lte LOW throughput
analysis & improvement
techniques
objective
Analysis flow for low throughput
› throughput investigation from
performance counters point of view
› Demonstrate throughput improvement with
practical cases.
› Key learnings.
objective
Analysis flow for low throughput
› throughput investigation from
performance counters point of view
› Demonstrate throughput improvement with
practical cases.
› Key learnings.
Content
Analysis flow for DL throughput
› throughput
Analysis flow for UL throughput
› throughput
› Sample case Dl throughput
thr oughput investigation
› Sample case Ul throughput
thr oughput investigation
› Parameter settings at E NodeB
› Essential features at E NodeB
› KPI/Counters to check for throughput analysis
› Key Learnings
Analysis Flow for Dl throughput
investigation
Total Check all important Accessibly, Mobility and Retainability KPI.
performance Holistic overview of perfo rmance makes analysis simpler
RRC connected Dl Throughput would reduce with increase in number of
users connected users (pmRrcConnMax).
CQI ,64/16QAM,TM Av CQI, % 64QAM samples and RI indicates DL SINR status
modes Average CQI should be high (>10%), % of 64QAM sample should be high (>10%),
PRB(DL) and High PRB and PDCCH utilization
util ization would impact the DL
PDCCH utilization Throughput (pmPrbUtilDl / pmPrbUtilUl).
DL Latency and High value of DL latency(>9ms) and RLC retransmission(> 1%)
RLC retransmission would impact DL throughput.
Power limited UE High occurrence of pmRadioTbsPwrRestricted and
and No of A2 events pmBadCovEvalReport indicates poor DL coverage.
Basic health check up and baseline parameter/feature Audit along
with all checks
Analysis Flow for UL throughput
investigation
Total Check all important Accessibly, Mobility and Retainability KPI.
performance Holistic overview of perfo rmance makes analysis simpler
Alarm & Parameter/
Make Baseline Audit for Parameter and feature
Feature Check
High uplink RSSI would impact the throughput
RSSI
(pmRadioRecInterferencePwrPUCCH / pmRadioRecInterferencePwr )
% of 16 QAM Low usage of 16 QAM modulations scheme in UL would impact
samples the UL throughput
PUCCH & PUSCH Poor UL SINR conditions would impact UL throughput
SINR (pmsinrpucchdistr & pmsinrpuschdistr )
High number of power limited UE indicates poor uplink coverage
Power limited UE
(pmRadioTbsPwrRestricted ,pmRadioTbsPwrUNRestricted)
DL throughput
investigation
DL throughput
investigation
Cell Root cause Action plan
090125002 Low average CQI/Low Physical optimization
use of 64 QAM samples
150166001 Low handover success Neighbor recreation/PCI
rate/PCI conflict Replan
094962003 Missing MO/Handover Parameter Audit
relations
095330003 High PRB utilization Add newsite/Bandwidth
Sample case 090125002
% of 64 QAM samples & Av CQI were low, Phy optimization ( EDT 2 4) was made.
Good improvement in all matrices were observed post optimization
Sample case 090125002
Good improvement in Throughput & NO reduction in Data Volume post
physical optimization
Sample case 090125002
Change in
Electric down tilt
from 2 to 4
Improvement in UE power limiting ratio post changes
Sample case 090125002
Clear improvement in LOW RSRQ samples post changes
(RSRQ comparison is from PM UE initiated measurement)
Sample case_150166001
CELL HAD poor MOBILITY & Retainability KPI
When ALL Neighbor relation were deleted ANR created relation with immediate Neighbor instead
of remote relation with SAME PCI
Sample case_150166001
NBR relation “150166001_33420-90427-3” prevented creation of
relation 150166001_33420-150164-3 BY ANR
As Both target cell have same PCI
Sample case_150166001
When All Relation were deleted ANR created relation with natural
neighbor 150164-3 and KPI turned good
Relation 150166001_33420-90427-3 was not created again
Sample case_150166001
.Good improvement in all matrices were observed post relation creation
with immediate neighbor
Sample case_150166001
% of 64 QAM samples ,Av CQI,MIMO usage improved post changes
Good improvement in all matrices were observed post relation creation with immediate neighbor
Samplecase_094962003
ANR was not adding any neighbor relations for this indoor cell.
Investigation confirmed missing “UTRANfreqrelation” & “EUtranFreqRelation”
MO for this cell
After the creation of these MO, KPI improved
Samplecase_094962003
Improvement in Throughput along with other KPI POST MO CREATION
Samplecase_094962003
Improvement in Throughput along with other KPI POST MO CREATION
Sample case 095330003
SITE DF5330 is highly utilized and throughput for all three cell reduces with
increase in data volume
No user license limitation in these cells as RRC success rate is quite good
95330003 Down link PRB
utilization and Dl throughput
In off peak hours -throughput is quite good
In peak hours -High downlink PRB utilization is limiting the
downlink throughput.
95330003 Down link PRB
utilization and Dl throughput
High value of connected user, DL PRB and scheduler utilization
Ways to reduce PRB utilization
› There are very limited means of reducing of PRB
utilization .How ever some of these methods can be tried
on case by case basis
› Traffic offload to less utilized neighboring cells
› Reduce control channel resources(Before that check
PDCCH utilization)
Add Bandwidth (if system is operating < 20Mhz) as
›
Bandwidth increase would increase number of PRB
› Reduce inactivity timer (tInactivityTimer) value so that
inactive user can be released early
DF5330-Data Volume
Qoffsetcelleutran to -2
Qoffsetcelleutran to -6
Qoffsetcelleutran to -4
Physical optimization
an and Tilt changes
Physical optimization & Qoffsetcelleutran applied for Traffic offloading to neighboring cell but it did not
help much.
OFFLOADING BY new site/ Bandwidth addition/ second carrier addition are the available options
UL throughput sample
cases
UL throughput
investigation
Cell Root cause Action plan
93704001 Low uplink DAS Check and Audit
coverage/uplink issue
in DAS(Distributed
Antenna system)
155980003 High RSSI/uplink UL Power control
interference parameter tuning /PIM
testing for internal
interference
Sample case_93704001
CELL 93704001 (IBS cell) has low throughput even though DL throughput
is quite good
Sample case_93704001
*Cell 15014002 & 15014002 shown just for comparison
Data from UL SINR counters confirms poor SINR condition at the cell
Sample case_93704001
UETR TRACES FOR CALL DROP
CTR also confirms uplink issue at the cell as there are many call
drop related to poor uplink coverage
ISSUE forwarded to o&M team for DAS AUDIT and CHECK as Faulty DAS
was suspected as primary cause of poor performance
Sample case_93704001
*Cell 15014002 & 15014002 shown just for comparison
DL throughput is quite good for the cell. DL coverage and SINR is also
good for the cell.
Sample case_155980003
CELL 15598003 has very low uplink throughput
Sample case_155980003
RLC retransmission rate and packet loss is very high in uplink for this
cell
Sample case_155980003
RSSI is very high at this cell which is impacting throughput and packet loss in uplink. Aggressive
setting of pZeroNominalPusch & pZeroNominalPucch used for this cell but KPI did not improve.
Cell forwarded to o&M team for PIM testing and ruling out internal interference
Essential Features
IRC (interference rejection combining) should be enabled.
Ul FSS (Ul frequency selective scheduling) should be enabled.
Channel Bandwidth (5, 10, 15 and 20) MHz should be equal to
licensed bandwidth.
64-QAM DL should be enabled.
16-QAM UL should be enabled.
Dual Antenna DL Performance Package or MIMO should be
enabled.
All these features should be enabled at eNodeB for Good UL and DL
throughput
Essential Parameters
› Radio Network MO parameters:
– EUtranCellFDD
› dlChannelBandwidth / ulChannelBandwidth=should be equal to license bandwidth
noOfTxAntennas for MIMO= 2
noOfRxAntennas for MIMO= 2
› transmission Mode default is 3, TxDiv and Open Loop Spatial Multiplexing (2x2
MIMO)
› pdcchCfiMode (number of OFDM symbols used for PDCCH, 1->3)
› maximumTransmissionPower = 460 (for 40 watt site)
› partOfSectorPower (for a sector)= 100
› pZeroNominalPucch some UEs need this to be increased or ACK/NACKs are not
received successfully on PUCCH.
› pZeroNominalPusch some UEs need this to be increased from default or lots of
errors seen on PUSCH.
– SectorEquipmentFunction
› confOutputPower = license power(20/40/60 Watt)
counters for throughput analysis
› UL Interference on PUCCH / PUSCH :
pmRadioRecInterferencePwrPUCCH / pmRadioRecInterferencePwr
› SINR of PUCCH / PUSCH: pmSinrPucchDistr / pmSinrPuschDistr
› PRB Utilization in DL / UL : pmPrbUtilDl / pmPrbUtilUl
› RLC ACK/NACK: pmRlcArqUlack|pmRlcArqUlNack
› Transport block PWR restricted: pmRadioTbsPwrRestricted /
pmRadioTbsPwrUNRestricted
› Scheduling activity per cell in UL and DL : pmSchedActivityCellDl
/pmSchedActivityUeDl
› Rank distribution MIMO/ SIMO : pmRadioTxRankDistr
› Number of A2 events(UE in poor coverage) :pmBadCovEvalReport
RSSI
UL Interference on PUCCH* :
UL interference on PUSCH :
pmRadioRecInterferencePwrPUCCH
pmRadioRecInterferencePwr
The measured Noise and Interference Power on PUCCH
The measured Noise and Interference Power on PUSCH
PDF ranges:
[0]: N+I <= -121 PDF ranges:
[1]: -121 < N+I <= -120 [0]: N+I <= -121
[2]: -120 < N+I <= -119 [1]: -121 < N+I <= -120
[3]: -119 < N+I <= -118 [2]: -120 < N+I <= -119
[4]: -118 < N+I <= -117 [3]: -119 < N+I <= -118
[5]: -117 < N+I <= -116 [4]: -118 < N+I <= -117
[6]: -116 < N+I <= -115 [5]: -117 < N+I <= -116
[7]: -115 < N+I <= -114 [6]: -116 < N+I <= -115
[8]: -114< N+I <= -113 [7]: -115 < N+I <= -114
[9]: -113 < N+I <= -112 [8]: -114< N+I <= -113
[10]: -112 < N+I <= -108 [9]: -113 < N+I <= -112
[11]: -108 < N+I <= -104 [10]: -112 < N+I <= -108
[12]: -104 < N+I <= -100 [11]: -108 < N+I <= -104
[13]: -100 < N+I <= -96 [12]: -104 < N+I <= -100
[14]: -96 < N+I <= -92 [13]: -100 < N+I <= -96
[15]: -92 < N+I [14]: -96 < N+I <= -92
Unit: dBm/PRB [15]: -92 < N+I
Unit: dBm/PRB
For good throughput in UL- RSSI value should be low (< -105 dbm)
UL SINR
SINR of PUCCH: SINR for PUSCH:
pmSinrPucchDistr pmSinrPuschDistr
Distribution of the SINR values calculated for P UCCH . Distribution of the SINR values calculated for P USCH .
› PDF ranges: › PDF ranges:
› [0]: SINR <= -15 › [0]: SINR <= -5
› [1]: -15 < SINR <= -12 › [1]: -5 < SINR <= -2
› [2]: -12 < SINR <= -9 › [2]: -2 < SINR <= 2
› [3]: -9 < SINR <= -6 › [3]: 2 < SINR <= 6
› [4]: -6 < SINR <= -3 › [4]: 6 < SINR <= 10
› [5]: -3 < SINR <= 0 › [5]: 10 < SINR <= 14
› [6]: 0 < SINR <= 3 › [6]: 14 < SINR <= 17
› [7]: 3 < SINR › [7]: 17 < SINR
› Unit: dB › Unit: dB
Condition: Each SINR value for PUCCH per UE calculated Condition: Each SINR value for PUSCH per UE calculated
on a TTI basis yields one sample in the distribution. on a TTI basis yields one sample in the distribution.
For good throughput in UL - FEW Samples in poor SINR ranges
PRB utilization
PRB Utilization in UL:
PRB Utilization in Dl : pmPrbUtilUl
pmPrbUtilDl
A distribution that shows the Physical
A distribution that shows the downlink Physical Resource Blocks (PRB) utilization (total
Resource Blocks (PRB) utilization (total number of used PRB by available PRB)
number of used PRB by available PRB) on on the Physical Uplink Shared Channel
the Physical Downlink Shared Channel (PUSCH).
(PDSCH).
PDF ranges:
PDF ranges: [0]: 0 % <= Utilization < 10 %
[0]: 0 % <= Utilization < 10 % [1]: 10 % <= Utilization < 20 %
[1]: 10 % <= Utilization < 20 % [2]: 20 % <= Utilization < 30 %
[2]: 20 % <= Utilization < 30 % [3]: 30 % <= Utilization < 40 %
[3]: 30 % <= Utilization < 40 % [4]: 40 % <= Utilization < 50 %
[5]: 50 % <= Utilization < 60 %
[4]: 40 % <= Utilization < 50 % [6]: 60 % <= Utilization < 70 %
[5]: 50 % <= Utilization < 60 % [7]: 70 % <= Utilization < 80 %
[6]: 60 % <= Utilization < 70 % [8]: 80 % <= Utilization < 90 %
[7]: 70 % <= Utilization < 80 % [9]: 90 % <= Utilization
[8]: 80 % <= Utilization < 90 % Unit: -
[9]: 90 % <= Utilization Condition: One sample should be the
Unit: - utilization during the sample gathering
Condition: One sample should be the utilization period.
during the sample gathering period.
For Good UL and DL throughput- UL and DL PRB utilization should be
low respectively ( <80%)
UL RLC NACK
RLC ACK: RLC NACK:
pmRlcArqUlack pmRlcArqUlNack
› The total number of successful › The total number of unsuccessful
RLC PDU transmissions (ACKs) RLC PDU and RLC PDU
in the uplink direction. segment transmissions (NACKs)
in the uplink direction.
› Unit: -
› Unit: -
› Condition: Continuous
measurement for Radio Bearers › Condition: Continuous
aggregated to cell level.. measurement for Radio Bearers
aggregated to cell level.
For good throughput in UL – UL RLC NACK RATIO should be low
Power restricted
transport block-UL
Transport block PWR restricted: Transport block PWR restricted:
pmRadioTbsPwrUNRestricted
pmRadioTbsPwrRestricted
› The number of Transport Blocks on MAC level
› The number of Transport Blocks on MAC scheduled in uplink where the UE was NOT
level scheduled in uplink where the UE was considered to be power limited.
considered to be power limited. › Unit: -
› Unit: - › Condition: A Transport Block is considered to
› Condition: A Transport Block is considered be power limited when the estimated required
to be power limited when the estimated transmit power is higher than the UE maximum
required transmit power is higher than the transmit power.
UE maximum transmit power.
› Counter type: PEG
› Counter type: PEG
› Scanner: Not included in any predefined
› Scanner: Not included in any predefined scanner
scanner
Ratio of power limited to non power limited blocks should be low
For Good throughput in UL- Ratio should be low(< 50%)
CQI
CQI Is a feedback mechanism from UE to eNB in downlink
– pmRadioUeRepCqiDistr gives CQI distribution in DL
– pmRadioUeRepCqiDistr is a PDF counter and it gives value in range
from 0 to 15
– CQI 1-6 map to QPSK
– CQI 7-9 map to 16QAM
– CQI 10-15 map to 64QAM
For good throughput in DL- Average CQI should be high (>10)
MIMO RANK distribution usage
Rank distribution MIMO/ SIMO : pmRadioTxRankDistr
The transmission mode / rank distributions gives more detailed information on how much each each
transmission mode and rank is used.
PDF ranges:
[0]: Transmit diversity same information both branches
[1]: Open Loop SM Rank 1 SIMO
[2]: Open Loop SM Rank 2 MIMO
[3]: Closed Loop SM rank 1
[4]: Closed Loop SM rank 2
Unit: -
Counter type: PDF
Scanner: Not included in any predefined scanner
SM stands for Spatial Multiplexing
For Good throughput in DL- High samples of MIMO Rank 2
MIMO is used currently used in DL
Ericsson supports TM2(PDF range 1) & TM3(PDF Range 2) currently
Modulation scheme usage DL
› QPSK Samples (%) = 100 * (pmMacHarqDlAckQpsk/
(pmMacHarqDlAckQpsk + pmMacHarqDlAck16qam +
pmMacHarqDlAck64qam))
› 16 QAM Samples (%) = 100 * (pmMacHarqDlAck16qam /
(pmMacHarqDlAckQpsk + pmMacHarqDlAck16qam +
pmMacHarqDlAck64qam))
› 64QAM Samples(%) = 100 * (pmMacHarqDlAck64qam/
(pmMacHarqDlAckQpsk + pmMacHarqDlAck16qam +
pmMacHarqDlAck64qam))
For good DL throughput- High usage of 64 QAM and 16 QAM scheme in DL
Modulation scheme usage UL
QPSK Samples (%) = 100 * ((pmMacHarqUlSuccQpsk/
(pmMacHarqUlSuccQpsk+ PmMacHarqUlSucc16qam ))
16 QAM Samples (%) =100 * ((mMacHarqUlSucc16qam/
(pmMacHarqUlSuccQpsk+ PmMacHarqUlSucc16qam ))
For good UL throughput - High usage of 16 QAM scheme in UL
KEY LEARNING
› Check all important KPI in cases of low throughput as it
make analysis simple and help in finding the root cause
faster.
› Do basic parameter and feature audit before proceeding
further
› Check RSSI at the cell and UL power control parameter
setting.
› Rule out any transmission issue by checking DL latency
and packet loss rate.
›DL throughput is mostly impacted by poor DL SINR. Low
Avg CQI/ 64 QAM sample/RI indicated poor DL SINR.
› UL throughput is mostly impacted by poor UL coverage
check CTR traces & UL SINR counters for this.
