5G Harq Retransmissions

5G NR
#1

Question about 5G harq retransmissions.
What is the average number you see in logs?
For example I see minimum 3 and maximum 16 (16 when TB failed to be sent and retransmission occurs at RLC layer).
Distribution is like in attached picture for a log.

image

I think there are a lot of retransmissions and wonder why MCS is so badly chosen.
I thought BLER is 10% for PDSCH/
Anyone facing same issue while analyzing 5G logs?
In the graph is presented “max decoding iterations” ie harq retransmissions (up to 16 retrans per each HARQ process)/

#2

Overlapping may be high on this site from neighbouring sites.

#3

I think it’s not this…
In all logs I checked iterations is at least 3 and max 16.
There are no values like 2 or 1.
But average value looks more like 7-8 iterations, as in the graph above.
And code blocks I see max 24, in DL.
I am amazed but those numbers, as I always thought that IBLER is 10% for PDSCH and HARQ retransmissions is max 1-2 retransmissions, not 7-8 like in the graph.

#4

Which message are you looking in 5G logs to get this?

#5

NR 5G MAC PDSCH STATUS.
Check last IEs, Max decoding Iterations and Total CB Number.

#6

I checked in a 5G SA log file I can see 1, 2, 3 and up to 4 max ReTX.

#7

So you see no values like 16?
Maybe you have 4 max retransmissions for HARQ.
This is a Huawei log, where 16 retransmissions are possible.
Check few tens of them and let me know which is the highest number you found.
Those are HARQ retransmissions not RLC retransmissions.

#8

Checked in NSA log file now, 5 is MAX.

#9

Yes, we can go upto 8 max.

#10

To get a wider view I might need to check the whole db, currently this metric is not loaded.
Both SA and NSA is Nokia.

Is you log for sub-6 NSA?

#11

Yes, it is n78.