edit question
Solved by:
Chiu
Solved:
Last query:
Last reply:
Revision history for this message
Jan (jan-ellenbeck) said : 		#1

Hi Chiu,

Yes, that's what I did. Regarding your questions:

1) I guess you are seeing this for small BLERs below 10^-2 or so, right? In fact it is from running "only" 5000 trials per SINR value. You run the link-level simulation 5000 times per SINR step, count the number of ACKs and NACKs and get your BLER. For BLERs around 10^-3 this is just a handful of NACKs so that there can be quite a variation between independent SINR steps.

2) Yes, I have this option but actually this is not very necessary. I had introduced it while I was calibrating because for some time I suspected that the mapping curves might be a little bit off. But in the end I found them to be quite OK as they are but it seems I left that offset at 0.2 dB because it looked a tiny bit better than without it. There is no single "correct" mapping curve and also no single "correct" reference curve.

Best regards
Jan

Revision history for this message
Chiu (qoocoolqoo) said : 		#2

Dear Jan,

1) Yes, exactly.

2) I see. So it would be fine with or without the offset.

Thank you!

BRs,

Sung-En Chiu