Product Name Security Level

NPI INTERNAL
Product Version Total pages 38

V200R006

Root Cause Tree-based LTE Throughput

Optimization Technical Guide

Prepared By Guo Yufei 133600 Date 2014-12-28

Reviewed By He Lingling 105935 Date
Zhu Yijiong 177769
Qiu Yongping 00144639 2015-1-27
Cao Fei 206174, Yu Haili 225359, and
Zhang Yejun 210897
Granted by Date

Huawei Technologies CO., LTD.

All Rights Reserved
 Change History

Data Revision Change Description Author

Version

2014-12-28 V0.90 Completed the draft. Guo Yufei 133600

2015-1-27 V1.00 Modified the document according to Guo Yufei 133600
review comments.
 Contents

1 Overview...........................................................................................................................................7
1.1 Background Information and Benefits...............................................................................................................7
1.2 Restrictions and Requirements..........................................................................................................................7
1.2.1 Specifications............................................................................................................................................7
1.2.2 Application Scenarios...............................................................................................................................8
1.3 Maturity.............................................................................................................................................................8

2 Technical Principles......................................................................................................................9
2.1 Throughput Evaluation....................................................................................................................................10
2.2 Low Throughput Problem Isolation and Demarcation....................................................................................11
2.3 Air Interface Root Cause Location for Low Throughput.................................................................................14
2.4 Throughput Optimization Suggestions............................................................................................................18

3 Procedures.....................................................................................................................................22
4 Guidelines.....................................................................................................................................22
4.1 Overview..........................................................................................................................................................22
4.2 Tools.................................................................................................................................................................22
4.3 Data Collection................................................................................................................................................22
4.4 Solution Application Instructions....................................................................................................................22
4.4.1 Remarks on Parameter Settings..............................................................................................................23
4.4.2 Description of the LTE Throughput Optimization Report......................................................................25
4.4.3 Suggestion Descriptions and Implementation........................................................................................29

5 Cases...............................................................................................................................................30
5.1 Case 1: Routine Optimization in the XX Office of the XX Operator..............................................................30
5.1.1 Background.............................................................................................................................................30
5.1.2 Comparison Perspectives........................................................................................................................30
5.1.3 Comparison Results................................................................................................................................30
 Tables
Table 1-1 Tool specifications...................................................................................................................................6

Table 1-2 Application scenarios..............................................................................................................................7

Table 2-1 Counters for evaluation...........................................................................................................................9

Table 2-2 Root cause tree of LTE throughput problems and problem resolving capability..................................11

Table 2-3 Mapping between root causes and weak coverage problems causing low LTE throughput.................14

Table 2-4 Root causes of intra-frequency interference in the downlink...............................................................15

Table 2-5 Fields in the detailed interference analysis...........................................................................................16

Table 2-6 Optimization suggestions......................................................................................................................17

Table 4-1 Data sources..........................................................................................................................................21

Table 4-2 Settings of threshold related parameters...............................................................................................22

Table 4-3 Optimization suggestions on non-air interface root causes...................................................................26

Table 4-4 Optimization suggestions on air interface root causes..........................................................................26

Table 5-1 Location capability comparison............................................................................................................30

 Root Cause Tree-based LTE Throughput
Optimization Technical Guide
Keywords:
LTE, downlink throughput, accurate optimization

Abstract
This document provides the technical principles and application guidelines of the root cause tree-based LTE
throughput optimization solution for network optimization engineers.

List of Abbreviations

Abbreviation Full Spelling

LTE Long Term Evolution

MR measurement report
CHR call history record
SGW serving gateway
MME mobility management entity
1 Overview
This document describes the technical principles and application guidelines of the root cause tree-based
LTE throughput optimization solution for network optimization engineers.

1.1 Background Information and Benefits

Background
LTE network optimization involves the following difficulties:
 High delivery costs: Network optimization experience is not consolidated or standardized. Network
optimization delivery requires a large number of intermediate and senior personnel, causing a high
labor cost. The major optimization method is still drive test, which increases the cost of vehicles,
equipment, and labor.
 Low optimization efficiency: Various optimization tools with different functions are not integrated to a
platform for efficient batch service delivery. Current optimization tools, including foreground tools
Probe and Assistant, and background tools Nastar, NPmastar, OMstar and SmartRNO, provide
different functions and use different data formats, causing inconvenience for use.
 Coarse parameter settings: Parameters are set based on templates, limiting optimization methods. As a
result, optimizations are not pertinent and service supports are not competitive.
 Inefficient problem solving: Background traffic statistics analyses cannot identify root causes and
drive tests can only identify problems that occur on the road, leaving many problems of intensive
coverage undetected.
Benefits
The root cause tree-based LTE throughput optimization solution is designed. In this solution, data from
many sources (except drive test data) and in-depth analysis algorithms are used to identify all root causes of
throughput problems and provide specific and quantitative optimization suggestions, ensuring that
throughput problems are resolved quickly and effectively. The solution also solidifies network optimization
experiences and integrates optimization tools to increase analysis efficiency, reducing requirements on
optimization skills of engineers and lowering labor cost. Geographic display of problems will also be
provided, visualizing problem areas and providing guidance for radio frequency (RF) optimization.

1.2 Restrictions and Requirements

This solution does not apply to scenarios with discontinuous LTE coverage or super-distance coverage,
high-speed railway scenarios, or highway scenarios.

1.2.1 Specifications

Table 1.2.1.1.1.1.1.1.1 Tool specifications

Item Description Maximum Specification In Use
SIG (MR) data 8G 20 GB
Traffic statistics 500 MB 2 GB
Configuration files after 20 MB 50 MB
compression
Number of cells 100 300
 Item Description Maximum Specification In Use
Data collection period Four busy hours per day in two Two busy hours per day in thee
consecutive days consecutive days

1.2.2 Application Scenarios

Table 1.1.1.1.1.1.1.1.2 Application scenarios

Scenario Type Restriction Preventive
Measure
NE version Supported NE versions: BTS3900 V100R008C00, V100R008C01, None
V100R009C00, V100R010C00, and V100R010C10
These NE versions are supported in both LTE FDD and LTE TDD
networks.

Requirements for application:

1. Between data collection completion and optimization suggestion implementation, do not modify the
network, such as PCI adjustment, parameter modification, and RF adjustment. If you perform such an
operation, you need to re-collect data for analysis.
2. High-speed railway, highway, metro, and indoor coverage are not supported. Typical scenarios are
urban areas with continuous coverage.
3. The network transmission must be stable. Otherwise, decompression of some MR files may fail.
Analysis results are incorrect if more than 10% of MR files fail to be decompressed.
1. If the optimization suggestions cannot be implemented completely, the network performance
improvement may be affected.

1.3 Maturity
The solution is verified in terms of data source analysis of all topics, usability, tool efficiency, and basic
algorithm implementation in lab tests. It is also verified in terms of performance gains, application
scenarios, and efficiency increase in the beta test in Hangzhou. The results show that the solution meets the
requirements for commercial use and is recommended for commercial use.

Optimization Maturity Description

Suggestion
Classification
Parameters 1. In outdoor scenarios where Pa is not set to –3 and Pb is not set to 1, set Pa to –3 and
Pb to 1. In indoor scenarios where Pa and Pb are not set to 0, set Pa and Pb to 0.
2. Whether to set Pa and Pb to 0 depends on network requirements.
For cells where the RS power is lower than 62 W or higher than 122 W, set Pa to –3 and
Pb to 1.

Features 1. Determine the basis for triggering MLB and the parameter settings based on the on-
 site camping strategy.
3. Determine whether to disable discontinuous reception (DRX) based on network
conditions and the auxiliary test unit (ATU) policy.
4. Closed-loop MIMO improves LTE FDD network performance. Enable this feature in
LTE TDD networks only when required.

Neighboring cells The optimization suggestions on missing configuration of neighboring cells can be
directly implemented. For neighboring indoor cells, perform optimization manually based
on network requirements.
PCI No target PCIs are provided currently. Perform these optimization suggestions on cells
with strong interference and correlated cells at the same time based on interference
analysis results.
Antennas Specific data of antenna adjustment cannot be provided currently. Field engineers need to
determine the angles for adjustment based on the root causes and network requirements.

NOTE
When implementing PCI optimization suggestions, if no optimization results can be determined using tools on site, you
can decrease coverage overlap to reduce network interference. Optimization methods include downtilt and RS power
adjustment.

2 Technical Principles
The general idea of this solution is to identify low throughput cells and to isolate and demarcate all
problems causing low throughput. In this way, key factors and air interface problems affecting throughput
can be identified. Targeted optimization suggestions can be obtained based on the identification results and
taken up to resolve the low throughput problems. The fault scope is narrowed down by identifying low
throughput cells and demarcating low throughput problems for these cells. In-depth analyses on air
interface problems are also carried out. In this way, targeted optimization suggestions can be determined for
air interface problems and non-air-interface problems.
2.1 Throughput Evaluation
LTE throughput KPIs, which are average throughput and average throughput per UE, evaluate network
performance in terms of traffic and user perception. In this document, cell-specific average downlink
throughput and single UE throughput are measured to evaluate grids and identify top cells.

Table 2.1.1.1.1.1.1.1.1 Counters for evaluation

Counter ID Counter Name Description
1526728261 L.Thrp.bits.DL Total downlink traffic volume for the PDCP SDUs in a
cell
1526729005 L.Thrp.bits.DL.LastTTI Downlink traffic volume for PDCP SDUs sent in the
last TTI before the buffer becomes empty
1526728262 L.Thrp.Time.DL Total duration for transmitting downlink PDCP DSUs
in a cell
1526729015 L.Thrp.Time.DL.RmvLastTTI Data transmit duration except the last TTI before the
downlink buffer becomes empty
1526728232 L.Thrp.Time.Cell.DL Total duration of downlink data transmission in a cell
1526728997 L.Thrp.Time.Cell.DL.HighPrecision Total duration (with the precision of 1 millisecond) of
downlink data transmission in a cell
1526726737 L.ChMeas.PRB.UL.Used.Avg Average number of used uplink PRBs
1526726740 L.ChMeas.PRB.DL.Used.Avg Average number of used physical downlink shared
channel (PDSCH) PRBs
1526728969 L.Traffic.ActiveUser.DL.Avg Average number of active UEs in the downlink
 Cell throughput is calculated using the following formula:
Average downlink throughput (high-precision) (Mbit/s) =
L.Thrp.bits.DL/1024/1024/L.Thrp.Time.Cell.DL.HighPrecision/1000
Downlink throughput per UE (kbit/s) = (L.Thrp.bits.DL/1024)/(L.Thrp.Time.DL/1000)
Sin eRAN6.1, the following formula is used to calculate the downlink bearer throughput per UE to comply
with the protocol:
Downlink throughput per UE (except the last TTI) = ((L.Thrp.bits.DL-
L.Thrp.bits.DL.LastTTI)/1024)/((L.Thrp.Time.DL.RmvLastTTI)/1000)
Downlink throughput per UE indicates the downlink transmission rate of each data radio bearer (DRB) in a
measurement period. If each UE has only one DRB, downlink DRB per UE indicates the data transmission
rate of each UE.
Analyses on low throughput problems are performed based on the average number of UEs in a cell, average
downlink CQI, average downlink MCS, rank 2 proportion, PDSCH initial block error rate (IBLER), and
block error rate (BLER). At the same time, cell performance is evaluated using KPIs including the access
success rate, handover success rate, and service drop rate, and UE service models are determined based on
the proportion of light traffic services.

2.2 Low Throughput Problem Isolation and Demarcation

Many factors are affecting throughput. To resolve throughput problems, the QCI allocated by the MME,
limited transmission bandwidth, packet loss and long delay in transmission, deterioration of NE
performance, and poor transmission capability of air interfaces can be analyzed and located currently.

Problem isolation and demarcation narrows down the fault scope of a low throughput cell to a certain NE
or port based on traffic statistics, external CHRs, and alarms. The imported alarm data is compared with the
 critical alarm list preset in the solution, which includes transmission, antenna, and NE alarms. If any listed
critical alarm is detected, a prompt message is displayed, requesting a rectification.
 Packet loss or long delay in transmission are detected through IP performance monitoring (IPPM) and
GPRS tunneling protocol-user plane (GTPU) measurement. IPPM and BSTWAMP measurement
requires coordination of Huawei CN products (PS9.2 or later). IPPM is not enabled by default. If
IPPM is enabled, the transmission load increases. Communicate with frontline personnel before
enabling IPPM.
 UE problems include insufficient UE capability, slow SIM card registration, and QCI allocation
exceptions. These problems can be determined based on UE capability information, aggregate
maximum bit rate (AMBR), and QCI in CHRs.
 Cells with severe uplink interference can be identified by checking the uplink interference level of
antennas. You can adjust interference related thresholds based on local spectrum scanning data.
 Cell capacity related problems include long PDCP delay, high PDCP packet loss rate, CCE congestion,
excessive CPU usage, and SRS/PUCCH resource congestion. If cells are overloaded, expand the cell
capacity.
The following table lists the end-to-end root cause tree of low throughput problems. However, data need to
be collected using Wireshark for analyses on SP server and public network problems, which cannot be
resolved in batches currently. This table provides the root causes and problem resolving capability for each
low throughput problem.

Table 2.2.1.1.1.1.1.1.2 Root cause tree of LTE throughput problems and problem resolving capability
Problem Root Cause Analysis and Troubleshooting Automatic Analysis
Type Method Applicable
SP server Server capability If possible, connect the test PC No
directly to the SP server to test low
throughput cells where a certain
service is used by a large number of
UEs or the download speed of a
certain website is slow.
TCP parameters Check the TCP window size. No
Software Enable the firewall. SP servers do not No
configurations support TCP port extension.

Public NEs with flow Check DPI nodes. The GGSN No

network control communicates with the OCS at a too
high frequency.
Public bandwidth None No
limitation
Core network Registration Analyze MME CHR data. No, but technically feasible.
configurations
Rate limitation Analyze MME CHR data. No, but technically feasible.
Packet disorder Analyze MME CHR data. No, but technically feasible.
Insufficient Check the CN equipment capacity No. The capacity usage
equipment capacity usage. analysis depends on traffic
statistics of the CN.
Problem Root Cause Analysis and Troubleshooting Automatic Analysis
Type Method Applicable
Incorrect parameter Check core network parameters. No. The core network
settings parameter analysis depends on
MML commands of the CN.
Software defect This problem only occurs in the area No
covered by equipment provided by a
sole vender.
Equipment This problem only occurs in the area No
incompatibility covered by equipment provided by a
sole vender.
Transmission Bandwidth Analyze IPPM and GTPU counters. Yes
limitation
Long delay and Analyze IPPM and GTPU counters. Yes
large jitter
Packet loss and Analyze IPPM and GTPU counters. Yes
disorder
Too many eNodeBs Check the bandwidth usage of the No
in the same access eNodeB access ring. (This counter is
ring unavailable on the wireless network
side.)
Bandwidths Check the bandwidth usage of the No
occupied by services eNodeB access ring. (This counter is
on the access ring unavailable on the wireless network
side.)
Incorrect UGW Check the quintet configuration and No
configurations related feature switches. Packets are
lost due to high instant rate of the
UGW output port.
NE eNodeB failures or Analyze relevant alarms. Yes
performance
deterioration
eNodeB rate Check the licensed traffic volume. Yes
restriction
eNodeB processing Analyze traffic statistics (check Yes
capability whether the CPU usage exceeds the
limit).
Algorithm and Check known problems. Check Yes
feature restrictions whether features have any gain.
Check the strategies for enabling
features.
eNodeB capacity Check the usages of CCE, SRS, Yes
restricted PUCCH, and PRB resources.
Antenna Antenna faults and Analyze relevant alarms. Yes
 Problem Root Cause Analysis and Troubleshooting Automatic Analysis
Type Method Applicable
alarms
Reverse connection Another document is dedicated to Yes
or connection with this problem.
crossed pairs for
antennas
Uplink interference Another document is dedicated to Yes
this problem.
Imbalanced channels Check whether there is path loss Yes
imbalance of dual-channel
components in indoor distributed
eNodeBs with two antennas.
Air interface Air interface Analyze traffic statistics. Yes
resource (Grant/RB)
Weak coverage Analyze MRs. Yes
Interference Analyze MRs. Yes
UE UE capability Analyze signaling using CHRs. Yes
QCI configurations Analyze signaling using CHRs. Yes
AMBR rate Analyze signaling using CHRs. Yes
UE main and Perform UE trace test, because a No
diversity reception background analysis is not available.
imbalance

NOTE
Root causes highlighted in yellow are specified by China Mobile.

2.3 Air Interface Root Cause Location for Low Throughput

On the live network, factors related to the air interface or transmission resources are the major causes of
decreased LTE downlink throughput. Therefore, root cause location for LTE throughput problems focuses
on analyzing problems related to air interfaces and transmission resources using CHR data, MML
configuration data, and engineering parameters to determine the proportion of each air interface cause.
Targeted optimization measures can be chosen based on the analysis results to quickly resolve the
problems.
Air interface problems can be classified into weak coverage and interference. Transmission resource
problems are analyzed using traffic statistics (see section 2.2"Low Throughput Problem Isolation and
Demarcation").
 Root causes for weak coverage
Weak coverage is a cause of poor service quality. There are multiple causes of weak coverage. In the
optimization, MRs are analyzed to determine the root causes of weak coverage. The root causes are usually
missing configurations of neighboring cells, overshoot island coverage, delayed handover, link exceptions,
weak intensive coverage, no primary serving cells, and weak coverage at cell edges.
 Table 2.3.1.1.1.1.1.1.3 Mapping between root causes and weak coverage problems causing low LTE throughput
Isolation Root Cause Exception Remarks
and of Weak Proportion
Demarcat UL/DL
ion Coverage
Weak Missing Proportion of MRs 1. Missing configuration of neighboring cells can be
coverage configuration reporting poor determined if no or inappropriate handover to a cell
in the s of service quality
downlink neighboring caused by weak with strong signals is performed when handover
cells downlink coverage criteria are met because neighbor relationship with the
due to missing cell is not configured. (Handover criteria: The signal
configurations of
neighboring cells level of a neighboring cell exceeds the absolute
threshold within a certain period, and its difference
with the signal level of the serving cell exceeds the
relative threshold.)
5. Due to the complexity of wireless propagation
environment, UEs may receive signals from a remote
cell. Not all detected unconfigured neighboring cells
with strong signals are missing neighboring cells.
Detected unconfigured neighboring cells with strong
signals need to be filtered based on the contribution of
these cells to weak-coverage-caused poor service
quality.
6. At boundary between rural and urban scenarios, no
sites reside between the serving cell and the
unconfigured neighboring cells with strong signals
and therefore, no existence of overshoot coverage can
be determined. In this case, these neighboring cells
can be treated as the neighboring cells of the serving
cell.

Delayed Proportion of MRs A delayed handover can be determined if a UE detects a

handovers reporting poor configured neighboring cell with strong signals that meets
service quality the handover criteria (same as those used for
caused by weak identification of missing configuration of neighboring
downlink coverage cells), but no handover is performed before the network
due to delayed quality deteriorates.
handovers in the
downlink
Overshoot Proportion of MRs Overshoot island coverage can be determined if a UE
island reporting poor accesses the serving cell from a long distance away and
coverage service quality the unconfigured neighboring cell with the strongest
caused by weak signals meets the handover criteria and resides far from
downlink coverage the serving cell. In this situation, the UE cannot be
due to overshoot handed over to such a neighboring cell.
island coverage
Isolation Root Cause Exception Remarks
and of Weak Proportion
Demarcat UL/DL
ion Coverage
Link Proportion of MRs A link exception refers to component/channel faults that
exceptions reporting poor cause an excessively large difference between the uplink
service quality and downlink path losses. As a result, the uplink and
caused by weak downlink are imbalanced, leading to access failures and
downlink coverage poor service quality.
due to link
exceptions
Weak Proportion of MRs Weak extensive coverage can be determined if a UE
intensive reporting poor (usually indoors) is close to an eNodeB, but the receive
coverage service quality level is very low. This usually occurs when an outdoor
caused by weak UE calls an indoor UE. Therefore, indoor network
downlink coverage performance deteriorates.
due to weak
intensive coverage
No primary Proportion of MRs No primary serving cells occurs in the case of subtle
serving cells reporting poor differences between the receiving levels of the serving
service quality cell and that of the neighboring cells. Lack of a primary
caused by weak serving cell with strong signals reduces the signal-to-
downlink coverage noise ratio (SNR) and leads to frequent handovers. This
due to no primary situation usually occurs in scenarios overlapping areas
serving cell between multiple cells.
Weak Proportion of MRs Weak coverage at cell edges can be determined if the TA
coverage at reporting poor value in the MR is greater than the configured TA
cell edges service quality threshold input, or the value of TA multiplied by 4.89 is
caused by weak greater than the average inter-site distance and at the
downlink coverage same time, the signal level of neighboring cells is lower
due to weak than the weak coverage signal level threshold.
coverage at cell
edges

 Root cause analysis for network interference

Network interference is another cause for poor service quality. In densely-populated urban areas, LTE
network interference in the downlink is caused by overlaps with intra-frequency cells. Intra-network
interference is particularly more severe when PCI mod 3 conflicts occur with intra-frequency cells. In the
optimization solution, root causes for network interference are analyzed using MRs and engineering
parameters. The root causes include overshoot coverage, azimuth exceptions, PCI mod 3 conflicts, and
interference that can be reduced by inter-frequency handovers.

Table 2.3.1.1.1.1.1.1.1 Root causes of intra-frequency interference in the downlink

Isolation and Root Cause of Exception Proportion Remarks
Demarcation Downlink
Interference
Poor service Handover parameter Proportion of MRs Inter-frequency UE load transfer fails
quality caused exceptions reporting poor service due to missing configuration of inter-
by downlink quality caused by frequency neighboring cells or
interference interference (that can be incorrect handover parameter settings.
reduced by inter-frequency
handovers) due to
handover parameter
exceptions
Azimuth exceptions Proportion of MRs Interference is generated due to reverse
reporting poor service antenna connections, antenna quality
quality caused by and installation problems, and signal
interference due to azimuth reflection by the antenna installation
exceptions platform are identified.
Overshoot coverage Proportion of MRs Due to abnormal network structures,
reporting poor service excessive coverage generates network
quality caused by interference.
interference due to
overshoot coverage of the
serving cell
Proportion of MRs
reporting poor service
quality caused by
interference due to
coverage of neighboring
cells
PCI mod 3 conflicts Proportion of MRs Improper PCI planning causes PCI
reporting poor service mod 3 conflicts between nearby sites,
quality caused by generating intra-frequency
interference due to PCI interference.
mod 3 conflicts
Pilot pollution Proportion of MRs Because of intra-frequency cell
reporting poor service coverage overlap, subcarriers in
quality caused by neighboring cells affect the RS of the
interference due to pilot serving cell, generating interference.
pollution

Network interference may be generated by multiple cells. Therefore, when analyzing network interference, you
need to identify the following information in addition to the listed root causes of interference:

 Interference cell

 Proportion of interference traffic

This solution provides the interference cells for the serving cell in the detailed interference analysis table
(in the Root Causes for Interference sheet in the LTE downlink throughput rate problem analysis report file)
and lists key information, such as the interference value, TA, inter-site distance, interference causes, and
azimuth relationship.
 Table 2.3.1.1.1.1.1.1.2 Fields in the detailed interference analysis
Field Meaning
Proportion of MRs Traffic volume in the interference cell that interferes with the serving cell
reporting neighboring cells
with strong signal level
Average TA TA reported by the UE when network interference causes poor service quality
Inter-Site distance Distance between the interference cell and the serving cell
Cause Network interference type (PCI mod 3 conflict, coverage overlap of the serving
cell, overshoot coverage of the interference cell, and azimuth exceptions)
Azimuth relationship Azimuth relations (opposite or including) between the interference cell and
serving cell (for the analysis on network interference impact)　

2.4 Throughput Optimization Suggestions

Table 2.4.1.1.1.1.1.1.6 Optimization suggestions

Problem Root Cause Features/Parameters for Remarks
Optimization
Features/paramet Key LTE DRX Preferentially use features or
ers check features/parame parameters that can reduce network
ters Beamforming algorithm switch interference for optimization.
MLB
Downlink closed-loop power
control
PDCCH symbol adaption
DL frequency-selective
scheduling
CQI adjust algorithm switch
(CqiAdjAlgoSwitch)
Inner-loop power control for the
PUSCH in dynamic scheduling
mode switch
(InnerLoopPuschSwitch)
Enhanced uplink frequency-
selective scheduling switch
(UlEnhancedFssSwitch)
PUSCH IRC algorithm switch
(PuschIrcAlgoSwitch)
Problem Root Cause Features/Parameters for Remarks
Optimization
Weak coverage Link None For parameter exceptions that cause
exceptions uplink and downlink imbalance,
optimize these parameters based on
network requirements.

Missing Add the unconfigured Sort neighboring cells of

configurations neighboring cells. problematic cells according to the
of neighboring descending order of the times
cells neighboring cells fail to be
configured and export top 3
unconfigured neighboring cells and
configure neighbor relationship
between each top 3 cells and the
serving cell on these cells. (Top 3
cells must meet the condition that
UEs are handed over to such a cell
once every hour at least after
neighbor relationship configuration).
This suggestion is of high priority.
Overshoot Add the unconfigured remote A workaround for problems for
island coverage neighboring cells. overshoot island coverage is
provided, which is to add top 3
unconfigured remote neighboring
cells.
Delayed A3 measurement trigger Export handover parameters of the
handovers quantity serving cell that cause delayed
handovers, such as the downlink
Intrafreq handover offset receive level threshold for handovers
at the edge of the serving cell.
Intrafreq handover hysteresis
Export cell-specific handover
Intrafreq handover time to parameters (cell individual offset) of
trigger the serving cell that cause delayed
handovers related to specific
Frequency offset neighboring cells.
Cell individual offset
Inter-Freq HO trigger event
type
Interfreq A3 offset
A3-based interfreq A2 RSRP
threshold
A3-based interfreq A1 RSRP
threshold
Problem Root Cause Features/Parameters for Remarks
Optimization
CoverageBased Interfreq
RSRP threshold
Interfreq Handover Time to
Trigger
Interfreq handover hysteresis
Frequency offset
CoverageBased Interfreq
RSRP threshold
Interfreq Handover Time to
Trigger
Interfreq handover hysteresis
A3-based interfreq A2 RSRP
threshold
A3-based interfreq A1 RSRP
threshold
Weak intensive Optimize downtilts. Perform on-site engineering
coverage parameter check for cells with weak
intensive coverage. If the downtilt is
the same as the engineering
parameter, implement the downtilt
optimization suggestions. Otherwise,
you can choose not to implement
these suggestions.
No primary Optimize azimuths and downtilts. Select the primary serving cell that
serving cells covers the low throughput area
based on the geographic
demonstration of the coverage areas.
Then adjust the azimuths and
downtilts of other cells involved.
Weak coverage New sites are needed to resolve Build more sites for wider coverage.
at cell edges this problem. This suggestion is not provided in
this solution.
Others Unknown causes Other causes may occur in special
scenarios. The proportion of these
causes is larger in scenarios where
sites are located with a long distance
in between.
Poor service Inter-frequency A3-based Interfreq A2 RSRP For LTE cells with poor service
quality due to handover threshold quality, analyze inter-frequency cells
interference in parameter using neighboring cell lists in MRs.
the downlink exceptions A3-based interfreq A1 RSRP If inter-frequency cells meet the
threshold corresponding level threshold,
 Problem Root Cause Features/Parameters for Remarks
Optimization
Interfreq Handover Time to export handover parameters for
Trigger inter-frequency cells with the
highest level and provide parameter
Interfreq handover hysteresis adjustment suggestions. These
suggestions include adding
Cell individual offset
unconfigured neighboring cells and
optimizing handover parameters.
Azimuth Check antenna azimuths. For cells with azimuth exceptions,
exceptions check against the detailed
interference analysis table to find
out which cell pair has azimuth
exceptions. Then check whether a
repeater has been installed. If not,
perform on-site engineering
parameter check and adjust the
antenna.
Interference Optimize downtilts. Perform on-site engineering
due to parameter check for cells with
overshoot overshoot coverage based on the
coverage of the downtilt optimization suggestions. If
serving cell the downtilt is close to or larger than
20 degrees, do not perform
Interference optimization. If the downtilt is
due to coverage smaller than 20 degrees and will be
overlap of the smaller than 20 degrees after
interference optimization, reduce the downtilt by
cell three degrees (an experience-based
value) or perform adjustment based
on the automatic cell planning
(ACP) results.
PCI mod 3 Optimize the PCI settings. Select cells that need PCI
conflicts optimization and optimize
throughput KPIs of top cells based
on the number of MRs reporting
poor service quality in top cells. A
target PCI, which is not provided in
this solution, is available in the PCI
optimization topic.

3 Procedures

4 Guidelines
4.1 Overview
This chapter provides the delivery procedure for the solution and operation guidelines for field engineers.
4.2 Tools

4.3 Data Collection

The following table lists the data involved in the solution.

Table 4.3.1.1.1.1.1.1.1 Data sources

Downlink Data Format Collection Scope How to Obtain
Throughput
Topic
External CHR Mandatory SIG CHR data in two hours per NIC/U2000
day for consecutive three
days
Traffic Statistics Mandatory MRF CHR data in two hours per NIC/U2000
day for consecutive three
days
XML Mandatory XML Latest configurations NIC/U2000
Configurations
Engineering Mandatory xls/xlsx Latest configurations Collected by
Parameters frontline
personnel
Alarms Optional CSV Historical alarms and NIC
active alarms
MML Packets Optional TXT Outputs of the MML NIC
commands executed
recently
Operation Logs Optional Oprlog All operation logs NIC
Interference Optional None None Generated by
Matrixes tools

4.4 Solution Application Instructions

This section describes the outputs of an optimization report only.

4.4.1 Remarks on Parameter Settings

Threshold parameters are used to evaluate problems related to transmission and capacity in problem
isolation and demarcation, and evaluate problems related to weak coverage and network interference in RF
root cause analyses. The solution does not aim at optimizing the PCI. Therefore, no PCI rules are set in the
solution.
 Table 4.4.1.1.1.1.1.1.2 Settings of threshold related parameters
Object Mandatory/ Value
Threshold Default Value Description
Optional Range

LTE Transmission If the transmission packet loss

throughput packet loss rate Mandatory 0.06% 0-00% is higher than the threshold,
threshold transmission is abnormal.
If the transmission delay value
Transmission is greater than the threshold,
Mandatory 20 ms 0-1000
delay threshold the transmission delay is
abnormal.
If the uplink interference is
Uplink
-128 to higher than the threshold,
interference Mandatory -110 dBm
0 severe interference occurs in
threshold
the uplink.
If the PDCP packet loss rate is
PDCP packet 0- higher than the threshold,
Mandatory 5.00%
loss threshold 100% packets loss during scheduling
in the cell is abnormal.
If the PDCP scheduling delay
PDCP
is greater than the threshold,
scheduling Mandatory 32 ms 0-1000
the scheduling delay in the cell
delay
exceeds the allowed value.
If the CCE usage is higher than
CCE usage 0-
Mandatory 80.00% the threshold, CCE resource
threshold 100%
congestion occurs in the cell.
If the PRB usage is higher than
PRB usage 0-
Mandatory 70.00% the threshold, PRB resource
threshold 100%
congestion occurs in the cell.
If the CPU usage is higher than
CPU usage 0-
Mandatory 60.00% the threshold, CPU resource
threshold 100%
congestion occurs in the cell.
If the downlink RSRP is
smaller than the threshold, the
MR reports weak coverage. If
Downlink
the proportion of MRs
RSRP -140 to
Mandatory -110 reporting weak coverage for a
threshold for 40
cell is greater than the
weak coverage
threshold for filtering weak
coverage problems, the cell has
weak coverage.
If the downlink CQI is smaller
Downlink CQI
Mandatory 10 0-15 than the threshold, the MR
threshold
reports poor service quality.
Level threshold Mandatory -97 -140 to If the level of a neighboring
for identifying 40 cell is higher than the
neighboring threshold, UEs in the serving
Object Mandatory/ Value
Threshold Default Value Description
Optional Range

cells with cell can be handed over to this

strong signals neighboring cell.
If the path loss difference
Path loss between the uplink and
difference Mandatory 11 0-20 downlink is greater than the
threshold threshold, uplink and downlink
imbalance exists.
If the TA is smaller than the
threshold, intensive coverage is
weak. If the proportion of MRs
reporting weak coverage is
larger than the threshold, the
cell has weak coverage
TA threshold
problems. If the proportion of
for weak
Mandatory 3 0-20 MRs reporting poor service
intensive
quality due to weak intensive
coverage
coverage is higher than the
threshold for filtering sub-
problems related to weak
coverage, optimization needs
to be performed to improve
weak intensive coverage.
If the TA is greater than the
threshold, weak coverage at
cell edges occurs. If the
proportion of MRs reporting
TA threshold
weak coverage at cell edges is
for weak
Mandatory 3 0-20 greater than the threshold for
coverage at cell
filtering sub-problems related
edges
to weak coverage, optimization
needs to be performed to
improve the weak coverage at
cell edges.
Threshold for If the number of neighboring
the number of cells is greater than the
neighboring threshold, these neighboring
Mandatory 2 0-20
cells without a cells are treated as the
primary neighboring cells of the serving
serving cell cell.
RSRP If the RSRP difference
difference between the neighboring cells
threshold for Mandatory 3 0-20 and the serving cell is smaller
no primary than this threshold, coverage
serving cell overlap exists.
RSRP Mandatory 6 0-20 If the RSRP difference
difference between the neighboring cells
threshold for and the serving cell is smaller
network than this threshold, network
 Object Mandatory/ Value
Threshold Default Value Description
Optional Range

interference interference exists.

Threshold for
If the number of neighboring
the number of
cells is greater than the
neighboring Mandatory 2 0-20
threshold, pilot pollution
cells with pilot
exists.
pollution
If the RSRP difference
Level
between the neighboring cells
difference
Mandatory 6 0-20 and the serving cell is smaller
threshold for
than this threshold, network
pilot pollution
interference exists.
0- If the number of MRs reporting
Threshold for
100% poor service quality due to
filtering
Mandatory 5% interference is greater than the
interference
threshold, network interference
problems
exists.
0- If the number of sub-problems
100% related to interference is
Threshold for
greater than the threshold,
filtering
Mandatory 5% optimization needs to be
interference
performed to address a certain
sub-problems
type of network interference
problems in the cell.

Threshold description:
1. Set the downlink RSRP threshold for weak coverage based on network requirements.
2. The threshold for the number of neighboring cells without a primary serving cell cannot be too large.
Otherwise, such neighboring cells may not be detected.
1. The RSRP difference threshold for interference cannot be set to a value smaller than 6 dB. If the value
of this threshold is smaller than 9 dB, the performance of the serving cell is affected.
2. The downlink CQI poor service quality threshold refers to the 5-bit CQI reported in MRs. Mapping
relationships exist between 5-bit CQIs and 4-bit CQIs.

4.4.2 Description of the LTE Throughput Optimization Report

The following files are exported after the analysis.

 Evaluation report on TopN cells for LTE downlink throughput rates

 This report includes results of LTE downlink throughput evaluation, problem isolation and
demarcation, associated KPIs, license analysis, UE operation log query, and parameter checks.
Problems identified and optimization suggestions are also provided.
 LTE downlink throughput rate problem analysis report
This report includes results of throughput evaluation, problem isolation and demarcation, license
analysis, related operation logs, poor service quality due to weak coverage, and poor service quality
due to network interference.
 The LTE DLThroughputReport sheet in the LTE downlink throughput rate problem analysis report
In this report, all cells are evaluated and ranked based on indexes selected by customers. The average
number of UEs in a cell, average downlink CQI, average downlink MCS, rank 2 proportion, PDSCH
IBLER and BLER counters, and KPIs related to access, handover, and service drop are also provided.

 The DemarcationResultReport sheet in the LTE downlink throughput rate problem analysis report
Analyses on transmission, NE alarms, antenna channels, UEs, cell capacity, licenses, related operation
logs, and proportions of services with poor quality due to weak coverage and network interference are
performed by grid or top cell group in this report.

 Proportion of cell-level air interface root causes

In-depth analyses on the identified problems related to downlink weak coverage and network
interference are performed by grid and top cell group to locate root causes. All the location results are
shown in this file without filtering. The specific root causes are described in the preceding chapters.
The identified root causes will guide you in subsequent network optimization.
 LTE DLThroughputAdvice
Each sheet in this file describes suggestions of resolving an exception detected during root cause
identification.
 Suggestions about exceptions due to non-air interface root causes are in the following sheets in this file:
 Parameter Check Result
 Transmission Related Problems Advice
 Hardware Related Problems Advice

 Antenna Channels Related Problems Advice

 UE Problems Advice

Table 4.4.2.1.1.1.1.1.3 Optimization suggestions on non-air interface root causes

Non-Air Interface Root Causes Sheet in the This File
Parameter check Parameter Check Result
Transmission issues Transmission Related Problems Advice
NE faults Hardware Related Problems Advice

Antenna channel Antenna Channels Related Problems Advice

Cell capacity congestion Cell Capacity Problems Advice

UE UE Problems Advice

Optimization suggestions for each air interface root cause can be provided only when certain filtering
conditions are met.

Table 4.4.1.1.1.1.3.1.4 Optimization suggestions on air interface root causes

Root Cause Optimization Suggestion Filtering Condition
Missing Adding neighboring cells The proportion of MRs reporting poor service quality
configurations of due to weak coverage is higher than 2%. The proportion
neighboring cells of MRs reporting missing configurations of neighboring
cells is higher than 5%.
Root Cause Optimization Suggestion Filtering Condition
Overshoot island Adding neighboring cells The proportion of MRs reporting poor service quality
coverage due to weak coverage is higher than 2%. The proportion
of MRs reporting overshoot island coverage p is higher
than 5%.
Delayed handovers Cell-level optimization The proportion of MRs reporting poor service quality
suggestions on delayed due to weak coverage is higher than 2%. The proportion
handovers and cell-specific of MRs reporting delayed handovers is higher than 5%.
optimization suggestions on
delayed handovers
Link exceptions Optimization suggestions on The proportion of MRs reporting poor service quality
link exceptions due to weak coverage is higher than 2%. The proportion
of MRs reporting link exceptions is higher than 5%.
No primary Optimization suggestions on The proportion of MRs reporting poor service quality
serving cells no primary serving cell due to weak coverage is higher than 2%. The proportion
of MRs reporting no primary serving cell is higher than
5%.
Weak intensive Optimization suggestions on The proportion of MRs reporting poor service quality
coverage weak intensive coverage due to weak coverage is higher than 2%. The proportion
of MRs reporting intensive weak coverage is higher
than 5%.
Weak coverage at Optimization suggestions on The proportion of MRs reporting poor service quality
cell edges weak coverage at cell edges due to weak coverage is higher than 2%. The proportion
of MRs reporting intensive weak coverage is higher
than 5%.
Network Optimization suggestions The proportion of MRs reporting poor service quality
interference that is about adding neighboring cells due to interference is higher than 2%. The proportion of
due to missing MRs reporting missing configurations of inter-
configurations of frequency neighboring cells is higher than 5%.
inter-frequency
neighboring cells
can be reduced by
handovers
Network Cell-level optimization The proportion of MRs reporting poor service quality
interference that is suggestions on delayed due to interference is higher than 2%. The proportion of
due to missing handovers and cell-specific MRs reporting interference due to delayed inter-
configurations of optimization suggestions on frequency handovers is higher than 5%.
inter-frequency delayed handovers
neighboring cells
can be reduced by
handovers
Overshoot Optimization suggestions on The proportion of MRs reporting poor service quality
coverage overshoot coverage due to interference is higher than 2%. The proportion of
MRs reporting overshoot coverage is higher than 5%.
The proportion of MRs reporting interference from
neighboring cells to the serving cell is higher than 5%.
Interference due to Check PCIs. The proportion of MRs poor service quality due to
 Root Cause Optimization Suggestion Filtering Condition
azimuth interference is higher than 2%. The proportion of MRs
exceptions reporting azimuth exceptions of a single cell is higher
than 5%.
Interference due to
PCI mod 3
conflicts
Pilot pollution Optimization suggestions on The proportion of MRs reporting poor service quality
pilot pollution due to interference is higher than 2%. The proportion of
MRs reporting pilot pollution is higher than 5%.

NOTE
These filtering conditions can be modified using the configuration file on the stand-alone OMStar.
Optimization suggestions must be reviewed and approved by field engineers who know well about the network
conditions before the suggestions are implemented.

4.4.3 Suggestion Descriptions and Implementation

All optimization suggestions should be checked manually. If the parameter optimization suggestions
do not meet on-site network requirements, perform the suggestions as required.
 Feature parameter optimization suggestions: Features that increase throughput take effect in specific
scenarios. Exercise caution when enabling these features. Key parameter Pa and Pb must be adjusted
based on associated KPIs of the operator. Implement these suggestions after negotiation with the
operator.
 No optimization suggestions on remote weak coverage, no primary serving cell, pilot pollution, and
interference caused by azimuth exception are provided. Field engineers perform routine optimization
based on analysis results of these problems. Further adjustment can be made based on geographical
analysis results.
 Handover optimization suggestions: After performing root cause isolation and demarcation for all
cells, use the results to select cells to be optimized according to certain threshold rules and put forward
corresponding optimization suggestions.
 PCI optimization suggestions: No target PCIs can be recommended for adjustment. Adjust the PCI
based on experiences or use PCI optimization tools.
− If PCI mod 3 conflicts cannot be avoided, adjust the PCI to reduce interference traffic. This can
improve network performance.
− If the PCI cannot be adjusted, you can adjust the antenna to reduce coverage overlap with the
interference cell. This can greatly reduce downlink interference.
 Optimization suggestions need to be reviewed by field engineers based on on-site network
requirements before implementation.

5 Cases
5.1 Case 1: Routine Optimization in the XX Office of the XX Operator
5.1.1 Background
According to field engineers:
 1. Multiple tools are used in routine network optimization, and these tools are not integrated on the same
platform, causing inconveniences to network optimization.
2. When locating a problem using a tool, field engineers need to do a large number of drive tests or other
auxiliary work, causing a long problem-locating duration.
3. Some tools are complex and therefore can be used only by senior or experienced engineers.
4. Optimization suggestions are provided based on engineers' skills or experience, and no quantitative
evaluation standards are provided.
Therefore, a comparison between optimization for poor service quality (the most important topic for field
engineers) in the accurate network optimization solution and live-network delivery is required. The
comparison can provide the benefits of the accurate network optimization solution in both efficiency and
effects.

5.1.2 Comparison Perspectives

 Optimization efficiency: Select groups of cells of the same number and perform optimization analyses
on theses cell groups within the same period using the same tools. Then perform comparative analyses
on these cells in terms of problem location period.
 Problem location: Check whether problems can be located at the same granularity as or more
granularity than manual analysis onsite while ensuring problem location efficiency.
 Optimization methods: Ensure that the outputs of this solution are consistent with field engineers'
optimization methods. This helps facilitate subsequent implementations.
 Optimization effects: Check whether the provided optimization suggestions help improve network
performance.

5.1.3 Comparison Results

Efficiency comparison for areas of the same scale

The accurate optimization solution efficiently identifies KPI problems and isolates problems related to CN,
transmission and equipment faults, in order to identify weak coverage and network interference scenarios
and offer optimization suggestions for top cells. The traditional optimization requires more analyses on
associated KPIs and more time and manpower to locate problems. The location results are often not
accurate. Accurate optimization, however, incorporates data collection, data processing, and the
provisioning of optimization suggestions, increasing analysis efficiency by 30.9%. Accurate optimization
increases the overall optimization efficiency by 11.4%.
Problem location comparison: Accurate optimization excels drive tests for traditional optimization in
identifying more problems. In accurate optimization, background data is used to collect statistics about
 overshoot coverage on high floors, weak coverage for indoor UEs, and network interference in areas where
drive tests are impossible.

Table 5.1.3.1.1.1.2.1.1 Location capability comparison

Air Interface Root Cause On-site Root Cause Dedicated Root Cause
Identification Identification
Weak coverage: missing configurations of √ √
neighboring cells
Weak coverage/network interference: delayed × √
handovers
Weak coverage: overshoot island coverage × √
Weak coverage: no primary serving cell √ √
Weak coverage: poor service quality due to weak × √
intensive coverage
Weak coverage: link exceptions × √
Network interference: PCI mod 3 conflicts √ √
Network interference: azimuth exceptions × √
Network interference: interference due to √ √
overshoot coverage
Network interference: pilot pollution √ √

Examples
On-site identification of air interface root causes for weak coverage
PCI Proportion of MRs Proportion of MRs Missing Optimization
Reporting Poor Service Reporting Poor Configurations Suggestions
Quality Due to Weak Service Quality in the of Neighboring
Coverage in the Downlink Downlink (%) Cells
(%)
320 8.877691 12.16732 Yes Add a cell with PCI
391

Topic identification of air interface root causes for weak coverage

Servi eNodeB Cell PCI Total Number Neighb eNo Cell PCI Avera Averag
ng ID of ID of of numb of MRs oring deB ID of of ge e RSRP
Cell the the the er of Reporti Cell ID the the RSRP in the
Nam Serving Servin Serv MRs ng Poor Name of Neig Neig in the Serving
e Cell g Cell ing report Service the hbori hbori Neigh Cell
Cell ing Quality Neig ng ng borin
Poor Caused hbor Cell Cell g Cell
Servic by ing
e Weak Cell
Qualit Coverag
y Due e Due to
to Missing
Weak Configu
Cover ration
age of
Neighbo
ring
Cells
Gaos 722776 1 320 108 4 Wenyu 7227 1 391 -94.5 -107.5
ha an 95
Com Buildin
muni g
ty DM_ne
D_1 w_1

Optimization suggestions

Serving eNodeB ID Cell ID PCI of Total Number of Average Avera

Cell of the of the the number of MRs RSRP in ge
Name Serving Cell Serving Serving MRs Reporting the Serving RSRP
Cell Cell reporting Poor Service Cell in the
Poor Quality Due to Servin
Service Weak g Cell
Quality Intensive
Due to Coverage
Weak
Coverage
Dormitory 394863 2 250 1303 420 -108.83 3.05
buildings
on the
Xiasha
campus of
Hangzhou
Normal
University
SM_2

Tests were performed 140 meters away from the eNodeB serving the cell "Dormitory buildings on the
Xiasha campus of Hangzhou Normal University SM_2", in such a dormitory building.
The RSRP of the entire indoor area was smaller than -100 dBm.
Network interference problem 1
When Xiasha campus of Hangzhou Gongshang University SM_2 functioned as the serving cell, it was
interfered by overshoot coverage of the cell "h719870 Zhejiang Economic and Trade Polytechnic SM_2".

The following figure shows the test results. In the test performed in Zhejiang Gongshang University,
signals from Zhejiang Economic and Trade Polytechnic SM_2 were detected. High SINR was detected
because the drive test vehicle was on the road and the strong signal levels of the serving cell could be
received. When UEs were indoors, the RSRP of the serving cell decreased to a level close to the RSRP of
the neighboring cell, causing network interference.

Network interference problem 2

The cell "h718134 the canteen of Hangzhou Vocational and Technical College SM_3" caused interference
to the cell "Xiasha campus of Zhejiang University of Media and Communications SM_1".

The preceding figure shows that the site near Hangzhou Vocational and Technical College was built with a
great high near the road, while no barriers between the canteen of Zhejiang University of Media and
Communications and Hangzhou Vocational and Technical College blocking signals. Therefore, network
interference occurred.
Optimization effect comparison
Optimization was performed based on the list of overall optimization suggestions export in this topic.
Suggestions on parameter adjustment were implemented for 22 cells in Xiasha district after being reviewed
by field engineers (with no objection). Network performance of 18 cells improved. This number accounted
 for 81.8% of all cells. Network performance of two cells, for which RF adjustment had been performed,
improved.

Cell Name Action Average Average Absolute Relative

Downlink Downlink Gain Gain (%)
Throughput Throughput (Mbit/s)
Before After
Optimization Optimization
(Mbit/s) (Mbit/s)
Building 33 of Chengxi Adding 8.95 10.00 1.05 11.73
Garden SM_3 neighboring
cells and
adjusting the
CIO
Xiasha campus of Adding 2.34 3.29 0.95 40.60
Zhejiang Gongshang neighboring
University SM_1 cells and
adjusting the
CIO
Xiasha campus of Adding 6.38 7.74 1.36 21.32
Zhejiang Gongshang neighboring
University SM_3 cells and
adjusting the
CIO
Xiasha campus of Enabling 8.09 8.61 0.52 6.43
Hangzhou Normal MLB
University SM_1
Xiasha campus of Enabling 7.44 9.23 1.79 24.06
Hangzhou Normal MLB
University SM_2
Xiasha campus of Adjusting the 6.61 6.88 0.27 4.08
Hangzhou Normal CIO
University SM_3
Dormitory buildings on Enabling 8.89 9.91 1.02 11.47
the Xiasha campus of MLB
Hangzhou Normal
University SM_1
Dormitory buildings on Enabling 9.64 11.17 1.53 15.87
the Xiasha campus of MLB
Hangzhou Normal
University SM_2
Dormitory buildings on Enabling 7.65 9.07 1.42 18.56
the Xiasha campus of MLB
Hangzhou Normal
University SM_3
Xiasha campus of Adding 10.52 8.16 -2.36 -22.43
Hangzhou No.4 High neighboring
School SM_3 cells and
Cell Name Action Average Average Absolute Relative
Downlink Downlink Gain Gain (%)
Throughput Throughput (Mbit/s)
Before After
Optimization Optimization
(Mbit/s) (Mbit/s)
adjusting the
CIO
Northern Xiasha Adding 7.86 7.23 -0.63 -8.1
campus of Hangzhou neighboring
No.4 High School cells and
SM_1 adjusting the
CIO
Northern Xiasha Adding 7.54 7.56 0.02 0.27
campus of Hangzhou neighboring
No.4 High School cells and
SM_2 adjusting the
CIO
Northern Xiasha Adding 6.31 8.02 1.71 27.10
campus of Hangzhou neighboring
No.4 High School cells and
SM_3 adjusting the
CIO
Xiasha campus of Adding 8.12 8.59 0.47 5.80
Zhejiang Economic and neighboring
Trade Polytechnic cells
SM_2
Green belt at Xuelin Enabling 11.62 12.09 0.47 4.04
Street SM_1 MLB
Green belt at Xuelin Adjusting the 8.68 12.14 3.46 39.86
Street SM_2 CIO
Green belt at Xuelin Enabling 10.54 11.94 1.40 13.28
Street SM_3 MLB
Northern Yunshui Adjusting the 8.68 8.87 0.19 2.2
Residences SM_2 CIO
Zhejiang University Adjusting the 9.41 7.89 -1.52 -16.17
Zhongzi Integrated CIO
Control Co., Ltd SM_1
Zhejiang University Adjusting the 8.50 11.20 2.70 31.76
Zhongzi Integrated CIO
Control Co., Ltd SM_3
Administrative building Adjusting the 8.14 9.214 1.074 13.1941
of Zhejiang Gongshang PCI
University SM_3
Administrative building Adjusting the 11.67 12.298 0.628 5.38132
Cell Name Action Average Average Absolute Relative
Downlink Downlink Gain Gain (%)
Throughput Throughput (Mbit/s)
Before After
Optimization Optimization
(Mbit/s) (Mbit/s)
of Zhejiang Gongshang PCI
University SM_2

The second round of optimization focused on RF adjustment for cells with strong interference. Antenna
downtilts for 14 cells were adjusted. The throughput of these cells in December, 2012 increased
significantly by 0.67 Mbit/s. The growth rate was 7.63%.

Success rates increased. The service drop rate decreased. The downlink throughput increased significantly.
Major KPIs retained stable and did not fluctuate greatly except for special occasions like the civil service
examination. The average throughput and spectral efficiency increased after optimization suggestions were
implemented.