5G Data Transmission

Problem Analysis

Copyright © Huawei Technologies Co., Ltd. All rights reserved.

 Objectives
 Upon completion of this course, you will be able to:
 Know basic principles of 5G NR data transmission.
 Master the method of identifying the reasons for the low rate during 5G NR drive tests.
 Understand the method of optimizing the 5G NR rate.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 2

 Contents

1. Basic Principles of 5G Data Transmission

2. Procedure for Identifying 5G Data Transmission Faults

3. Cause Analysis and Optimization over the Air Interface

4. Typical Cases

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 3

Basic Concepts
 5G NR introduces new technologies and a new architecture upon the LTE system. NR inherits OFDMA, SC-FDMA,
and multiple-antenna technologies used in LTE, and introduces more MIMO layers than LTE. In addition, NR allows
for QPSK, 16QAM, 64QAM, and 256QAM modulation schemes to adapt to air interface quality.

 Both NR and LTE support flexible allocation of time-frequency resources in a specified bandwidth by using
frequency division multiplexing and time division multiplexing. However, unlike LTE, NR can work in both high and
low frequency bands, and allows for multiple subcarrier spacings (SCSs), such as 15 kHz, 30 KHz, 60 kHz, 120 kHz,
and 240 kHz. NR also provides a larger maximum bandwidth in a carrier than LTE, as listed in the following table
(3GPP TR 38.211).

μ SCS Number of Symbols per Slot Number of Slots per Frame Number of Slots per Subframe
0 15 14 10 1
1 30 14 20 2
2 60 14 40 4
3 120 14 80 8
4 240 14 160 16

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 4

Theoretical Peak Rate Calculation
 The NR theoretical peak rate is related to the system bandwidth, modulation scheme, MIMO mode,
and specific parameter settings. For example, in a 100 MHz cell, the quantities of symbols occupied by
the PDCCH in each slot, the broadcast channel, the synchronization channel, and the reference signal
need to be considered during the theoretical peak rate calculation. In addition, the impact of the BLER
over the air interface needs to be considered during actual rate calculation.
 On an NR TDD network: Peak rate = Number of RBs x Number of subcarriers per RB x Number of
symbols per slot x Number of timeslots per subframe x Coding efficiency x Modulation order x Number
of layers x (1 – BLER) x (1 – Control data overhead proportion) x slot configuration
 The theoretical peak rate is about 3 Gbit/s based on the protocol requirements and cell configurations
on the live network, such as the cell bandwidth of 100 MHz (273 RBs), subcarrier spacing of 30 kHz,
DL-UL slot configuration of DDDSU_4:1, self-contained slot configuration of SS2 (10:2:2), typical coding
efficiency of 0.892, a maximum of 8 layers for a single UE, BLER of 10%, and the control data overhead
proportion of 13.6%.
Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 5
 Contents

1. Basic Principles of 5G Data Transmission

2. Procedure for Identifying 5G Data Transmission Faults

3. Cause Analysis and Optimization over the Air Interface

4. Typical Cases

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 6

E2E Factors Affecting 5G Data Transmission (1)

 In-depth understanding of the E2E factors affecting 5G data transmission is the basis for
identifying data transmission faults. You need to learn how different factors affect the
variables (such as the grant count, RB quantity, MCS index, rank value, and BLER) in the 5G
uplink and downlink rate calculation.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 7

E2E Factors Affecting 5G Data Transmission (2)
4) SINR/CQI/RI/SRS 6) Alarm

5) Channel Conditions 7) Parameter

11) QCI/AMBR
8) Channel Calibration APP Server
3) UE Capacity
HSS
RB/Modulation/
Feature gNodeB
Router
BBU
CP X2
E 9) Transmission
Switch MME/UGW Firewall
Packet loss

1) TCP 10) Transmission Packet Loss,

Windows/Thread/RTT/MTU eNodeB BBU OutofOrder/Bandwidth

2) Hardware Performance

Factors 1, 2, 3, 6, 7, 9, 10, 11, and 12 affect the variables such as the grant count and RB quantity, and factors 4, 5, 6, 7,
and 8 affect the variables such as the MCS index, BLER, and rank value.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 8

Counter Observation on the Probe

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 9

 Contents

3. Cause Analysis and Optimization over the Air Interface

3.1 Analysis of Factors Affecting the 5G Air Interface Quality

3.2 Troubleshooting for the Low MCS Index

3.3 Troubleshooting for the High IBLER

3.4 Troubleshooting for the Low Rank

3.5 Troubleshooting for the Low Grant Count and RB Quantity

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 10

5G Rate Impact Analysis (1)
 The 5G single-UE (4T8R) downlink rate reaches the maximum when the following conditions apply:
dual-codeword transmission, rank 8 (rank 4 for 2T4R UEs), MCS index 27 (256QAM), no block errors,
1600 downlink grants (with the downlink-uplink slot configuration of 4:1), and a maximum of 273 RBs
that can be scheduled in the downlink in the 100 MHz bandwidth (with some scheduling TTIs reserved
for common channels)
 The air interface channel quality has the most significant impact on the rate, and is represented by
indicators such as RSRP, SINR, MCS, BLER, and RANK. The impact of these indicators on the rate is
shown in the following figure.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 11

5G Rate Impact Analysis (2)

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 12

5G Rate Impact Analysis (3)
 Common faults related to the air interface channel quality are as follows:
 Low MCS index
 High IBLER
 Low rank
 Low grant count and RB quantity

 Resource scheduling insufficiency may involve RBs or grants. UDP packet injection on the
base station side can be used to determine whether the resource scheduling insufficiency is
caused by a TCP fault or an air interface fault.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 13

5G Air Interface Factor Check (1)
Air Interface Factors Affecting the Rate

Grant count and RB quantity Low MCS index

Conditions for reaching the peak rate: Conditions for reaching the peak rate:
1600 grants in DL and 400 grants in UL MCS index 27 (256QAM) in DL, MCS index 28 (64QAM)
(with the slot configuration of 4:1) and in UL
about 265 RBs in the 100 MHz Possible factors:
bandwidth (1) Test location, in which the environments with SSB
Possible factors: SINR greater than 20 dB, CSI RSRP ranging from –75
(1) Insufficient inbound traffic dBm to –65 dBm, and multipath propagation
(2) AMBR rate limitation (2) Neighboring cell interference, which can be
(3) Missing downlink DCI detection mitigated by deactivating neighboring cells
(4) Problems at the PDCP, RLC, and MAC (3) External interference, which can be mitigated by
layers wideband spectrum scanning
(5) Multi-user scheduling (4) Fixed MCS index
(5) CQI reporting faults

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 14

5G Air Interface Factor Check (2)
Air Interface Factors Affecting the Rate

High IBLER Rank

Conditions for reaching the peak rate: 0% IBLER Conditions for reaching the peak rate: Rank 4 in DL and rank 2 in UL
Possible factors: (for 2T4R); Rank 8 in DL and rank 4 in UL (for 4T8R)
(1) Test location, in which the environments with Possible factors:
SSB SINR greater than 20 dB and CSI RSRP ranging (1) DMRS parameter settings
from –75 dBm to –65 dBm are recommended (2) Channel environment, in which an outdoor area (NLOS) with
(2) Neighboring cell interference, which can be surrounding buildings or trees is selected, but not an open LOS area
mitigated by deactivating neighboring cells (3) Channel calibration results
(3) External interference, which can be mitigated (4) Fixed rank value
by wideband spectrum scanning (5) SRS power insufficiency, which can be solved by gradually increasing
(4) Abnormal MCS convergence the UE transmit power in a step of 3 dB
(5) Abnormal CQI adjustment (6) Uplink interference, which can be checked by the factors that affect
(6) Abnormal weight adaptation the SRS channel quality (Interference Detect on the MAE-Access, which
(7) Downlink frequency offset is described in the channel troubleshooting guide)
(8) Abnormal uplink TA

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 15

 Contents

3. Cause Analysis and Optimization over the Air Interface

3.1 Analysis of Factors Affecting the 5G Air Interface Quality

3.2 Troubleshooting for the Low MCS Index

3.3 Troubleshooting for the High IBLER

3.4 Troubleshooting for the Low Rank

3.5 Troubleshooting for the Low Grant Count and RB Quantity

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 16

Test Location (1)
 Conditions in DL and UL for reaching the peak rate
 MCS index 27 (256QAM) in DL, MCS index 28 (64QAM) in UL

 Requirements for the location with the peak rate test

 Channel quality requirements: CSI-RS SINR greater than 30 dB, CSI RSRP ranging from –75 dBm to –
65 dBm, and RSRP not greater than –65 dBm. If the power received by a UE is overly high, clipping
occurs on the receive component. As a result, the SINR decreases, causing the MCS index and then
the rate to decrease.
 Multipath environment selection: The high rank in the downlink depends on the surrounding
multipath environment. As such, select the area with reflection of surrounding trees and buildings
and avoid a direct path between the test point and base station antenna.
 Neighboring cell control: It is recommended that the test be performed at a point where the SSB
RSRP of the neighboring cell is 6 dB lower than that of the serving cell, to avoid SSB and TRS
interference.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 17

Test Location (2)
 Case
 In a 5G trial, the mechanical downtilt of site 1083 is adjusted from 0° to 6°. A test is performed
at a test point 30 m away from site, and the test result shows that the downlink rate increases.
Before the mechanical downtilt adjustment, the test result shows that the rank value is always 4,
but the MCS index is low, indicating that the interference between beams is large. After the
adjustment, the rank value is changed to 3 and the MCS index is fixed at about 26.

Average
Mechanical MCS
Test Time Downlink Rank CQI SSB SINR SSB RSRP CSI SINR CSI RSRP
Downtilt Index
Rate

December 8 0 474 Mbit/s 3.99 17.2 12.6 34.5 -72 36.5 -71

December 13 6 654 Mbit/s 3 26.6 11.88 33.8 -64.4 33.5 -67.9

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 18

Interference (1)
 In peak rate tests, the DMRS SINR must be greater than 30 dB to ensure that the MCS index reaches 27
and the IBLER approximates 0%.
 When interference signals exist in a cell, the uplink and downlink services of the cell are affected. As a
result, faults such as access failure, service drop, and low rate occur. If the RSRP is high but the SINR is
low (for example, average RSRP>= –80 dBm, average SINR <= 15 dB) and the MCS index is low, start
the interference troubleshooting.
 Interference includes but is not limited to neighboring cell interference, overshoot coverage
interference, and external interference.
 Neighboring cell interference is caused by multiple strong neighboring cells. You can deactivate the neighboring
cells to check whether the problem is solved.
 Overshoot coverage interference mainly refers to a situation where the coverage of cells far away from the
serving cell falls into the serving cell coverage.
 External interference can be solved by spectrum scanning. Then, start the interference troubleshooting.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 19

Interference (2)
 Case
 Symptom: In an FOA site, both NR 2.6 GHz cells and LTE TDD 2.6 GHz cells work on 2.6 GHz. After the site is
deployed, the interference is severe and the uplink and downlink rates are low. When NR and LTE TDD work on
the same frequency band, their frame structures must be consistent. The interference can be mitigated only when
the frame structures of NR and LTE TDD are the same, meaning that the NR frame length plus 3 ms is used.
 Solution: When the special subframe configuration of band D on the LTE side is SSP7 (10:2:2), the corresponding
subframe configuration on the NR side is SS54 (6:4:4). When 2.6 GHz cells are added on the NR side, set the frame
structure to 8:2 and the special subframe configuration to SS54.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 20

Abnormal CQI Reporting
 SRS exception
 In massive MIMO, the reciprocity between uplink and downlink channels is implemented through
the SRS. The base station delivers CSI measurement only after receiving the SRS reported by the UE.
Then, the UE reports CSI results (CQI/PMI/RI). The network side performs scheduling based on the
CQI reported by the UE. In other cases, the network side cannot perform downlink data scheduling.
Therefore, the SRS has a significant impact on CSI measurements, and an SRS exception will cause
abnormal CQI reporting.
 CSI exception
 If the number of ports configured on the base station side is inconsistent with the number of ports
on the test UE, the CSI/CQI measurement is abnormal. As a result, the UE cannot report the CQI,
and the MCS index and rank value are low because the base station cannot receive the CQI.
 Configure the number of CSI-RS ports: four ports for the CPE, eight ports for the 4T8R UEs

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 21

Question
 What are the maximum MCS indexes for DL
256QAM and UL 64QAM, respectively?

 A. MCS index 27 (256QAM)

 B. MCS index 28 (64QAM)
 C. MCS index 28 (256QAM)
 D. MCS index 27 (64QAM)

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 22

 Contents

3. Cause Analysis and Optimization over the Air Interface

3.1 Analysis of Factors Affecting the 5G Air Interface Quality

3.2 Troubleshooting for the Low MCS Index

3.3 Troubleshooting for the High IBLER

3.4 Troubleshooting for the Low Rank

3.5 Troubleshooting for the Low Grant Count and RB Quantity

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 23

Abnormal MCS Convergence (1)
 Similar to troubleshooting for the low MCS index, troubleshooting for the high IBLER focus
on changes in the air interface, such as poor coverage, strong neighboring cell interference,
and external interference. If the SINR is abnormal, such as a sharp decrease or fluctuation,
block errors are subject to unstable channel conditions.
 Downlink/Uplink IBLER
 Initial BLER (%): It is the ratio of the number of blocks incorrectly transmitted at initial transmission
to the number of blocks effectively transmitted. It is recommended that the IBLER be 0% in fixed-
point peak rate tests and 10% in field mobility tests.
 Residual BLER (%): It is the ratio of the number of blocks incorrectly transmitted after the last
transmission to the number of blocks effectively transmitted. In normal cases, residual BLER is far
less than initial BLER.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 24

Abnormal MCS Convergence (2)
 If the SINR is normal, favorable radio conditions exist in the current system. A high BLER may be
caused by the MCS index selection algorithm. If the MCS index is too high, the IBLER does not
converge. If the IBLER is normal (the target value is 10%), it must be stable at about 10%.
 If the IBLER exceeds 10% for a long time, it does not decrease. Check whether the MCS index decreases to 0. If
not, the MCS index adjustment on the base station side is abnormal.
 If the IBLER is lower than 10% for a long time, it does not increase. Check whether the MCS index increases to
27. If not, the MCS index adjustment on the base station side is abnormal.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 25

Abnormal Uplink TA
 With uplink timing, the gNodeB requires that the signals from different UEs in the same
subframe but with different frequency domain resources arrive at the gNodeB at basically
the same time. This ensures orthogonality of the uplink transmission and avoids intra-cell
interference.
 TA = 0.51 ns * 512*3*10^8 /2 = 39 m, which is a half of LTE.
 Downlink block errors may also be caused by uplink out-of-synchronization. Therefore, when
a high BLER occurs, you can also check the TA for any drastic TA adjustment.
 TA problems occur in mobility scenarios. TA adjustment is generally not performed at a fixed
point or near point. If a large number of TA adjustments occur, troubleshoot the fault.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 26

Question
 What is the maximum number of uplink
and downlink data streams for a single UE?
 A. 8 streams in DL and 4 streams in UL
 B. 16 streams in DL and 8 streams in UL

 What are the differences between IBLER

and RBLER?

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 27

 Contents

3. Cause Analysis and Optimization over the Air Interface

3.1 Analysis of Factors Affecting the 5G Air Interface Quality

3.2 Troubleshooting for the Low MCS Index

3.3 Troubleshooting for the High IBLER

3.4 Troubleshooting for the Low Rank

3.5 Troubleshooting for the Low Grant Count and RB Quantity

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 28

Weight Adaptation Algorithm (1)
 Adaptation between SRS-based weights and PMI-based weights allows beamforming weight selection
based on SRS or PMI. The SRS-based weight selection applies to UEs with a large SRS SINR, and the
PMI-based weight selection applies to UEs with a small SRS SINR. PMI-based weights are more
accurate than SRS-based weights for cell edge users, which improves their SINRs to positively affect the
data rate.
 When the uplink SRS SINR of a UE is greater than the value of ThSRS (–2 dB by default), SRS-based weights are
selected.
 When the uplink SRS SINR of a UE is less than the value of ThPMI (–8 dB by default), PMI-based weights are
selected.
 When the SRS SINR of a UE is between the value of ThSRS and that of ThPMI, the weights remain unchanged.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. ThPMI Page 29
Weight Adaptation Algorithm (2)
 When the SRS-based weight is used, the base station uses the rank adaptation algorithm to
determine the final rank value to be used.
 When the SRS SINR is less than the value of ThPMI, the PMI-based weight is used and the
rank adaptation algorithm does not take effect. The RI reported by the UE is directly used.
 DFT-based weights are used when the PMI is not reported, a handover is just performed, or
channel calibration fails. The rank value is selected based on the RI reported by the UE
according to the following rules:
 If RI in the CSI reported by the UE is 1, rank 1 is used.
 If RI in the CSI reported by the UE is 2 or 3, rank 2 is used.
 If RI in the CSI reported by the UE ranges from 4 to 8, rank 4 is used.
 Adaptation between SRS-based weights and PMI-based weights needs to be enabled on the
base station.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 30

Terminal Capabilities and Related Parameters (1)
 Factors affecting the maximum number of downlink layers for a single UE:
Min(number of gNodeB transmit antennas, number of UE receive antennas),
maximum number of layers supported by a cell, and protocols-defined
maximum eight downlink layers for a single UE
 Factors affecting the maximum number of uplink layers for a single UE:
Min(number of gNodeB receive antennas, number of UE transmit antennas),
maximum number of layers supported by a cell, protocols-defined maximum
four uplink layers for a single UE

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 31

Terminal Capabilities and Related Parameters (2)

Maximum Maximum Maximum Maximum Maximum

Maximum
Number of Number of Number of Number of Number of
Number of
Downlink Downlink Downlink Uplink Layers Uplink
gNodeB UE Uplink Layers
Layers for a Multiplexing Layers for a for a Single Multiplexing
for a Single UE
Single UE in Layers in MU- Single UE in UE in SU- Layers in MU-
in MU-MIMO
SU-MIMO MIMO MU-MIMO MIMO MIMO
64T64R 2T4R 4 16 4 2 8 2
64T64R 4T8R 8 16 4 4 8 4
32T32R 2T4R 4 16 4 2 8 2
32T32R 4T8R 8 16 4 4 8 4
8T8R 2T4R 4 8 4 2 4 2
8T8R 4T8R 8 8 4 4 4 4

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 32

Terminal Capabilities and Related Parameters (3)
 Turning on the SU-MIMO Multiple Layers Switch
 Maximum MIMO Layer Number Quota
 Downlink: LAYER_8(8 Layers) for 8T8R scenarios; LAYER_16(16 Layers) for other scenarios
 Uplink: LAYER_4(4 Layers)
 DMRS configuration: The DMRS type and the number of symbols pose restrictions on the
maximum rank for a single UE. The following table lists the maximum rank for a single UE
in different configurations.
Uplink/Downlink DMRS Type DMRS Length Maximum Rank for a Single UE
Type1 1 4
Type1 2 8
Downlink
Type2 1 6
Type2 2 12
Type1 1 4
Type1 2 8
Uplink
Type2 1 6
Type2 2 12

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 33

Alarm Handling and Channel Calibration Failure
 For details about how to handle alarms, see Operation and Maintenance > Fault
Management > 3900 & 5900 Series Base Station Alarm Reference in the product
documentation.
 Run the DSP NRLOCELLCHNCALIB command on the NR base station side to query the
channel calibration result. If the channel calibration fails, run the STR NRLOCELLCHNCALIB
command to perform the manual calibration.
 Once channel calibration fails, the system cannot accurately evaluate the SRS-based weights
and uses the DFT-based weights by default to initiate services. The gNodeB selects a rank
based on the RI reported by the UE according to the following rules:
 If RI in the CSI reported by the UE is 1, rank 1 is used.
 If RI in the CSI reported by the UE is 2 or 3, rank 2 is used.
 If RI in the CSI reported by the UE ranges from 4 to 8, rank 4 is used.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 34

Frequent Handovers/Switching
 Frequent UE handovers between cells
 The low rank and low MCS index ensure the handover success rate rank and the initial access success rate after
a handover. After a handover, the initial rank value is 1 by default and can be changed back to the original value
after about 30 ms, which has a small impact on network performance. However, if frequent UE handovers occur,
the rank value cannot quickly increase. In this case, the areas with frequent handovers need to be optimized.
 The NRDUCellRsvdParam.RsvdU8Param40 parameter indicates the initial rank value used after UEs are handed
over. The initial rank value equals the parameter value plus 1. For example, the default parameter value 0
indicates that the initial rank value equals 1. The valid value range of this parameter is from 0 to 7.
 Frequent switching between MCS index tables
 In MCS index table switching, the MCS index tables of 64QAM and 256QAM are frequently switched under given
conditions. The rank value is fixed at 1 for scheduling during switching.
 The DL_MCS_TABLE_ADAPT_SW option of the NRDUCellPdsch.DlLinkAdaptAlgoSwitch parameter determines
whether to enable MCS index table adaptation. Enabling this function is defaulted and recommended. When the
SINR fluctuates greatly, the MCS index table is likely to switch frequently. Therefore, check the impact of the
switching on the rank value.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 35

Radio Environments
 Test experience summary
 The energy of each path for points close to the base station is strong. Therefore, the rank
is high.
 A high rank is difficult to be achieved in open areas.
 In open areas, there is no refraction and reflection from objects such as buildings, and multipath
is unavailable. Therefore, RF adjustment can be performed to construct the refraction and
reflection from objects such as the ground and buildings to create multiple paths and improve
the rank value.
 According to the test experience, in open scenarios, if the mechanical downtilt of the AAU is set
to a small value, it is difficult to achieve a high rank and high MCS index. In this case, increase
the mechanical downtilt of the cell to construct reflection from the ground to improve the rank
value and MCS index.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 36

Question
 What are the types of weights for beamforming?
What are their application scenarios?

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 37

 Contents

3. Cause Analysis and Optimization over the Air Interface

3.1 Analysis of Factors Affecting the 5G Air Interface Quality

3.2 Troubleshooting for the Low MCS Index

3.3 Troubleshooting for the High IBLER

3.4 Troubleshooting for the Low Rank

3.5 Troubleshooting for the Low Grant Count and RB Quantity

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 38

AMBR Rate Limitation
 Important UE registration information on the core network: AMBR and QCI
 The AMBR limits the non-GBR of a UE.
 The QCI information of a UE is associated with the QCI-level PDCP and RLC parameters (including
the number of SN bits and RLC mode) on the base station side. As a result, the UE throughput is
affected.
 The UE-specific AMBR and QCI can be viewed in the S1AP_INITIAL_CONTEXT_SETUP_REQ message
traced over the S1 interface or the SgNB_Add_Req message traced over the X2 interface.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 39

Missing DCI Detection
 Check the RSRP value of CSI to determine whether missing DCI detection is caused by poor
coverage. Check the setting of NRDUCellRsvdParam.RsvdU8Param7 (PDCCH aggregation
level configured for UEs in a cell). If this parameter is set to 0, the adaptive aggregation level
is used. If this parameter is set to 1, the aggregation level is always 2. If this parameter is set
to 2, the aggregation level is always 4. If this parameter is set to 3, the aggregation level is
always 8. If this parameter is set to 4, the aggregation level is always 16. If this parameter is
set to other values, the adaptive aggregation level is used. An excessively low aggregation
level will cause DCI missing detection. The adaptive aggregation level is recommended.
Alternatively, the aggregation level 3 or 4 is used to perform the test again to check whether
the problem is resolved.
 Insufficient DCI resources
 A large number of scheduled UEs

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 40

Problems at the PDCP, RLC, and MAC Layers (1)
Protocol Layer Impact Factor Analysis Method
Insufficient ingress traffic
Packet loss due to full
Analyze the traffic, packet loss due to full buffer, and packet loss due to
buffer
timeout based on cell DT tracing results.
Packet loss due to
timeout
PDCP The SN length is specified by the QCI-level DlPdcpSnSize parameter. In
SCG scenarios, the PDCP layer is on the NR side. It is recommended that
the PDCP SN length on the NR side be the same as the RLC SN length,
Insufficient PDCP SN which is 18 bits.
length
The RLC SN length is 18 bits by default. If the PDCP SN length is set to 12
bits, the RX side of the PDCP may experience the hyper frame number
asynchronization.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 41

Problems at the PDCP, RLC, and MAC Layers (2)
Protocol Layer Impact Factor Analysis Method
They are specified by the QCI-level DlRlcSnSize and RlcMode
RLC SN+RLC mode parameters, respectively. Their respective recommended values are 18 bits
and AM.
Check whether the inbound PDCP packets are stable or inbound traffic is
RLC ingress traffic
insufficient
RLC
Number of RLC
retransmission times
Scheduling stopped due to Analysis based on the tracing on the base station side
full buffer in the RLC
window
HARQ resource allocation failure is common. Determine whether the
fault is caused by abnormal ACK feedback or abnormal scheduling
HARQ process exhaustion
MAC process. Abnormal feedback indicates that no HARQ feedbacks are
provided or there are a large number of HARQ feedbacks with DTX.
Missing DCI detection Check whether there are a great number of HARQ feedbacks with DTX.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 42

Question
 When the SCS is 30 kHz and the subframe
configuration is 8:2, how many scheduling times
are planned within 1s? What are the respective
values in the downlink and uplink?

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 43

 Contents

1. Basic Principles of 5G Data Transmission

2. Procedure for Identifying 5G Data Transmission Faults

3. Cause Analysis and Optimization over the Air Interface

4. Typical Cases

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 44

Case 1: Low Rate Due to a Low Rank in Open
Areas (1)
 Symptom
 A CPE is used to perform packet injection tests in two cells served by the
base station. The packet injection is normal, but the rate in one cell is only
400+ Mbit/s.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 45

Case 1: Low Rate Due to a Low Rank in Open
Areas (2)
 Analysis
 The analysis results of configuration data show that all parameter settings are normal.
 The analysis results of operation logs show that no operation is performed when the problem occurs.
 The analysis results of alarm logs show that no rate-related alarm is reported.
 According to the logs on the Probe, the number of times the base station schedules the UE, RB quantity, and
MCS index are all close to the maximum. However, the rank value is always 2 and doe not increase to 4. As a
result, the rate is low. However, such test in other cells is normal. It is suspected that the problem is caused by
the test environment.
 According to the analysis of the onsite test environment, the correlation between multiple channels of signals is
too high due to LOS propagation to the antenna (meaning no obstacle around the antenna), and the rank
value cannot increase to 4. After the test location is changed (under a tree), the rank value increases to 4 and
the rate meets the requirements.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 46

Case 1: Low Rate Due to a Low Rank in Open
Areas (3)
 Conclusion
 It is confirmed that there is no obstacle between the CPE
and the base station antenna, the correlation between
multiple channels of signals is too high, and the rank
value cannot increase to 4. As a result, the rate is low.

 Solution
 The low rank is caused by the inappropriate test
location. Therefore, change the test location (for
example, under the tree in this example) where there
are obstacles around the antenna and the SINR is
favorable (greater than 28 dB). The rank value increases
to 4.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 47

Case 2: Low Rate Due to Insufficient Downlink
Grants (1)
 Symptom
 During the drive test of an operator, it is found that the number of times the base station schedules
UEs in the 60 MHz cells does not exceed 1350 per second, which should be 1600 per second in
normal cases. DL grants (left) and PCI (right)

 Analysis
 TTI data analysis results show that system information (SIB1) subframe contains a large number of
error codes related to CCE allocation failures.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 48

Case 2: Low Rate Due to Insufficient Downlink
Grants (2)
 The beam time-domain position of SIB1 is as follows:
 When there are seven SSB beams, SIB1 occupies slots 10 to 17.
 When there are four SSB beams, SIB1 occupies slots 10 to 13.

 Reason for CCE allocation failures: In RAN2.1, SIB1 is introduced in NSA mode. When data services and
SIB1 are scheduled at the same time, CCE resources are insufficient, causing the scheduling to fail.
According to the PDCCH tracing result, SIB1 is located in slots 10 to 17 of an even-numbered frame,
and the common DCI aggregation level is 8 (48/6 = 8). For details, see the table on the next page.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 49

Case 2: Low Rate Due to Insufficient Downlink
Grants (3)

 If a slot may be used to transmit uplink DCI, some CCEs are reserved for the uplink DCI transmission.
The reservation percentage is specified by NRDUCellPdcch.UlMaxCcePct parameter, which is 50% by
default. According to the current specifications of CPE 1.0, a maximum of 96 CCE RBs can be processed
in one PDCCH symbol, meaning that 48 (96 x 50%) CCE RBs are reserved for downlink scheduling.
However, the common DCI in SIB1 always occupies 48 CCE RBs (aggregation level 8 x 6 RBs per CCE).
As a result, no downlink CCE is available for user data scheduling, and downlink scheduling fails.
Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 50
Case 2: Low Rate Due to Insufficient Downlink
Grants (4)
 According to verification on the live network, changing the value of NRDUCellPdcch.UlMaxCcePct to
33 (that is, the CCE reservation percentage for uplink scheduling is 33%) can prevent scheduling
failures in single-user scenarios.
 In addition, run the following command with related parameter settings so that the number of times
the base station schedules UEs reaches 1600 per second in the areas where the problem occurs
 MOD NRDUCELLPDSCH: NrDuCellId=0, RateMatchSwitch=PDCCH_RATEMATCH_SW-1; MOD NRDUCELLPDCCH:
NrDuCellId=0, OccupiedRbNum=2;

 Principle analysis: In RAN2.1, when PDCCH rate matching is enabled (that is, the PDCCH and PDSCH share
symbols) and the number of CCE RBs occupied by UEs is configured (Actual effective value = Configured value x
12), the allocated CCEs that are in the CCE range for public resource scheduling are not counted in the
measurement of the uplink-downlink CCE ratio. Therefore, when a slot is used to transmit SIB messages, the
downlink data scheduling in the slot can also succeed.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 51

Case 2: Low Rate Due to Insufficient Downlink
Grants (5)
 Conclusion
 In versions later than RAN2.1, SIB messages are also sent (required by ANR) in NSA networking. In addition, the
uplink-downlink CCE ratio is 50% by default. As CPE 1.0 has a limit on the maximum number of CCE RBs, CCEs
fail to be allocated in the slot for transmitting SIB messages. As a result, downlink data scheduling is triggered.
 The degree of impacts of CCE allocation failures on the downlink throughput is related to the subframe
configuration and the number of SSB beams. For example, the downlink throughput decreases by 15.6% when
the subframe configuration is 8:2 and seven SSB beams are used.

 Solution
 Change the uplink-downlink CCE ratio (it has no impact on a single UE with downlink-only services but affects
uplink scheduling in other scenarios).
 Final solution: The uplink-downlink CCE ratio adaptation function will be supported in later versions.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 52

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (1)
 Symptom
 After the NR version of an operator is upgraded, the fixed-point test results of the 4T8R UE show
that the downlink throughput decreases sharply and then remains low. The rank value used for
scheduling decreases from 6 to 4.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 53

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (2)
 Analysis
 According to the statistics of other associated counters on the Probe, the number of downlink
grants per second and the number of scheduled RBs per second are sufficient.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 54

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (3)
 The average MCS index and IBLER also deteriorate. According to the deterioration amplitude
of the downlink throughput, the deterioration in the MCS index and IBLER is not the main
cause for the problem.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 55

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (4)
 The UE-measured CSI RSRP and SINR values are normal, and the rank value (PCC Measured
RI) is also normal.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 56

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (5)
 It can be determined that the decrease in the rank value is caused by the base station
instead of the UE or channel conditions. The cause of the sharp decrease in the rank value
needs to be further analyzed. Run the DSP NRDUCELLCHNCALIB command on the base
station side to query the channel calibration result. It is found that channel calibration fails
at the corresponding time.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 57

Case 3: Downlink Rank Value Remained Low Due to
the AAU Channel Calibration Failure (6)
 Conclusion
 Check how the rank adaptation function works. When channel calibration fails, uplink and downlink channel
reciprocity is not supported. In this case, only DFT-based weights can be used for scheduling, but rank
adaptation does not take effect when DFT-based weights are used.
 When rank adaptation does not take effect, the maximum rank value is 4 if channel calibration fails in a cell
with eight CSI ports, and the maximum rank value is 2 if channel calibration fails in a cell with four CSI ports.
The base station initiates channel calibration every 30 minutes. If strong interference occurs during the
calibration (the uplink full power test is performed in the center of the neighboring cell at the time when the
problem occurs), the channel calibration will fail. The channel calibration failure affects only the air interface
performance and is not a function or component problem. Therefore, the alarm reporting is not considered into
the product implementation.

 Solution
 If the channel calibration failure is caused by strong external interference, check for external interference.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 58

 Summary
 In this course, we have learned the following contents:
 Understand the 5G NR data transmission principles and learn how to evaluate the 5G
peak rate.
 Master how to identify the reasons for the low rate during 5G NR drive tests, understand
the factors that affect the rate, and learn common troubleshooting methods.
 In addition, this course describes only the factors that affect the rate on the air interface
side. The factors that affect the rate on the transmission side, core network, and cloud
side are beyond the scope of this course.

Copyright © Huawei Technologies Co., Ltd. All rights reserved. Page 59

Thank You
www.huawei.com