RA57331-V-23R1 5G Radio

Optimization Part 2 Nokia internal

version 23R1

Latency

1 © 2023 Nokia

1
Copyright and confidentiality

The contents of this document are proprietary Such Feedback may be used in Nokia products and purpose, are made in relation to the accuracy,
and confidential property of Nokia. This document related specifications or other documentation. reliability or contents of this document. NOKIA
is provided subject to confidentiality obligations of Accordingly, if the user of this document gives SHALL NOT BE RESPONSIBLE IN ANY EVENT FOR
the applicable agreement(s). Nokia Feedback on the contents of this document, ERRORS IN THIS DOCUMENT or for any loss of data
Nokia may freely use, disclose, reproduce, license, or income or any special, incidental, consequential,
This document is intended for use by Nokia’s distribute and otherwise commercialize the indirect or direct damages howsoever caused, that
customers and collaborators only for the purpose feedback in any Nokia product, technology, service, might arise from the use of this document or any
for which this document is submitted by Nokia. No specification or other documentation. contents of this document.
part of this document may be reproduced or made
available to the public or to any third party in any Nokia operates a policy of ongoing development. This document and the product(s) it describes
form or means without the prior written permission Nokia reserves the right to make changes and are protected by copyright according to the
of Nokia. This document is to be used by properly improvements to any of the products and/or applicable laws.
trained professional personnel. Any use of the services described in this document or withdraw
contents in this document is limited strictly to the this document at any time without prior notice. Nokia is a registered trademark of Nokia
use(s) specifically created in the applicable Corporation. Other product and company names
agreement(s) under which the document is The contents of this document are provided mentioned herein may be trademarks or trade
submitted. The user of this document may "as is". Except as required by applicable law, no names of their respective owners.
voluntarily provide suggestions, comments or warranties of any kind, either express or implied,
other feedback to Nokia in respect of the including, but not limited to, the implied warranties
contents of this document ("Feedback"). of merchantability and fitness for a particular

2 © 2023 Nokia

2
Revision history and metadata
Please delete this slide if document is uncontrolled
Document ID: DXXXXXXXXX
Document Location:
Organization:
Version Description of charges Date Author Owner Status Reviewed by Reviewed Approver Approval
date date

DD-MM-YYYY DD-MM-YYYY DD-MM-YYYY

3 © 2023 Nokia

3
Module Objective
After completing this learning element, the participant should be able to:

Describe User Plane Latency Performance Analysis

Identify User Plane Latency Features
Report User Plane Latency Optimization Procedure
Record Control Plane Latency Performance Analysis
Recognize Control Plane Latency Features and Optimization Procedure
Express Quiz/Exercise

4 © 2023 Nokia

4
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

5 © 2023 Nokia

5
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

6 © 2023 Nokia

6
User Plane Latency Performance Analysis
Introduction
• Low latency is a key improvement of 5G over LTE
• It improves end-to-end performance and enables new applications
• User plane (UP) latency is the time it takes to send a packet containing user information
• UP latency measurements consider the radio connection is already available
• The time to establish the connection is part of control plane (CP) latency
• Latency can refer to one-way or round-trip-time (RTT)
• Latency can be reported as:
• Average: It can be improved using features and parameter tuning
• Distribution or percentile (e.g. 99% latency): It can be improved with good RF and mobility
performance
• Users are mainly interested in end-to-end latency (i.e. including RAN, core and server
components)
• The scope of this material is the RAN latency

7 © 2023 Nokia

7
User Plane Latency Performance Analysis
Introduction: Factors impacting User Plane Latency
Network Configuration: Probability of collision:
• Determines the frequency of the • After a transmission opportunity, latency depends on if
transmission opportunities the transmission is allocated to another user or not.
• Some network configuration • Increases for high traffic levels impacting the
factors are outside operator distribution of the latency.
control ( e.g. slot duration)
whereas others can be controlled
( e.g. scheduling interval) Retransmissions:
When the packet is transmitted retransmissions can
increase the latency. They can be minimized if RF
UE implementation: performance is optimized
• The behaviour of the chipsets • MAC/RLC retransmissions due to degradation of the
can impact the latency too. radio link quality
• PDCP layers retransmission due to loss of radio link

8 © 2023 Nokia

8
User Plane Latency Performance Analysis
Performance Analysis: Field tests (1/2)
• E2E round trip time can be measured using the ping application
• Pings can be UE originated or server originated
• Field tests use small packets (e.g.32 bytes) to avoid segmentation of the packets into
multiple RLD PDUs
• To isolate RAN delays, it is necessary to take S1-U traces too:

UE originated pings: RAN RTT = E2E RTT (from UE log)- core/server RTT (from S1 trace)
Server originated pings: RAN RTT directly visible in S1 trace

• Experience in 5G networks shows many factors impact the measured RTT (e.g. packet size,
interval between pings…). There is no best practice for doing ping tests
• Applications to measure throughput like Speedtest provide some latency although only
reporting smallest RTT, not the average

9 © 2023 Nokia

9
User Plane Latency Performance Analysis
Performance Analysis: Field tests (2/2)
Example of MO ping traces:
• UE log provides E2E RTT (e.g. ping #233 = 11.1 ms)
• S1-U trace provides the core + server RTT (ping #233 = 294.031 – 293.369 = 0.662 ms)
• Difference provides the terminal + RAN RTT (using the sequence number to link the two
measurements): Ping #233 = 11.1 – 0.7 = 10.4 ms

UE log

S1- U trace

10 © 2023 Nokia

10
User Plane Latency Performance Analysis
Performance Analysis: Interface testing (1/2)
• Best interface to trace the RTT when taking traces for mass traffic analysis is the S1-U
interface between gNB and SGW
• TCP handshake timestamps allow to calculate the core RTT and the RAN & Transport RTT for
every TCP connection
• Information allows to create latency distributions

11 © 2023 Nokia

11
User Plane Latency Performance Analysis
Performance Analysis: Interface testing (2/2)
• Example of TCP handshake latency distribution from one city aggregated to gNB level
SCS=120kHz, 5G19A
• Only 18 worst gNBs shown (x-axis)
• Big difference between median latency and 95th percentile latency

12 © 2023 Nokia

12
User Plane Latency Performance Analysis
Performance Analysis: OSS PM Counters (1/2)
• There are 5G counters to measure the different RAN latency components:
• PDCP delay, RLC delay, F1 delivery delay (estimated), X2 delivery delay (estimated)
• From current experience counters are not reliable to measure actual latency but useful to
investigate causes of high latency (e.g. congestion or packet retransmissions)

DL PDCP Delay

DL RLC Delay
UL RLC Delay
UL PDCP Delay

13 © 2023 Nokia

13
User Plane Latency Performance Analysis
Performance Analysis: OSS PM Counters (2/2)
• Counters may not be reliable to measure the actual latency
• But counters are still useful to investigate causes of high latency such as congestion or packet retransmissions
Counter id Counter name Comment
DL PDCP delay per QoS group (calculated dividing the cumulative delay with number of SDUs). KPI: NR_70b. From filed test, the
M55319C09001..20 DL_DELAY_QOS_GRP_01..20
main reason for it is the congestion of the radio interface.

DL PDCP delay per QoS group (calculated dividing the cumulative delay with number of SDUs). KPI: NR_70b. From filed test, the
M55319C01001..20 DL_SDU_TX_NSA_QOS_GRP_01
main reason for it is the congestion of the radio interface.
M55319C10001..20 UL_DELAY_QOS_GRP_01..20 UL PDCP delay per QoS group (calculated dividing the cumulative delay with the number of SDUs). KPI: NR_71b
M55319C05001..20 UL_SDU_RX_NSA_QOS_GRP_01..20 UL PDCP delay per QoS group (calculated dividing the cumulative delay with the number of SDUs). KPI: NR_71b
M55310C05001..20 DL_RLC_DELAY_L_QOS_GRP_01..20 DL RLC delay per QoS group
M55310C17001..20 DL_RLC_INI_PDU_TX_L_QOS_GRP_01..20 DL RLC delay per QoS group

M55311C10001..20 UL_RLC_DELAY_H_QOS_GRP_01..20 UL RLC delay per QoS group

M55311C13001..20 UL_RLC_COML_RX_H_QOS_GRP_01..20 UL RLC delay per QoS group
Requires Nokia Flow Control algorithm (part of 5GC000630 and 5GC000570 features). The algorithm checks the amount of data
M55323C05001..08 EST_F1_INTF_LATENCY_BIN_00…08
sent over the interfaces and together with the data split algorithm decides the interface used to send the PDUs.
Requires Nokia Flow Control algorithm (part of 5GC000630 and 5GC000570 features). The algorithm checks the amount of data
M55323C06001..09 EST_F1_RLC_QUE_LATENCY_BIN_01..09
sent over the interfaces and together with the data split algorithm decides the interface used to send the PDUs.
Counter behavior depends if 3GPP or Nokia flow control is used.3GPP flow control: it represents the RTT estimation from packet
M55323C11001..09 EST_F1_PDU_DELV_DELAY_BIN_01..09 sent by CU until packet is ack by DDDS (DL Data Delivery Status) message. Nokia flow control: it represents one-way delivery
estimation from packet sent by CU until eNB transmits packet in the air for the first time
M55313C22001..09 EST_PDU_DELV_DELAY_X2_BIN_01..09 Counter behavior depends if 3GPP or Nokia flow control is used. Same comment as above

14 © 2023 Nokia

14
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

15 © 2023 Nokia

15
User Plane Latency Features
Network Characteristics impacting Latency (1/3)

Operating Band: Results from latency networks with different slot durations are not comparable.
Operating band defines subcarrier spacing and slot duration. Shorter slot duration means shorter
median RTT:
• Low band: 15 kHz SCS, 1 ms slot duration
• Mid band: 30 kHz SCS, 0.5 ms slot duration
• mmWave: 120 kHz SCS, 0.125 ms slot duration

F1 topology: Extra latency on cloud developments versus classical ones.

• F1 delay negligible for classical BTS as F1 interface is inside the AirScale.
• F1 delay for cloud BTS may impact latency. Recommended one-way latency in F1 for cloud
BTS is 1-2ms but can be higher if transport network for F1 is not dedicated to 5G
• Impacts the average RTT and the tail of RTT distribution

16 © 2023 Nokia

16
User Plane Latency Features
Network Characteristics impacting Latency (2/3)
NSA versus SA: Differences which potentially can impact the latency:
• SA uses 5G Core Network (5G CN) instead of EPC
• Different core network locations may cause SA traffic to have longer or shorter delay than NSA
• NSA has a risk that some packets go via the X2 interface with longer latency
• SA might have increased risk that the radio connection is very bad or maybe even drop
Example: test network, 3.5 GHz, 5G20A. Similar results for SA and NSA (213.0 ms caused by a single outlier):

17 © 2023 Nokia

17
User Plane Latency Features
Network Characteristics impacting Latency (3/3)
EN-DC configuration and topology (NSA deployments only): Possible additional delay if UL
transmissions are sent with the LTE radio interfaces and forwarded to gNB via X2
• In case LTE and 5G are no co-located the X2 performance may add to the delay

FDD vs TDD: Additional delay in TDD due to the need of waiting for transmission until the uplink
or downlink slot is available

TDD frame structure: Interval between uplink slots affects how fast a packet can be
sent/received on the air interface
• Frame structure to be the same for all macro sites but it is possible to have cases of special frame
structure optimized for isolated micro cells (indoor cells in factories, mines).
• Impacts the average RTT
• It impacts the possible number of beams, RACH formats and achievable DL/UL throughput

18 © 2023 Nokia

18
User Plane Latency Features
Features impacting latency (1/3)
Connected mode DRX
• The UE saves battery by listening occasionally to the PDCCH
• It impacts the median downlink one-way latency if the UE is into sleep mode as it takes longer
to transfer a packet to the UE
• Behavior is controlled via parameterization

Feature id Feature name Release

5GC000772 Common DRX 5G18A
5GC001876 Spillover from 5GC00772 Common DRX 5G19A
5GC002060 Enhanced Common DRX 5G20A

Note: there is potential conflict between the timers for proactive UL scheduling and the
timers for Connected Mode DRX

19 © 2023 Nokia

19
User Plane Latency Features
Features impacting latency (2/3)
Uplink Scheduling
• Scheduling request period impacts latency. If high, the UE has longer waiting times
• Scheduling request are transmitted by the UE on the PUCCH during UL or special slots
• Legacy: Scheduling Request Period = CSI Reporting Period
• With long PUCCH: dedicated parameter for SR (srPeriodicity) removes dependency of latency and
csiReportPeriodicity
• The scheduling request period mainly impacts applications like mobile-originated pings
Short PUCCH case:
• Need for trade-off between latency and CSI reporting overhead:
• Small scheduling requests improve latency but increase the CSI reporting overhead leading to
PUCCH capacity problems

Recommendation: To avoid PUCCH capacity problems, keep default value for

csiReportPeriodicity = 320 slots and rely on proactive UL scheduling for better latencies

20 © 2023 Nokia

20
User Plane Latency Features
Features impacting latency (3/3)
Uplink Scheduling (cont.)
• UL Proactive Scheduling (5GC000252)
• UE is proactively allocated transmission possibilities on the
PUSCH
• UE has short waiting time to transmit: every 15 ms while
csiReportPeriodicity is 160 ms (typical values)
• Proactive UL scheduling will reduce the median uplink
latency
Disadvantage:
• The UE must transmit something every time it is scheduled
• If it does not have real data to transmit, it will transmit
dummy bits
• Starting from 22R2, 5GC002271: Scheduling request
The use of proactive scheduling is therefore a compromise
handling permits the UE to skip transmissions* There is a set of parameters which control
• how frequent the scheduling is done,
• Dummy transmissions increase the PUSCH interference in
the network and reduce the UE battery • for how long period, and
*This requires UE support, so cannot be always used • the size of the scheduled grants

21 © 2023 Nokia

21
User Plane Latency Features
UE implementation impacting latency
Ping RTT depends largely on the UE used for testing. Reasons:
• UE capability e.g. chipset behaviour
• UE Performance issues e.g. if UE does not transmit with enough UL power

Example ping times from different

handsets.
Drive test, MO ping, pro-active
scheduling enabled. 3.5 GHz, 5G19A

Huawei chipset uses same

slot for UL RLC Ack and ping
reply meaning faster RTT
than Qualcomm chipset (MTP
and Xiaomi)

22 © 2023 Nokia

22
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

23 © 2023 Nokia

23
User Plane Latency Optimization Procedure
Optimization procedure overview

• Average latency values are optimized activating features and tuning parameters.
• Those features may impact negatively the network performance (e.g. higher UE battery
consumption, lower throughput) but they can be considered in specific scenarios like
showcasing low latencies or indoor cells in factories

• Optimal distribution of UP latency is related to other optimization aspects:

• Good RF performance: to minimize HARQ/RLC re-transmissions and delays for radio bearer
re-establishments
• Good control plane latency to quickly re-establish the connections if broken
• Good mobility procedures to avoid situations of bad RF
• Good LTE/5G interworking to minimize usage of LTE with its higher air interface latency
• Good network capacity to avoid packet collisions

24 © 2023 Nokia

24
User Plane Latency Optimization Procedure
Optimization procedure (1/3)
1. Check for congestion areas
• Test showing the impact of different radio link loads on average and 99% percentile RRT
Results:
• TCP throughput reached about 400 Mbps
• Average latency was not impacted until the radio link load was high (250 Mbps) while the 99%
latency was immediately impacted with just 10 Mbps / 400 Mbps = 2.5% radio link load

Test conditions Median RTT 99% RTT

FiVe test results with single UE in
No background traffic 12 ms 16 ms stationary conditions.
10 Mbps UDP 13 ms 24 ms SCS = 30 kHz, constant UL scheduling,
50 Mbps UDP 13 ms 42 ms BTS SW = pre-P7 (5G19_5.3835.322).

250 Mbps UDP 15 ms 118 ms

TCP 50 ms 117 ms

25 © 2023 Nokia

25
User Plane Latency Optimization Procedure
Optimization procedure (2/3)
2. Check for bad RF Performance Areas
• HARQ retransmissions can add 3-5 ms additional delay. RLC retransmissions can add 100ms delay
(SCS=30 kHz, 5G19A)
• Test showing ping RTT from drive testing
Results:
• Average RTT is only significantly impacted by very bad RSRP (<-110 dBm) while the 99% percentile is
already different between the “good” and “medium” scenarios

SCS = 30 kHz, proactive UL

scheduling, 5G19A

26 © 2023 Nokia

26
User Plane Latency Optimization Procedure
Optimization procedure (3/3)

3. Check for handover problems

4. Check for frequent use of LTE network (NSA deployments)

• LTE network can be used when 5G radio link is lost so 5G radio link failures need to be investigated
• Field tests with SCS=120 kHz have shown ~10 ms additional RTT delay when the uplink uses LTE
radio interface instead of 5G

27 © 2023 Nokia

27
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

28 © 2023 Nokia

28
Control Plane Latency Performance Analysis
Introduction (1)

• Control plane (CP) latency is the time it takes to establish a 5G radio connection
• NSA deployments:
• Time between when the UE is in idle mode to when it has completed the RACH procedure
• LTE establishment time impacts the CP latency
• SA deployments:
• Time between when the UE is switched on or when the UE is in 5G idle mode to when it has
completed the 5G RACH procedure
• If considering the time from when the UE is switched on then the attach procedure is included
• 5G idle mode DRX parameters and the AMF paging settings have an impact in case of mobile-
terminated signalling

29 © 2023 Nokia

29
Control Plane Latency Performance Analysis
Introduction (2)

• Control plane latency is specially important if 5G handover is not fully implemented and connections
between 5G sectors need to be released and set up again
• High Control plane latency implies lower average 5G throughput
• Control plane latency is usually considered as average instead of as distribution (i.e. percentile)
• Next slides show the factors impacting the CP in case of NSA and SA deployments

30 © 2023 Nokia

30
Control Plane Latency Performance Analysis
Impacting factors (NSA deployments)

Coverage:
UE with marginal 5G coverage: control plane latency increases because multiple access attempts are
needed
UE within 5G coverage: control plane latency depends on if blind or measurement bases SgNB addition
is used
SgNB addition can be delayed:
• Intentionally, allowing time for completing the LTE Carrier Aggregation
• Because ongoing VoLTE calls

31 © 2023 Nokia

31
Control Plane Latency Performance Analysis
Impacting factors (SA deployments)

Expected factors:
• Paging policy in the core network
• Idle mode DRX parameters
• “RRC Inactive” state enables fast connection setup but it is not supported in 22R2

32 © 2023 Nokia

32
Control Plane Latency Performance Analysis
Performance Analysis
• There are no counters to measure the control plane latency.
• Control plane latency is measured by tracing the UE or the BTS.
• NSA deployments: both, LTE and 5G messages are needed:
• UE side: it is possible to collect them simultaneously
• BTS side: it is necessary to trace separately the 4G and the 5G BTS
• CP latency is measured differently depending if the random access is contention based or not and on
the deployment type (NSA or SA)

Contention based RA
NSA deployments:
CP Latency = time of 5G RACH ‘msg3’ – time of 4G preamble (or 4G RRC Conn. Req)

SA deployments:
CP Latency = time of 5G RACH ‘msg3’ – time of 5G preamble (or 5G RRC Conn. Req)

33 © 2023 Nokia

33
Control Plane Latency Performance Analysis
Performance Analysis (cont.)

Contention free RA
NSA deployments:
CP Latency = time of 5G RACH ‘msg2’– time of 4G preamble (or 4G RRC Conn. Req)

SA deployments:
CP Latency = time of 5G RACH ‘msg2’ – time of 5G preamble (or 5G RRC Conn. Req)

• Long CP latency can be due to unsuccessful 1st call setup attempt so 5G accessibility issues need to
be analysed

34 © 2023 Nokia

34
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

35 © 2023 Nokia

35
Control Plane Latency Features and Optimization Procedure
Features
• Following features (LTE and 5G) are relevant when analysing the control plane latency:
Feature id Feature name Release
LTE4193 Dynamic Trigger for LTE-NR Option 3X LTE19
LTE5524 Ongoing QCI1 prevents EN-DC setup LTE19A
5GC001874 Non contention based random access 5G19A
5GC000723 RRC state handling - Idle to Connected (SA mode) 5G19B
5GC000733 Paging support in SA mode 5G19B
LTE5003 Data buffer trigger for EN-DC SRAN20A
LTE5388 VoLTE without EN-DC SRAN20A
LTE5667 Extended B1 measurement report for NR SRAN20C

36 © 2023 Nokia

36
Control Plane Latency Features and Optimization Procedure
Features: LTE4193 Dynamic Trigger for LTE-NR Option 3X (1/3)
NSA only

• Feature enables the measurement based SgNB addition i.e. it is only attempted in areas with
minimum 5G coverage
• UE needs to measure the coverage and report it to the 4G BTS incurring in additional time
• Recommendation: use the feature to avoid switching the user plane to the gNB if 5G radio
connection is not likely to be established
• However, blind addition can be considered in cells where faster 5G CP latency is required and
the 4G/5G coverage are similar
• Note: blind addition is currently slower than measurement based addition due to a guard timer
of 1000ms introduced by R&D to avoid deadlock on S1 interface (in cases of other MME
vendors)

37 © 2023 Nokia

37
Control Plane Latency Features and Optimization Procedure
Features: LTE4193 Dynamic Trigger for LTE-NR Option 3X (2/3)

Test: Control plane latency measurements with and without measurement based
SgNB addition
• Guard timer set to 1000 ms NSA only
• Blind addition takes longer than measurement-based addition

38 © 2023 Nokia

38
Control Plane Latency Features and Optimization Procedure
Features: LTE4193 Dynamic Trigger for LTE-NR Option 3X (3/3)

Test: SgNB addition time including the delay components breakdown NSA only
• b1TimeToTriggerRsrp =256 ms
• FiVe drive test with 5G19B/ SRAN20A. 29 samples

Dominant factor is the

B1measurement delay

39 © 2023 Nokia

39
Control Plane Latency Features and Optimization Procedure
Features: Ongoing VoLTE calls
NSA only

• Feature LTE5524: Ongoing QCI1 prevents EN-DC setup delays the SgNB if there is an ongoing
VoLTE call (using QCI1) so its quality is not impacted
• VoLTE traffic is always routed through the 4G BTS
• Waiting until the VoLTE call is completed can impact largely on the SgNB addition delay

• Feature LTE5388: VoLTE without EN-DC extends the functionality to VoLTE calls using QCI65
and QCI66

40 © 2023 Nokia

40
Control Plane Latency Features and Optimization Procedure
Features: Contention free random access
NSA only
Non contention based random access for NSA: 5GC001874
• Allows faster and more reliable access than contention-based RA
• Requires SRAN20C feature LTE5667: Extended B1 measurement report for NR
• Contention-free random access can accelerate the 5G RACH procedure due to:
• Msg3 is not used making the procedure faster and improving the average latency
• Avoids preamble collisions delays improving the latency distribution
• Shorter RA procedure decreases the risks of radio failure during the procedure and to
start a new connection improving the latency distribution

Field test results with contention-free random access for the handover procedure do not
show significant improvements in the RACH time

41 © 2023 Nokia

41
Control Plane Latency Features and Optimization Procedure
Features: Idle Mode and SA paging

SA

• 5GC000733: Paging support in SA mode enables paging to allow for mobile-terminated calls
• 5GC000723: RRC state handling – Idle to Connected (SA mode) controls the idle to connected
mode transition
• Both features have parameters for repetition attempts and intervals, which impact the control
plane latency
• The core network (AMF) also has paging-related messages that can impact the control plane
latency for mobile-terminated calls

42 © 2023 Nokia

42
Control Plane Latency Features and Optimization Procedure
Features: SA versus NSA
• SA mode allows much faster connection setup time SA and NSA
• It needs to be kept in mind that the time for the UE to camp on the 5G cell in
SA / 4G cell in NSA is not included in the measurement
• The time for the UE to do the B1 measurements on the 5G cell is included

Measurements of NSA and SA control plane

latencies in FiVe test network. NSA result
measured from RRC setup request (LTE) to RACH
OK (NR). SA result measured from RRC setup
request (NR) to RRC reconfiguration complete
(NR). SA mode is clearly shorter. 5G20A

43 © 2023 Nokia

43
Control Plane Latency Features and Optimization Procedure
Optimization Procedure

1. Optimize the accessibility as it will impact the control plane latency. In case of NSA, LTE RRC
establishment time also has an impact on it
2. Usage and/or parameterization of features although they may impact negatively other
performance aspects:
• Disable the measurement based SgNB addition
• If keeping the measurement-based SgNB addition, decrease the time to trigger. Note: it may
increase the failures in 5G RACH procedure, and it also may create issues with 4G carrier
aggregation and multiple DRB use
• Activate contention free random access. Note: It requires LTE20C feature
• In principle, the frequency of the RACH occurrences plays a role for how fast the UE can send the
5G preamble. Also, the periodicity of the SSB burst set can impact the time it takes for the UE to
synchronize to the 5G cell. In both cases, the impact is probably negligible compared to the total
duration of the access procedure and can probably be ignored. Especially the B1 measurements
(NSA) and the interactions with the core network is likely to take much longer

44 © 2023 Nokia

44
Index

User Plane Latency Performance Analysis

User Plane Latency Features
User Plane Latency Optimization Procedure
Control Plane Latency Performance Analysis
Control Plane Latency Features and Optimization Procedure
Quiz/Exercise

45 © 2023 Nokia

45
Quiz/Exercise
Exercise 1

Deploy the Log file:

2019-12-04-17-46-35-0000-0000-0412-5514-S

Q:
a) Which type of SgNB addition is in use (blind or measurement based) ?

b) Check for Control plane latency. Is the measured value in the expected range?

46 © 2023 Nokia

46
47