You are on page 1of 15

Uplink Leg Switch Mid-Band

Uplink-Downlink Decoupling
FNI STC NW
SBS
• STC NW has deployed NSA NR with anchoring from LTE FDD layers (L1800, L2100,
L700). Due to weak coverage of UL NR STC customers complaint a lot about very low
throughput and UL leg was fixed to LTE to ensure acceptable UL throughput.

• Latest SW version NR 19Q4 brought the possibility to automatically switch leg (NR or
LTE) based on radio conditions and it is successfully tested in the live NW.

• This feature implements Uplink Leg Switching for EN-DC split bearers in 5G/NSA. It Switch from SCG to MCG
introduces detection of good/poor NR uplink quality using measurements made by cellgroup
the gNB.

• Based on NR uplink quality, operator configured thresholds and the system will PDCP
initiate UL cell group switch from SCG to MCG or from MCG to SCG.

• A configurable prohibit timer prevents switching back to NR to avoid ping-pong


between NR and LTE. LTE NR

Switch from MCG to SCG


cellgroup
•UL leg switch is functional working as expected

•Before activation it is important to understand the feature functionality, because it requires tuning based on the actual network
configuration.

•The MOM parameter range for the low quality threshold is not sufficient. Its expected to be revised in coming SW releases and
higher values would be allowed higher values.

•Optimum thresholds for the leg switching has been derived and checked during FNI.

•Thresholds strategy : SCG throughput >= MCG [our majority of test sample showed NR throughput higher than LTE, though
testing was done in medium radio conditions and few samples were collected from cell edge]

•endcUlNrLowQualThresh: 35 (SBS recommend to rollout this setting) ~ 11 SNIR Reported

•[Refer to next slide for details, how reported SINR is Normalized and used in algorithm]
•endcUlNrQualHyst: 6 (SBS recommendation)

•Ookla samples for latency are improved as SCG is in use for maximum of the time
•User measured UL throughput = 48 -> 77mbps ~ 60% Improvment (Mid cell range)
•Latency measured = 23 -> 17ms ~ 35% Improvement(much lower than LTE UL leg)
Assumption: UE max power 20 dBm UL SINR Max @ 1 RB
Interfence + Noise = -115 dBm/RB
=>20dBm-(-115 dBm/RB) = 135 dB @ used for UL leg switch
• UL Leg switching quality thresholds are checked
UL Pathloss 0.
UL Path loss (this fig.) includes all
gains and losses between UE Tx and against a standardized UL SINR Max.
gNB Rx. •
(normalized to the maximum achievable SINR; that which would be
measured if the UE were using only a single resource block)
• In order to understand the UL Leg switch decision thresholds, we need to
realize it’s based on the ulSinrMax (normalized to 1RB), i.e. green curve
in the figure. The UL SINR for a UE a this point depends on actual number
2 of RBs, power used by the UE, the path gains and losses and Interference
plus Noise over the scheduled RBs at the time .
• If the minimum requirement for acceptable NR performance is N
resource blocks with a SINR of X dB on each of those resource blocks,
1
then endcUlNrLowQualThresh should be set to X + 10 * log10(N).
• Example related to the FFI tested thresholds for 60 MHz:
For 60MHz the maximum number of RBs for PUSCH is 158.
10*log(158) = 22dB.
2
1 • 1. Switch from SCG to MCG triggered at endcUlNrLowQualThresh = 35dB
At this point the maximum achievable UL SINR on 158 RBs will be:
X_low = 35dB – 22dB = 13dB
• 2. Switch from MCG to SCG triggered at
endcUlNrLowQualThresh+endcUlNrQualHyst = 35+6 = 41dB
At this point the maximum achievable UL SINR on 158 RBs will be:
X_high = 41dB – 22dB = 19dB
• FFI tests (BB trace) logged measured UL SINR(@158 RBs) = ~16 dB
Uplink Leg Switching allows Uplink User plane to be
dynamically switched between NR and LTE based on NR
uplink SINR measurements
1. Resource initially configured to be used is for example
SCG – NR (based on MOM attribute initialUplinkConf)
2. When UL SINR falls below endcUlNrLowQualThresh, UL
is switched through the LTE leg
3. UL SINR becomes endcUlNrQualHyst / hysteresis better
than low quality threshold but cant switch until prohibit
timer expiry
4. Switch back from LTE to NR happens after UL SINR has
been hysteresis better than low quality threshold for a
prohibit timer.

Parameters: (Tried)

endcUlNrLowQualThresh (-2, 7, 25, 35) endcUlNrQualHyst (9,7,6)

endcUlNrRetProhibTimer (1000, 8000msec) endcUlLegSwitchEnabled (True/False)


Challenges Limitations Possibilities

Measure - Trigger NR to LTE OR LTE to NR leg switch No PM counters available to check We have to completely rely on field test and utilize
from stats some test-counters for live tests

Optimal threshold to trigger UL leg NR > LTE > NR Reported UL SNIR is normalized In absence of UL NR traffic, the noise floor will vary
internally and used by Algo; its big after shifting the NR traffic. Different values will be
challenge to find the optimal value tried to reach to final threshold

Ping-pongs NR->LTE->NR switch With default values, lot of Prohibit timer was extended to 8sec, later we will
switches were observed. bring it down step by step

Target UL throughput from subscriber perspective irrespective It need to be agreed with STC and Its open for discussion yet.
of UL LEG (NR or LTE) MS team to finalize the excepted
throughput based on cell load and
related strategy
 initialUplinkConf: MCG/SCG (both tried)
 NR BW 100MHz-TDD
 LTE BW 20 MHz - FDD
 UL Leg Switching-ON

 endcUlNrQualHyst =9
 endcUlNrLowQualThresh = -2 (default)
 endcDlNrRetProhibTimer (1s)
 Minimum time between an uplink leg switch from NR to
LTE and an uplink leg switch from LTE to NR

 UL PRB – MAX allowed


 LTE Anchor Network: UE Locked to LTE
700/1800/2100 MHz

From below observed frequent


switching, therefore further parameter
tuning carried on
 initialUplinkConf: MCG/SCG (both tried)
 NR BW 100MHz-TDD
 LTE BW 20 MHz - FDD
 UL Leg Switching-ON

 endcUlNrQualHyst =7
 endcUlNrLowQualThresh = 25
 endcDlNrRetProhibTimer (1s)
 Minimum time between an uplink leg switch from NR to
LTE and an uplink leg switch from LTE to NR

 UL PRB – MAX allowed


 LTE Anchor Network: UE Locked to LTE
700/1800/2100 MHz

Better than previous but not optimum


 initialUplinkConf: MCG/SCG (both tried)
 NR BW 100MHz-TDD
 LTE BW 20 MHz - FDD
 UL Leg Switching-ON

 endcUlNrQualHyst =6
 endcUlNrLowQualThresh = 35
 endcDlNrRetProhibTimer (8s)
 Minimum time between an uplink leg switch from NR to
LTE and an uplink leg switch from LTE to NR

 UL PRB – MAX allowed


 LTE Anchor Network: UE Locked to LTE
700/1800/2100 MHz

Stable leg
switching
• initialUplinkConf: MCG/SCG (both tried)
• NR BW 100MHz-TDD
• UL Leg Switching-ON
• endcUlNrLowQualThresh = -2,25, 35
• endcUlNrQualHyst =9,7,6
• UL PRB – MAX allowed
• LTE Anchor Network: UE Locked LTE 700/1800/2100 MHz FDD
20MHZ
• TEMS Investigation
• ZTE UE

numBearerModificationResponseSuccess The number of successful uplink leg switches.


numBearerModificationResponseFailure The number of failed uplink leg switches.
Average throughput increased from
48 -> 77mbps ~ 60% higher throughput

Scenario tested (Fixed):


• Pre: LegSwitch OFF, fixed UL LTE
• Post: LegSwitch ON, leg is automated

Scenario planned:
• Mobility
• NR leg fixed to 5G

Average latency improved from


23 -> 17ms ~ 35% Improvement
Parameters witch gave stable switching results are applied on a small cluster. Major KPIs including UL
throughput and latency are benchmarked

1. UL leg switch OFF - UL leg fixed on LTE (completed)


2. UL leg switch ON – UL leg automated (completed). Redrive will be done as one of the sites found
was down
3. UL leg switch OFF – UL leg fixed on NR (planned on Monday 2020-01-13

After customer confirmation, it the trial will be extended on large cluster for about 50 sites and it will
be monitored for some days.

Further tuning of endcUlNrLowQualThresh and endcUlNrRetProhibTimer will be done to ensure that


user perceived throughput isn’t compromised and NR is used maximum.
Thanks

You might also like