Hello 5G Experts,
In 5G NSA Option 3X, the data split occurs at the gNB. However, how does the gNB determine the proportion of data that should be carried by 4G and 5G?
Thank you!
2 Likes
Data Split Mechanism in 5G NSA Option 3X
In Option 3X of 5G NSA, the data split occurs at the gNB. To determine how much data should be carried by 4G (LTE) and 5G (NR), it’s essential to consider that in Option 3X, the Secondary Node (SN) controls the termination of the User Plane (UP) data flow, particularly in SCG Split Bearers.
A Split Bearer refers to a bearer where traffic can be routed via either the LTE (eNB) node or the NR (gNB) node. Split Bearers are classified into two types:
- MCG Split – Traffic is split at the Master Node (MN) (usually the LTE eNB).
- SCG Split – Traffic is split at the Secondary Node (SN) (typically the NR gNB).
This configuration allows dynamic load balancing and optimized data flow management between LTE and 5G.
1 Like
Thank you!
In the case of SCG Split Bearers, how is the data amount determined and split between 4G and 5G? What parameters influence this decision?
The gNodeB dynamically adjusts the data split between 4G and 5G to optimize the overall user experience.
This ensures efficient utilization of both 4G and 5G resources, providing users with the best possible performance for their applications.
The split is determined based on radio conditions for each technology, allowing for adaptive and optimized data distribution.
1 Like
Check Flow Control Algorithms in Your gNB Configuration
The flow control algorithm monitors the amount of data sent over a given interface. If the estimated delay exceeds the target, PDCP PDUs are no longer transmitted over that interface.
The data split algorithm determines which interface will be used for transmitting the current PDU, selecting the path with the least latency to optimize performance.
The downlink (DL) data split and flow control operations can function in the following modes:
- Complete downlink user data transfer via the F1-U interface
- Complete downlink user data transfer via the X2-U interface
- Downlink user data transfer with data split over both X2-U and F1-U interfaces
This mechanism ensures efficient data distribution between 4G and 5G networks, improving overall system performance.
There are various flow control methods and algorithms, including the 3GPP-recommended Downlink Data Delivery Status (DDDS) mechanism.
For multivendor interoperability, 3GPP TS 36.425 defines the Downlink Data Delivery Status (DDDS), which enables feedback from the Secondary eNB (SeNB) to the Master eNB (MeNB). This feedback allows the MeNB to regulate the downlink user data flow through the SeNB for the corresponding E-RAB, ensuring efficient and controlled data delivery across the network.
1 Like
There are separate parameters governing the data split process:
- Downlink (DL): The split is determined based on data volume, network load, and other parameters to ensure efficient resource utilization.
- Uplink (UL): The split is controlled by the ulDataSplitThreshold parameter, which defines the threshold for distributing uplink data between 4G and 5G.
These parameters help optimize traffic flow, balancing performance and network efficiency.
1 Like
In a split bearer, the user plane data flows through PDCP and is distributed to:
- NR RLC via the F1 interface
- LTE RLC via the X2 interface
Your question is essentially: How does NR PDCP become aware of the data scheduled in LTE?
This is managed through DDDS (Downlink Data Delivery Status) from LTE RLC. The highest PDCP SN (Sequence Number) delivered by LTE RLC is reported over a periodic timeframe, allowing NR PDCP to track data delivery.
How does PDCP estimate the amount of data for LTE RLC and NR RLC?
There are multiple algorithms for this, depending on the desired optimization:
- Load Balancing: The system can estimate buffer occupancy on both LTE and NR paths to distribute data evenly.
- Throughput Maximization: The highest delivered SN number can be used as a weighting factor to allocate data, ensuring that throughput is maximized relative to the DDDS reporting interval.
The choice of algorithm depends on whether the goal is to balance network load or maximize throughput efficiency.
1 Like
Thanks a lot for your response!
To clarify further - if a 5G NSA UE is connected to two radio links (4G and 5G) under an SCG configuration and needs to download 100 MB of data, how does the gNB determine the amount or ratio of data to be transmitted via eNB (4G) and gNB (5G)?
Is this ratio configurable?
Is it based on RF conditions or other parameters?
1 Like
The parameter may vary between vendors and can either be hardcoded or configurable, depending on the vendor’s implementation. If I recall correctly, Nokia also relies on DDDS for such cases. Let me illustrate this with an example to explain the concept of how this could be achieved.
Assume the DDDS period is 40 ms, and NR reports around 385 PDCP PDUs successfully delivered in sequence to the UE (based on the highest PDCP PDU Sequence Number). In contrast, LTE delivers around 192 PDUs in the same timeframe. For simplicity, let’s assume all packets are of the same size, 1300 bytes.
If packets are sent only in NR, using this assumption, the achievable throughput would be approximately 100 Mbps (based on the 385 delivered PDCP PDUs). However, if the implementation supports a split of 100/50, the throughput could reach around 150 Mbps. The same logic applies in reverse, depending on how the traffic is split.
In essence, throughput can be estimated by normalizing the delivered PDCP SNs over the DDDS periodicity, providing a basis for performance estimation.
1 Like
Thank you so much!
Everything is clear now, and I truly appreciate your help. 
1 Like
PDCP-Config parameters, such as primaryPath and ul-DataSplitThreshold, configured by RRC, play a crucial role in scheduling and distributing data across NR and LTE legs.
This process also considers DDDS feedback from the NR RLC and LTE RLC layers.