If we increase the number of UEs supported per TTI in 5G from 16 or 32 UEs/TTI (to let’s say 64 or 128 UEs/TTI), then how will it impact the L2 or upper layer processing?
What could be the bottleneck?
Basically If I change UEs/TTI from 8 to 64 given a fixed resource allocation.
What my question is If I schedule 8 UE/slot vs 64 UEs/slot or 128 UEs/slot, what becomes the bottleneck for L2 and upper layers.
Is it the storage, processing?
Can anyone with L2/L3 expertise comment?
In my view no of PRB can be bottleneck.
From processing side there is no as such bottleneck seen from hardware side used.
If we have proper ecpri link between Base Band and Radio module.
I think PDCCH capacity is can be bottleneck.
If UEs are in poor radio condition, then due to high AL requirement, we cannot schedule as many user in 1 TTI.
PRB is also possible, but in NR with very high BW less chances.
If I have 8 UEs I schedule 32-35 PRBs per UE where if I have 64 UEs.
Then I schedule 4-5 PRBs per UE.
Given the full 100 MHz bandwidth, I can put 64 DCIs with agregation level 1-2, given that the channel is very good.
Think it’s L2 processing.
Higher UEs per slot is impacted by HARQ feedback processing, etc.
So in my view higher UEs per slot is L2 processing capacity.
Not an L2 expert, but came across this request in one of my projects.
From development perspective, tracking all HARQ threads in gNB L2 is extensive processing task. Higher UEs need higher buffering, processing, scheduling, etc.
This means, due to the aggregation level of the PDCCH components (Control-Channel-Element) the UEs number reduced in the TTI, typically when the user in bad radio condition, and due to the critical nature of the PDCCH (very very critical that the user gets it correct) the coding rate increases, which increases the resources consumed to transfer the PDCCH, that increases the level of aggregation (since each CCE contains a fixed amount of PRB [REG] - Take a look at the table below).
So, to recap, when the user in bad condition that site tends to increase the coding rate (since the modulation is fixed in this channel to QPSK) of the sent information to make it more immune to errors, this typically means the useful data in these PRBs is less, that requires the site to aggregate more CCE to send these data, which means in a fixed format framing nature of the mobile NW, the number of user is reduced (per TTI).
I hope that make sense to you.
You can get more theoretical / practical information from this website:
It s dynamic, and max agg level you can use is restricted per slot by 3gpp.
There are two limitations for UE side as 3GPP 38.213 indicates: PDCCH parameters can be adjusted / optimized based on field traffic as long as not exceeding above two tables.