Yes in logs we can see 1TB is split in about 22 CBG.
Agreed. But in mostly simulated environment I donât see anybody using CBG based retransmission till now.
Itâs only TB.
So for now, I understand that there is no way to retransmit data from a single layer in MIMO.
It has to be whole TB(if we donât consider CBG).
But max CBG per TB can only have value 2,4,6 and 8. How come 22?
This is what I see in logs.
21, 22 or 23 CBGs.
Maybe reporting is per msec not per slot.
So you mean once initial data transmission is done with 2x2 MIMO, then all retransmissions will follow 2x2. Right?
(here actually i am not convinced)
Maybe not per TB.
Yea, maybe.
Do you mean rank might be different? Which would then mean a different TB size?
Nope rank is not change.
Only RV is changed.
And mcs will be 28 29 30 or 31.
This I agree too!
If this is true, then using TB based retransmission is worst idea.
Because RF can change during retransmission, and even any one CBG got decoding failure then whole TB need to re transmit again and again.
RF doesnât change in one slot or one msec.
Mcs 28-31 means retrans. For a DL retrans TB, mcs is reduced and rb increased, right?
What is intresting that CB has values like 23 24 although one line represents one slot.
This can still be valid meaning that there are max 8 CBG but CB can be 23 or 24.
I think CBG is not configured maybe.
We nee to check RRC for that.
Yes, this looks true.
If CBG is configured, I donât expect the whole TB to be retransmitted.
It should look like this in the log but I could not find it:
And I have one question about CBG: how the gnodeB is informed that one CBG is failing? At what layer: at phy layer in uplinkin PUSCH? Does anyone have an example of such a message?
Even I would want to know this.
I havenât really seen CBG being used until now in.our testing.
But am sure itâs harq level retransmissions.
I donât think so.
HARQ is MAC layer that so much higher.
It has to be phy layer i.e. layer1.
But isnât phy retransmission triggered by Harq.
Is there a different mechanism?