If HARQ fail, how is NACK sent for different MIMO streams of data?

CRC failure for CB, for CBG or for TB?
Retransmissions possible at CBG level only, not at CB level.

At CBG level.
Even if one CB fails, it would be CBG retransmission.

I also understand the same.
Yet I am not clear if this CBG CRC is checked at MAC layer or at phy layer.
It has to be at phy layer in my opinion but I find no evidence for this.

I think it would be at PHY layer.

Crc is always as phy layer, right?

I am confused about that excel file that I shared.
There is CB information inside yet the entire TB is retransmitted.
CRC for TB is at MAC layer.
CRC for CBG we believe is at PHY layer but need to double check this.

CBs will always be there.
But whether retransmission to perform at CBG level or TB level is configured.
Isn’t it?

I think you are right here.

This is a good article:
Transport Block - an overview | ScienceDirect Topics).

I think these segments are nothing but code blocks.
So CBs are always there.
Whether you want to base your retransmissions based on CBGs is the optional part.

Yes, right. :+1:

I think the concept of code blocks is pretty clear now.

Yes, it seems usage of CB is mandatory while usage of CBG is optional.

From NR logs no info for CBG. Seems not used as of now.

For HARQ, retransmission is performed through redundancy versions of the codeword. Since HARQ is adaptive, the MCS and number of layers can change for every retransmission.

To add, once TB is sent from MAC to PHY for initial transmission, it does not change. TB is used to generate a codeword with mother code rate (1/3 for LTE) the rate matching is used to generate the actual transmitted bits (taking into consideration mcs and layers in each (re)transmission).

1 Like