Training Course - 5G RAN3.0 KPI Introduction V1.00
Training Course - 5G RAN3.0 KPI Introduction V1.00
Training Course - 5G RAN3.0 KPI Introduction V1.00
0 KPI
www.huawei.com
2. 5G KPI Description
Measurement object
Remarks: The design of counters related to QCI and 5QI is different in LTE and 5G. In LTE, Such counters
are designed at the counter level in LTE and at the counter object level in 5G. That is, such counters are
reported only after the corresponding counter objects are activated. The maximum number of QCI and 5QI
objects must not exceed 20.
Major Impact
GNBCUUPFunction (SDAP and PDCP layer
measurement) in the protocol does not include
gNodeB-level cell-level measurement. That is,
1) Currently, NR cell PDCP counters are
measurement incompatible with the protocol.
in 3GPP 2) PDCP counters cannot measure cell-level
performance, including traffic, packet loss, and
delay.
NR DU cell NR cell 3) Retainability KPIs cannot be designed
following LTE. This is because LTE retainability
KPIs count data transmission in the user buffer
but CUCP in 5G protocols cannot perceive the
data transmission in buffer.
HO.NRCellRelation.Out.Cell Intra-RAT Inter-cell Outgoing Handover Measurement RAN2.1 Added counters related to intra-RAT handover between two cells.
IRATHO.NrCell.ECell Inter-RAT Inter-cell Handover Measurement RAN2.1 Added counters related to inter-RAT handover between two cells.
SecurMode.Cell Security Mode Measurement RAN2.1 Added counters related to security mode.
Traffic.RLC.DuCell.Operator RLC Measurement for All Users of Operator RAN2.1
Added operator-related counters.
Traffic.User.Cell.Operator Cell User Quantity Measurement of Operator RAN2.1
User.DuCell.QCI DU Cell QCI User Quantity Measurement RAN2.1
Added QCI-related counters in the NSA networking.
Thp.DuCell.QCI QCI Throughput Measurement RAN2.1
Thp.PDF.DuCell Throughput Distribution Measurement RAN2.1 Added counters related to throughput range distribution.
PRB.DuCell PRB Measurement RAN1.0 Added PDCCH CCE aggregation level statistics.
Added maximum transmit power and average power of a cell (for
ChMeas.Pwr. DuCell Channel Quality Measurement RAN1.0
EMF).
MAC.DuCell MAC Measurement RAN1.0 Added downlink IBLER statistics.
RA.DuCell Random Access Measurement RAN1.0 Added TA range distribution.
Measurement per MCS Index in the MCS Index Table for
ChMeas.PDSCH.64QAM.MCS.DuCell RAN2.1
PDSCH with 64QAM Added counters related to the number of RBs of downlink MCS
Measurement per MCS Index in the MCS Index Table for indexes corresponding to ranks in different MCS index tables.
ChMeas.PDSCH.256QAM.MCS.DuCell RAN2.1
PDSCH with 256QAM
Measurement per MCS Index in the MCS Index Table for
ChMeas.PUSCH.64QAM.MCS.DuCell RAN2.1
PUSCH with 64QAM
Added counters related to the number of RBs of uplink MCS
Measurement per MCS Index in the MCS Index Table for
ChMeas.PUSCH.256QAM.MCS.DuCell RAN2.1 indexes corresponding to ranks in different MCS index tables.
PUSCH with 256QAM
Measurement per MCS Index in the MCS Index Table for
ChMeas.PUSCH.64QAMTP.MCS.DuCell RAN2.1
PUSCH with Transform Precoding and 64QAM
ChMeas.64QAM.CQI.DuCell Measurement of CQIs in the CQI Table with 64QAM RAN2.1 Added counters related to CQIs corresponding to ranks in
ChMeas.256QAM.CQI.DuCell Measurement of CQIs in the CQI Table with 256QAM RAN2.1 different CQI tables.
3GPP has discussed E2E KPIs in SA5. In November 2018, some SA networking counters and KPIs are discussed.
However, due to the separation of CUCP and CUUP, retainability KPIs are introduced in 3GPP specifications in June
2019. The implementation in RAN3 is still pending and is not supported in this version yet.
RAN KPIs (Layer 2 Measurements) are still under discussion and have not been officially released. It is expected that
KPIs inherit the SA5 conclusion.
28.554 5G end to end Key Performance Indicators (KPI) (5G E2E KPI)
Add performance measurements for ng-eNB in terms of connectivity with 5GC (management and organization networks are sliced; performance
32.425
measurement and guarantee data of NR and NG-RAN)
32.426 Add performance measurements for EPC in terms of connectivity with NR (added gNodeB performance measurement for the 5G core network)
32.450 Add KPIs for ng-eNB in terms of connectivity with 5GC (added NR performance measurement for the EPC)
32.451 Add KPI requirements for ng-eNB in terms of connectivity with 5GC (added KPI requirements for gNodeB for the 5G core network)
32.455 Add KPIs for EPC in terms of connectivity with NR (added KPI requirements for gNodeB for the 5G core network)
2. 5G KPI Description
In SA architecture, 5G-dedicated
Accessibility accessibility KPIs, retainability KPIs,
and mobility KPIs will be designed.
Retainability SA only
In NSA architecture, it is
recommended that accessibility (planning)
Mobility KPIs, retainability KPIs, and mobility
KPIs be obtained from the LTE side.
KPI architecture
• User Uplink/Downlink Average Throughput
(counter-based) Service integrity • Cell Uplink/Downlink Average Throughput
• PRB Usage
Utilization • CPU Usage
SA & NSA
Availability • Radio Network Unavailability Rate
Range
Description
Name
KPI
Formula
Associated
counter
A D
Traffic Evaluation
PDCP PDCP A Upper layer <--> gNodeB PDCP
MCG
SCG SCG MCG split SCG
split
MCG C gNodeB PDCP <--> eNodeB RLC
bearer split bearer 1 split bearer
bearer 2
bearer 1 bearer 2
B F C E D Upper layer <--> eNodeB PDCP
Note: Currently, these counters are measured at the cell level under the NR cell. It does not comply with the
NRM model specified in 3GPP TS 28.541.
Service integrity KPIs are used to evaluate the service quality of end users in the 5G RAN, which include:
User Uplink/Downlink Average Throughput
Cell Uplink/Downlink Average Throughput Refer to 3GPP TS 28.554 and 3GPP 28.552.
Successful transmission,
buffer not empty
The last slot shall always
be removed from
Failed transmission block error
ThpTimeDl
calculations since it
Successful transmission,
can be impacted
buffer empty
by packet size of
User Plane (UP) packets.
No transmission, buffer not
empty (e.g. due to contention)
These KPIs indicate the average uplink and downlink UE throughputs in a cell.
Name User Uplink Average Throughput Name User Downlink Average Throughput
Measurement Measurement
Cell or RAN Cell or RAN
Object Object
UserULAveThp = UserDLAveThp =
Formula UserULRmvSmallPktTrafficVolume/UserULR Formula UserDLRmvLastSlotTrafficVolume/UserDLRm
mvSmallPktTransferTime vLastSlotTransferTime
Name Cell Uplink Average Throughput Name Cell Downlink Average Throughput
Measurement Measurement
Cell or RAN Cell or RAN
Object Object
CellULAveThp =
Formula CellDLAveThp =
CellULTrafficVolume/CellULTransferTime Formula
CellDLTrafficVolume/CellDLTransferTime
Cell Uplink Average Throughput =
Cell Downlink Average Throughput =
N.ThpVol.UL.Cell/N.ThpTime.UL.Cell N.ThpVol.DL.Cell/N.ThpTime.DL.Cell
Associated Note: Associated Note:
Counter Counter
When comparing this KPI with the theoretical When comparing this KPI with the theoretical
peak rate for TDD, pay attention to the peak rate for TDD, pay attention to the uplink-
uplink-downlink subframe configuration. downlink subframe configuration.
Unit Gbit/s Unit Gbit/s
Utilization KPIs are used to evaluate the capabilities, such as the capability to meet
traffic demands, in specific internal conditions. Utilization KPIs include:
Uplink/Downlink Resource Block Utilizing Rate
Average CPU Load
These KPIs indicate the busy-hour uplink and downlink RB usages in a cell or RAN.
Name Uplink Resource Block Utilizing Rate Name Downlink Resource Block Utilizing Rate
Measurement Measurement
Cell or RAN Cell or RAN
Object Object
RB_URUL = (RB_UsedUL/RB_AvailableUL) x RB_URDL = (RB_UsedDL/RB_AvailableDL) x
Formula Formula
100% 100%
Uplink Resource Block Utilizing Rate = Downlink Resource Block Utilizing Rate =
Associated Associated
(N.PRB.UL.Used.Avg/N.PRB.UL.Avail.Avg) x (N.PRB.DL.Used.Avg/N.PRB.DL.Avail.Avg) x
Counter Counter
100% 100%
Unit % Unit %
Formula MeanCPUUtility
Unit %
This KPI indicates the percentage of time when cells in a radio network are
unavailable.
Measurement
Radio network
Object
RAN_Unavail_Rate =
Formula (ΣCellUnavailTime/(TheTotalNumberOfCellsInCluster × {SP} x 60)) x
100%
Traffic KPIs are used to measure the traffic volume on the 5G RAN. Traffic KPIs
include:
Uplink/Downlink Traffic Volume
Average/Maximum User Number
These KPIs indicate the uplink and downlink traffic volumes in a cell, which are
measured at the Radio Link Control (RLC) layer.
Associated Associated
Uplink Traffic Volume = N.ThpVol.UL Downlink Traffic Volume =N.ThpVol.DL
Counter Counter
Unit kbit Unit kbit
Accessibility KPIs reflect the probability that a UE successfully accesses the network
and initiates services, including:
RRC connection setup
NGSIG setup
QoS flow setup
This KPI is used to evaluate the RRC connection setup success rate.
Measurement
Cell or RAN
Object
Unit %
This KPI is used to evaluate the RRC connection setup success rate of services.
Unit %
This KPI is used to evaluate the RRC connection setup success rate of signaling.
RRCS_SRsignaling = (RRCSetupSuccesssignaling/RRCSetupAttemptsignaling) x
Formula
100%
Unit %
This KPI is used to evaluate the signaling connection setup success rate of the NG
interface.
NGSIGS_SR =
Formula
(NGSIGConnectionEstablishSuccess/NGSIGConnectionEstablishAttempt) x 100%
Unit %
This KPI is used to evaluate the QoS flow setup success rate of all services.
Unit %
This KPI is used to evaluate the call setup success rate of all services.
CSSR = (RRCSetupSuccessservice/RRCSetupAttemptservice) x
Formula (NGSIGConnectionEstablishSuccess/NGSIGConnectionEstablishAttempt) x
(QosFlowSetupSuccess/QosFlowSetupAttempt) x 100%
Unit %
Mobility KPIs are used to evaluate the mobility performance of the NR network, which
directly reflect user experience. Based on the handover type, mobility KPIs are
classified into the following types:
Intra-RAT intra-frequency handover KPIs
Intra-RAT inter-frequency handover KPIs
Inter-RAT handover KPIs
This KPI is used to evaluate the intra-NR intra-frequency outgoing handover success
rate.
IntraFreqHOOut_SR = (IntraFreqHOOutSuccess/IntraFreqHOOutAttempt) x
Formula
100%
Unit %
This KPI is used to evaluate the intra-NR inter-frequency outgoing handover success rate.
Unit %
This KPI is used to evaluate the intra-NR incoming handover success rate.
Intra-RATHOIn_SR = (Intra-RATHOInSuccess/Intra-RATHOInAttempt) x
Formula
100%
Unit %
This KPI is used to evaluate the inter-RAT outgoing handover success rate from NR
to LTE.
IRATHO_N2E_SR = (IRATHO_N2E_Success/IRATHO_N2E_Attempt) x
Formula
100%
Inter-RAT Handover Out Success Rate (NR to LTE) =
Associated Counter (N.HO.InterRAT.N2E.ExecSuccOut)/(N.HO.InterRAT.N2E.ExecAttOut) x
100%
Unit %
Remarks: If the failures caused by the core network do not need to be counted in some scenarios,
the following formula can be used:
Inter-RAT Handover Out Success Rate (NR to LTE) = (N.HO.InterRAT.N2E.ExecSuccOut –
N.HO.InterRAT.N2E.AMFAbnormRsp)/(N.HO.InterRAT.N2E.ExecAttOut) x 100%
EPSFB_N2E_BasedHO_SR =
Formula (EPSFB_N2E_BasedHO_Success/EPSFB_N2E_BasedHO_Attempt) x
100%
Unit %
Measurement
Type Name Formula Unit
Object
Average downlink cell throughput (Cell
N.ThpVol.DL.Cell/N.ThpTime.DL.Cell Cell or RAN Gbit/s
Downlink Average Throughput)
Average uplink cell throughput (Cell Uplink Cell Uplink Average Throughput =
Cell or RAN Gbit/s
Average Throughput) N.ThpVol.UL.Cell/N.ThpTime.UL.Cell
Service
(N.ThpVol.UL –
Integrity KPIs Average uplink UE throughput (User Uplink
N.ThpVol.UE.UL.SmallPkt)/N.ThpTime.UE. Cell or RAN Gbit/s
Average Throughput)
UL.RmvSmallPkt
(N.ThpVol.DL-
Average downlink UE throughput (User
N.ThpVol.UE.DL.SmallPkt)/N.ThpTime.UE. Cell or RAN Gbit/s
Downlink Average Throughput)
DL.RmvSmallPkt
Uplink PRB usage (Uplink Resource Block (N.PRB.UL.Used.Avg/N.PRB.UL.Avail.Avg)
Cell or RAN %
Utilizing Rate) x 100%
Utilization Downlink PRB usage (Downlink Resource (N.PRB.DL.Used.Avg/N.PRB.DL.Avail.Avg)
KPIs Cell or RAN %
Block Utilizing Rate) x 100%
Average CPU usage (Average CPU Load) VS.BBUBoard.CPULoad.Mean CPU %
Measurement
Type Name Formula Unit
Object
(RRCSetupSuccessservice/RRCSetupAttemptservice) x
Call Setup Success Rate (NGSIGConnectionEstablishSuccess/NGSIGConnectionEstablishAttempt) x
(QosFlowSetupSuccess/QosFlowSetupAttempt)
Measurement
Type Name Formula Unit
Object
(N.HO.IntraFreq.Ng.IntergNB.ExecSuccOut +
Intra-Frequency N.HO.IntraFreq.Xn.IntergNB.ExecSuccOut +
Handover Out N.HO.IntraFreq.IntragNB.ExecSuccOut)/(N.HO.IntraFreq.Ng.IntergNB.ExecAtt
Success Rate Out + N.HO.IntraFreq.IntragNB.ExecAttOut +
N.HO.IntraFreq.Xn.IntergNB.ExecAttOut)
(N.HO.InterFreq.IntragNB.ExecSuccOut +
Mobility
Inter-frequency Cell or RAN %
KPIs N.HO.InterFreq.Xn.IntergNB.ExecSuccOut +
Handover Out N.HO.InterFreq.Ng.IntergNB.ExecSuccOut)/(N.HO.InterFreq.IntragNB.ExecAtt
Success Rate Out + N.HO.InterFreq.Xn.IntergNB.ExecAttOut +
N.HO.InterFreq.Ng.IntergNB.ExecAttOut)
Measurement
Type Name Formula Unit
Object
(N.HO.InterRAT.N2E.ExecSuccOut)/(N.HO.InterRAT.N2E.ExecAttOut)
Remarks: If the failures caused by the core network do not need to be
Inter-RAT Handover counted in some scenarios, the following formula can be used:
Out Success Rate Inter-RAT Handover Out Success Rate (NR to LTE) =
(N.HO.InterRAT.N2E.ExecSuccOut –
N.HO.InterRAT.N2E.AMFAbnormRsp)/(N.HO.InterRAT.N2E.ExecAttOut) x
Mobility 100% Cell or RAN %
KPIs
EPSFB Success
Rate Based N.HO.InterRAT.N2E.EPSFB.ExecSuccOut/
Handover (NR to E- N.HO.InterRAT.N2E.EPSFB.ExecAttOut
UTRAN)
or ava is f cise is
CQI 5 reporting times in 256QAM Table /(Total number of Proportion of RBs with Rank 2 MCS 10 in 256QAM size). If the last timeslot is excluded, the downlink cell
CQI reportings, including different CQI tables)
t e
Table = Number of RBs with Rank 2 MCS 10 in
h is . It er
256QAM Table/Total number of RBs corresponding to
th
throughput is not affected by the service model.
Proportion of CQI spectral efficiency (spectral efficiency:
a
s data ork y. E otin
the MCS
x g
i
1.9141 x 2) measured by UE = Proportion of Rank 2 CQI Proportion of scheduling spectral efficiency (spectral
is o etw nl m .
5 in 256QAM Table efficiency: 2.5703 x 2) = Proportion of RBs with Rank 2
Measurement point: UE h n n epoint:
T dMeasurement
MCS 10 in 256QAM Table
o MAC r o rs
n liv us n e p e
a l he to Impact m
Factor Impact e
Factor
thUE channel a
n quality
w s Factor Impact
f ea
Air interface resource Affected
Air interface resource Affected
Air interface resource Not affected Number of scheduling times Affected Number of scheduling times Affected
(traffic model, timeslot, and other (traffic model, timeslot, and other
CQI reporting frequency Affected numerologies are also affected.) numerologies are also affected.)
Example: Problem:
Spectral efficiency of table2 MCS index10 with RANK2 = 2.5703 x 2 = 5.1406 1. Did not distinguish MCS index table: Different spectral efficiencies in
different tables.
Ratio of spectral efficiency to 5.1406 = (Number of scheduled RBs in table2
MCS index10)/(Number of RBs scheduled by all MCS) 2. Did not distinguish RANK: With the same MCS, the spectral efficiency
Notes: 2.5703 is from "Table 5.1.3.1-2: MCS index table 2 for PDSCH". of RANK2 may be twice that of RANK1.
2. 5G KPI Description
3. RRCConnectionReconfiguration
B 1911816753 SgNB.Abn releases in an SgNB Release Required message to
4. RRCConnectionReconfigurationComplete ormRel the LTE-NR the eNodeB, the cause in the SgNB
5. SN Status Transfer NSA DC Release Required message is "Radio
6. Data Forwarding
scenario Connection With UE Lost" or "Failure in the
7. Secondary RAT Data Volume report
Radio Interface". The counter value is
8. Path Update procedure accumulated in the PSCell of the LTE-NR
9. UE Context Release NSA DC UE.
4. RRCConnectionReconfiguration
Attempts in the LTE- SgNB Change Required message to the eNodeB.
5. RRCConnectionReconfigurationComplete ange.Att
NR NSA DC Scenario The counter value is accumulated in the PSCell of
8. Random Access Procedure
6. SgNB Change Confirm
B
7. SgNB Reconfiguration Complete the LTE-NR NSA DC UE.
9a. SN Status Transfer
9b. SN Status Transfer As shown at point B in Figure 2, the
Number of Successful N.NsaDc.InterSgNB.PSCell.Change.Succ counter
10. Data Forwarding
N.NsaDc.InterS
Inter-SgNB PSCell is incremented by 1 each time the gNodeB
11. Secondary RAT Data Volume Report 1911816749 gNB.PSCell.Ch
Changes in the LTE- receives an SgNB Change Confirm message from
12. E-RAB Modification Indication
ange.Succ
14. End Marker Packet
13. Bearer Modification
NR NSA DC Scenario the eNodeB. The counter value is accumulated in
15. New Path the PSCell of the LTE-NR NSA DC UE.
16. E-RAB Modification Confirm
Figure 1
Counter Counter
Counter ID Measurement Point
Name Description
1. Do access and service drops exist in NSA networking? If they do, how to
define them? Is SgNB addition or release equivalent to service drops in
traditional networks? nly
n o
sio
2. What are the objectives of NSA DC networkus optimization? How to evaluate
sc
the quality of an NSA network? al di
r n
nt e
r i
F o
c u tio rs
is au te
1 Number of NSA UEs N/A Average number of RRC connections of an NSA UE Operator UE promotion
h
state/Average number of RRC connections of an NSA UE
r n it c ted
Top causes for low proportion of NSA UEs in dual-
N/A
t e w e N/A Registration on the operator
in se om en
connectivity state - registration on core network
r
Fo ly. U se s plem
(Number of normal SgNB releases triggered by SgNB
Top causes for low proportion of NSA UEs in dual- Service promotion by operators
– Number of SgNB releases triggered by
connectivity state - dual-connectivity release due to no and the requirement of user
on cau t im
coverage)/Number of SgNB releases triggered by
requirement activation
SgNB
3
be e no
Top causes for low proportion of NSA UEs in dual- Resource and configuration
SgNB access: SgNB addition success rate
connectivity state - resource configuration problems
Top causes for low proportion of NSA UEs in dual- Retainability of the SgNB for NSA UEs in dual-
Coverage
connectivity state - NSA DRB service drop rate connectivity state: NSA DRB service drop rate
SgNB handover (mobility): SgNB change success MeNB handover (mobility): Intra-RAT handover success
4 Coverage continuity (handover) Coverage
rate rate with SN
Figure 1
Counter ID Counter Name Counter Description Measurement Point
UE MeNB SgNB
Total number of SgNB As shown at point A in figure 1, the
addition attempts for UEs L.NsaDc.SgNB.Add.Att counter is
SgNB Addition Request L.NsaDc.SgNB.
1526747851 that treat the local cell as incremented each time the eNodeB
Add.Att
A SgNB Addition Request
their PCell in the LTE-NR sends an SgNB Addition Request
Acknowledge NSA DC state message to the gNodeB.
RRCConnectionReconfiguration
RRCConnectionReconfiguration
Complete
Total number of successful As shown at point B in figure 1, the
SgNB Reconfiguration Complete SgNB additions for UEs L.NsaDc.SgNB.Add.Succ counter is
L.NsaDc.SgNB.
1526747852 that treat the local cell as incremented each time the eNodeB
B Add.Succ
their PCell in the LTE-NR sends an SgNB Reconfiguration
NSA DC state Complete message to the gNodeB.
UE MeNB SgNB
Counter ID Counter Name Counter Description Measurement Point
SgNB Release Request
A As shown at point A in figure 1, the
RRCConnectionReconfiguration L.NsaDc.SgNB.Rmv.Att counter is
incremented each time the eNodeB
RRCConnectionReconfiguration Total number of SgNB
sends an SgNB Release Request
Complete removal attempts for
L.NsaDc.SgNB message to the gNodeB.
1526747855 UEs that treat the local
.Rmv.Att As shown at point A in figure 2, the
cell as their PCell in the
L.NsaDc.SgNB.Rmv.Att counter is
LTE-NR NSA DC state
incremented each time the eNodeB
Figure 2 sends an SgNB Release Confirm
message to the gNodeB.
UE MeNB SgNB
Counter Counter
Counter ID Measurement Point
Name Description
UE EUTRAN
As shown at point A in the figure, the
L.NsaDc.ScgFailure counter is incremented
RRC connection reconfiguration Total number of
(Scg-Configuration) each time the eNodeB receives an
SCG failures for
SCGFailureInfomationNR message from a
L.NsaDc.Scg UEs that treat the
SCGFailureInformation 1526747856 UE. Before that, on the NR side, the UE
Failure local cell as their
experiences a random access failure, the
PCell in the LTE-
A number of RLC retransmissions reaches its
NR NSA DC state
maximum value for the UE, or the SCG
configuration fails for the UE.
After receiving the SCG Failure sent by the UE, the MeNB triggers delivery of the SgNB Release message.