What is the criteria for allocating 5G NR PDCP entity?

Hello 5G Experts,

For NSA n/w, I have an LTE anchor site MCG in 2.3 GHz band and 5G NR mmwave SCG in the 38Ghz. band.
In the rrc reconfiguration, I can see the PDCP entity allocated in 5G but there is no 5G mmwave coverage at the Cell edge as expected and no B1 event for SCG addition in Phy layer,.
So the Flow goes like 5G PDCP>4G RLC-MAC> 4G PDSCH

And in the cell center I have good 5G mmwave coverage and so the flow does like


My question is, what is the criteria for allocating 5G NR PDCP entity ?
What I observe is that PDCP entity allocation does not have a dependency on 5G coverage presence. Is it a default 5G PDCP allocation for all LTE anchor sites with X2 to gNB ?

Hi without 5G addition at cell edge you mean split bearer there…
Sorry, not able to understand completely.

Depending on vendor there are special anchoring LTE carriers that can be paired with 5G.
Also the amount of data that flows via 5G or via 4G depends on many parameters like buffer in data, delay, etc.
Is not related to coverage.

We have flow control option like DL over F1u X2 u or F1 u only.

Yes, after 5G addition, it can be Split bearer (5G + 4G PDSCH )or Direct bearer with 5G PDSCH alone.
Any case.

SCG addition is not just at PHY layer.
It is also at 5G RLC/MAC layer.
That is because because 5G PDCP data is split between 4G RLC and 5G RLC.

The UE is in light green area, where no mmWave coverage is present.


It will be RLF in 5G at that point or event A2 that will remove SCG.

So this pairing criteria you mentioned is it blind from the RLC MAC towards the PHY?
I.e. it just pairs and allots PDCP in 5G without considering the lower layer (RLC/MAC/PHY) availability in 5G?

It is parameters dependant: amount of data in the buffer, delay, thresholds, etc.
There are many parameters to decide how much data goes via 5G and how much data goes via 4G.

But in your case when UE is in green area all data will go via 4G because It will be RLF in 5G at that point or event A2 that will remove SCG.

So does this flow control is applicable to which all layer?
Is it from MAC towards PHY?

Yes that’s known, but assume my UE is stationary in this area from switch on, I still gets 5G PDCP and other lower layers in LTE.
So this 5G PDCP allocation criteria is what I’m trying to figure out here.

Yes from top to bottom.

That means the 5G PDCP allocation is also based on flow control criteria?

As per understanding, split happen at PDCP layer then on which route data will flow, i.e. via 4G or 5G or both depend on flow control in data over f1u and X2 u.
You will see flow like:
So for 4G+5G combined throughput check at 5G PDCP layer.
5G only throughput at 5G MAC layer.

It’s SN bit transfer only, i.e. bearer configured with NR PDCP.
But actually it’s not split bearer.
Even I seen through counter leading to confusion that 5G is activated.

Ok, So for now we don’t have a proper conclusion on the PDCP entity allocation criteria in non-5G coverage area served by an LTE anchor.

Depending on the vendor, UEs which report NSA capability, wi automatically be configured with PDCP with a 5G VERSION of the protocol regardless if PSCELL is configured or not, and will be using 5G PDCP as long as it is connected to a NSA cell.

I understand Nokia and Huawei works like this, Ericsson will require a LTE intra-cell handover to add the PSCELL, the change versions.

Is this what you curious about?


Absolutely! Great Help, Appreciate the response :+1:

I hope by NSA Cell you meant the LTE anchor cell in NSA?

I am having doubt.
How UE is using 5G PDCP if it is not having 5G Coverage means 5G coverage is below A2 threshold or completely no coverage zone.
What we see in NSA till now, if NSA UE goes out of EN-DC coverage, PDCP is shifted from 5G to 4G and flow will be 4G PDCP >>> 4GRLC >>> 4GMAC >>> PHY.

Are you sure about 5G PDCP entity for stationary Users where no 5G coverage?
It may possible some delay in PDCP entity transfer on cell edge, but you should not have persistent scenario like 5G PDCP + 4G (RLC to PHY) in no 5G zone.

Could you please double confirm.

1 Like

We have seen this scenario in initial testing itself.