Product Name Security Level

GENEX Discovery INTERNAL

Product Version
Total 34 pages
All

GENEX Discovery Geo-location Technical Paper

Prepared By Wu Zhibiao (employee ID: 00302686) Date 2016-10-31

Li Xi (employee ID: 00344634)
Wang Runting (employee ID: 00334421)
Reviewed By Meng Qingdong (employee ID: 00224353) Date 2016-12-06
Zheng Kaisi (employee ID: 00266114)
Zhao Qunqun (employee ID: 00387977)
Approved By Date

Authorized By Date

Huawei Technologies Co., Ltd.

All Rights Reserved.
(REP01T01 V2.6/IPD-CMM V3.0/For Internal Use Only)
 GENEX Discovery Geo-location Technical Paper INTERNAL

Change History

Date Revision Version Change Description Author

2016-10-31 1.00 Completed the draft. Wu Zhibiao (employee

ID: 00302686)
2016-12-10 1.01 Updated the content based on review Wu Zhibiao (employee
suggestions. ID: 00302686)
2018-3-10 1.02 Updated the Geo-location Technology Wu Zhibiao (employee
Specifications ID: 00302686)

2020-02-21 Huawei confidential. No spreading without permission. Page 2 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Contents

Change History.................................................................................................................................2
1 About This Document..................................................................................................................2
1.1 Purpose...............................................................................................................................................................2
1.2 Application Scope..............................................................................................................................................2

2 Geo-location Technology Overview..........................................................................................2

2.1 Definition...........................................................................................................................................................2
2.2 Geo-Location Application Panorama.................................................................................................................2
2.3 Related Concepts...............................................................................................................................................2
2.3.1 Location Precision....................................................................................................................................2
2.3.2 Location Confidence.................................................................................................................................2
2.3.3 Grid-based Processing..............................................................................................................................2
2.3.4 Feature Database.......................................................................................................................................2
2.4 Main Geo-location Technologies in Industry.....................................................................................................2
2.5 Geo-location Technologies in the Discovery.....................................................................................................2
2.5.1 Location Principle Overview....................................................................................................................2
2.5.2 A-GPS Location........................................................................................................................................2
2.5.3 Indoor Distribution Location....................................................................................................................2
2.5.4 RTT Location............................................................................................................................................2
2.5.5 Feature Matching Location.......................................................................................................................2
2.5.6 WCCL Location........................................................................................................................................2
2.5.7 Subscriber Environment Feature Analysis...............................................................................................2
2.5.8 Map Matching and Location Calibration..................................................................................................2

3 Geo-location Technology Specifications..................................................................................2

3.1 Location Delay...................................................................................................................................................2
3.2 Location Precision.............................................................................................................................................2
3.3 Indoor and Outdoor Differentiation Accuracy...................................................................................................2
3.4 Road Subscriber Identification Accuracy..........................................................................................................2
3.5 Location Success Rate.......................................................................................................................................2

4 Terms................................................................................................................................................2

2020-02-21 Huawei confidential. No spreading without permission. Page 3 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

GENEX Discovery Geo-location Technical Paper

Keywords
MR location, indoor and outdoor differentiation, road matching, building matching

Abstract
This document is intended for the marketing personnel, MO, network optimization service design
personnel, and product development and design personnel of service geographic observation products. It
clarifies the geo-location technologies and specifications.

List of Acronyms and Abbreviations

Acronym/Abbreviation Full Spelling

A-GPS Assisted Global Positioning System

CHR call history record
DCH Dedicated Transport Channel
GPS Global Positioning System
IPDL idle period downlink
LCS location service
LMU location measurement unit
LTE Long Term Evolution
MDT Minimization of Drive Tests
MR measurement report
OTDOA Observed Time Difference of Arrival
RNC radio network controller
RRU remote radio unit
RTD relative time difference
RTT round-trip time
SAS stand-alone SMLC
SMLC serving mobile location center
SPU signaling processing unit
TDOA time difference of arrival
TOA time of arrival
TOW time of week
UE user equipment

2020-02-21 Huawei confidential. No spreading without permission. Page 4 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Acronym/Abbreviation Full Spelling

UTDOA Uplink-Time Difference of Arrival

WCCL weighted centroid correction localization

2020-02-21 Huawei confidential. No spreading without permission. Page 5 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

1 About This Document

1.1 Purpose
Based on the main geographic applications of the Discovery, this document describes the basic concept,
main methods and principles, and implementation dependencies and restrictions of geo-location
technologies, and clarifies geo-location capability specifications.

1.2 Application Scope

This document serves as a technical clarification of R&D engineers for marketing personnel. It unifies
the promotional statement of the geo-location capability based on the implementation principles of the
Discovery geo-location technology algorithm. Contents in this document are used as basic input for
market promotion, sales and service product design, and implementation.
It is intended for R&D engineers, marketing personnel, and MOs. Marketing personnel can use it to
communicate with technical personnel of customers. However, providing this document to customers or
peer vendors is prohibited.
Unless otherwise specified, all geo-location algorithms described in this document have been used in
GENEX Discovery V200R002C50.

2 Geo-location Technology Overview

2.1 Definition
A geo-location technology determines the location information of a UE and its call behavior attributes
(indoor or outdoor call, and moving speed) based on radio signals measured on a mobile communication
network.
A geo-location technology is essentially used to determine the location of a UE when it reports an MR
during a call. Therefore, the technology is also called MR location. The technology correlates events
occurring during a call with located MRs to further determine the locations where the events occur. The
events include control-plane events (such as call drops) and user-plane events (such as periodic traffic
volume events and web page access events).

2.2 Geo-Location Application Panorama

Through locating reported MRs and events occurring during calls of all UEs on the network, a geo-
location technology achieves visualized display of the radio network environment status (coverage and
quality), traffic distribution, and event distribution (call drops and handovers), supporting network
performance monitoring and optimization evaluation.
The Discovery is a geographic evaluation tool in the service domain tool system. Based on an accurate
and effective geo-location engine and various GIS display skills, the Discovery provides network- and
subscriber-level geographic evaluation and optimization applications.

2020-02-21 Huawei confidential. No spreading without permission. Page 6 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Figure 1.1 Geo-location applications of the Discovery

2.3 Related Concepts

2.3.1 Location Precision
Location precision refers to the difference (deviation or error) between the location point calculated
using a location algorithm and the actual location point of a UE.
In MR location applications, the difference between calculated and measured location points of each
MR varies. Therefore, the deviation of a certain MR cannot determine the location precision of a batch
of MRs. Currently, several key points on the cumulative distribution function (CDF) curve are mostly
used to describe the location precision in industry. The curve can reflect the location precision under a
certain probability. For example, the probability that the location deviation is less than 100 m is 67%
(which is described as 100m@67% in the following sections of this document). The probability reflects
the location precision confidence. Probabilities 67% and 95% are mostly used in industry. 67% is close
to the +1σ position on the normal distribution curve, that is, the location deviation corresponding 67% is
consistent with the standard deviation of the normal distribution. 95% is close to the +2σ position on the
normal distribution curve.

2020-02-21 Huawei confidential. No spreading without permission. Page 7 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Figure 1.2 CDF curve

From a statistical point of view, the number of MR location times at a certain point may affect the
overall location precision of this point. At a certain point on a network, if the MR collection duration is
long, the probability that more MRs are reported at the point will be high. Geo-location applications
locate the weak-coverage, strong-interference, and high-traffic problem areas based on multi-MR
statistical results to evaluate networks. Therefore, Huawei introduces a new location precision concept
— traffic precision.
Traffic precision refers to the deviation between the position reflected by MR location–based traffic
map statistics (such as the signal strength and traffic volume) and the actual geographic position. Traffic
precision involves two types of deviations. One is the position location deviation, which is generated by
multiple times of location. The other is statistical information deviation.
For example, if an area is determined as a high-traffic area based on the location result, the deviation
between this area and the actual high-traffic area can be regarded as the traffic precision. Based on
experience data statistics, if the same batch of data is used, the traffic precision is about 30% higher than
the single-point location precision in most cases.

Figure 1.3 Traffic precision measurement

2020-02-21 Huawei confidential. No spreading without permission. Page 8 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

2.3.2 Location Confidence

Location confidence is a theoretical location precision range of an MR determined based on the location
input conditions and calculation method of the MR. The confidence can be classified into different
levels.

Figure 1.4 Location confidence

This figure shows a location precision range. If the location confidence of MR1 in the grid with a red
circle is 50 meters, the actual location of the UE is within the grid, and the MR1 counter data
represented by the grid is accurate. If the location confidence of MR2 in the grid with a blue circle is
100 meters, the actual location of the UE is within the nine grids in blue, and the MR2 counter data
represented by the grid is inaccurate. MR2 location confidence is lower than that of MR1.
Table 4.1 describes the confidence levels determined based on theoretical precision range of the MR
location method in the Discovery.

Table 4.1 Location confidence levels in the Discovery

Location Description
Confidence

The average precision is 20 m to 80 m. Only GPS/A-GPS (including LTE MDT data)

Excellent
or indoor distributed site location results are provided at this level.
The average precision is 80 m to 120 m. Quick location results in urban inter-site
Good
distance scenarios or high-score feature matching results are provided.
The average precision is 120 m to 200 m. Quick location results in suburban inter-site
Normal
distance scenarios or medium-score feature matching results are provided.
The average precision is 200 m to 350 m. Location results in suburban inter-site
Poor
distance (dual-site) scenarios are provided.
The average precision is greater than 350 m. Location results in other scenarios that
Worse
do not meet the preceding conditions, such as isolated sites, are provided.
If MR location fails or combination of event counters and MRs fails, the location of
Failed
the site where the event occurs is provided.

2020-02-21 Huawei confidential. No spreading without permission. Page 9 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

2.3.3 Grid-based Processing

Online subscribers report massive MRs during calls, and massive geographic counters are measured
based on the MRs. If the counter data is directly loaded and displayed on the GIS map, data loading and
refreshing is slow in most cases. To resolve this problem, a grid-based map is widely used in industry to
display the counter data.
Numerous evenly-spaced grids with equal size form a grid array. Each grid can be viewed as a pixel and
is defined using rows and columns. Grid-based MR processing is to normalize the latitudes and
longitudes of all MRs within a grid. All MRs in one grid are normalized into one data record. The MR
data is rendered using the grid-based rendering service provided by the GIS and displayed on the grid
layer.
Pay attention to the following notes:
2. Deviations exist in grid-based MR processing results.
Plane coordinates are used in grid-based MR processing. Longitudes and latitudes on a spherical
coordinate system are converted to X and Y points (meter) on a plane coordinate system. As the
Earth is an irregular ellipsoid, no matter which project method is used, conversion deviations are
generated.
3. Grid precision is not equal to location precision.
Grids are N m x N m (such as 50 m x 50 m, or 100 m x 100 m) squares. The square size is grid
granularity, not grid precision.
4. User appropriate grid granularity.
From the information fineness aspect, 50 m x 50 m grids provide more and better differentiated
information than 100 m x 100 m grids. However, when the location precision is low and the grid
size is small, the MR confidence in grids will be low, so will the confidence of service counters
measured based on the grids. In addition, the data volume of a 50 m x 50 m grid map is four times
of that of a 100 m x 100 m grid map. In grid rendering and display, the performance of 50 m x 50
m grids is lower than that of 100 m x 100 m grids in most cases.
5. Some information is missing in grid-based display.
In service features of the Discovery, as some data is missing in grid-based processing, grid maps
are not used in some geographic applications. For example, a vector-based display method is used
in the Key Road Evaluation feature and the single-subscriber moving track playback function.

2.3.4 Feature Database

A feature database contains information about the geographic distribution of network signals. A single
geographic location is usually covered by multiple cells. The cell signal strength and other cell
information constitute the signal feature of this geographic location.
Feature databases are the key input of the feature matching geo-location technology. Feature database
accuracy is essential for ensuring feature matching accuracy. Feature databases are generated using the
propagation prediction function (simulating the signal strength from cells to locations) or generated and
updated based on DT and A-GPS data. The generation and update methods improve consistency
between the feature database and data on the live network, further improving the location precision.

2.4 Main Geo-location Technologies in Industry

Currently, the MR-based geo-location methods in industry are classified into the following types:
 GPS/A-GPS location
GPS/A-GPS modules embedded in terminals measure, calculate, and report MR position
information. This method has high location precision, but it requires terminal support. A-GPS

2020-02-21 Huawei confidential. No spreading without permission. Page 10 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

location requires auxiliary devices on the network side, and the location fails when indoor UEs fail
to receive satellite signals. Therefore, in large-scale location scenarios, A-GPS location is used as
an auxiliary method for improving the location precision.
 Feature matching location
Feature matching location is also referred to as Database Correlation Method (DCM) or fingerprint
technology. The technology constructs a feature database based on the level or SINR information
of each cell at each geographical point within the network coverage range, compares the
information reported from UEs with the information stored in the feature database, and locate the
UE positions in grid points based on the matching degree. If the feature database information is
accurate, the location precision is high. This method applies to all RATs and most network
scenarios. Feature information in the feature database can be obtained through simulation and can
be constructed or corrected based on DT data, A-GPS data, and Wi-Fi. A three-dimensional feature
database can be constructed to support three-dimensional location.

 Ranging-based geometrical location

This method measures the distance between a UE and a reference object and use an algorithm to
calculate the two- or three-dimensional coordinates. The calculation can be performed on the
network side with measurement data provided by UEs (referred to as UE-assisted calculation) or
performed on the UE side (referred to as UE-based calculation). The UE-based calculation method
can be categorized into three types:
− Path loss–based model location: Path loss is calculated based on the signal strength. (Path loss
refers to the difference between the known pilot power and the pilot receive power measured by a
UE). When a UE receives signals from multiple cells, distances between the UE to the cells are
measured based on a propagation mode, and then the UE is located based on trilateral or
plurilateral geometric calculation. Because of the propagation environment complexity (building
blocking, reflection, or diffraction), the deviation of the distance calculated using path loss is big.
The location precision is low.

− Propagation delay location: Radio signals are transferred at a constant speed (light velocity c). The
propagation distance L is calculated using the formula: L = c x t. The distance is derived based on

2020-02-21 Huawei confidential. No spreading without permission. Page 11 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

propagation delay information. The location method measures delays or relative delays of multiple
cells (greater than or equal to three non-co-site cells), and determines UE positions through multi-
circle intersection or multi-hyperbolic intersection geometric calculation. OTDOA, UTDOA, and
E-TOD in UMTS and LTE 3GPP specifications are all developed based on this location method.
However, this method requires base station synchronization (by adding LMU devices) or high-
precision measurement functions on UEs. These device and functions are rarely used in the market
and are hard to promote (the LTE market is to be ripened). This method attenuates synchronization
requirements, referred to as TOA. It uses the absolute-time triangle location algorithm to calculate
positions. Because this method strictly requires concurrent signal measurement of multiple sites, its
application scenarios are limited and it is usually used as an auxiliary location method.

− AOA-based location: This method does not require base station synchronization, but intelligent
antennas for measuring incident angles should be installed on base stations. It calculates the Angle
of Arrival (AOA) and then uses the triangle measurement method to determine UE positions. This
method applies to only TD-SCDMA networks. This method has the following problems:
 If the distance between a UE and a base station is long, the location distance deviation will be
large due to the angle measurement error.
 In non-line-of-sight scenarios, the location deviation is large.
 This method is not suitable for indoor location. It is suitable for suburban areas with few
multipath signals.

 Cell ID–based topological-centroid geometrical location

Based on the site position of a cell ID in an MR, this method draws a polygon and then calculates
the centroid location of the polygon to determine the position whether the UE makes a call. With
Cell-ID signal strength used as a weighting factor, this weighting-based topological-centroid
location method can improve the location precision. Pseudo-range measurement or calculation is
not required. Besides basic engineering parameter information, such as cell site addresses and
azimuths, this method does not require extra input. Compared with the path loss–based model
location method, its location precision is relatively low, but it takes low costs and has high location

2020-02-21 Huawei confidential. No spreading without permission. Page 12 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

efficiency. In suburban areas with a lot isolated sites, the precision can be degraded only to the cell
level.

 OTT location
This method provides location services for a large number of users based on Internet apps. It uses
app software to extract valid subscriber position information, and uses subscriber identifiers to
correlate MRs. Longitude and latitude information reported by apps may be measured based on
different coordinate systems, and disturbance data may be reported. Therefore, valid OTT
longitude and latitude information can be obtained only after coordinate conversion and calibration
and data cleansing. Most apps use GPS/A-GPS, Wi-Fi, and base stations to locate UEs. The
precision of OTT longitudes and latitudes is high. Probes need to be deployed for collecting user-
plane xDR data.
Table 1.1 describes the location technologies and precision of the main third-party manufacturers or
vendors in industry.

Table 1.1 Location technology and precision summary in industry

Vendor Location Method Declared Location Precision

Groundhog A self-developed chaos theory algorithm is In competitive analysis materials, the

used to identify subscriber behavior location precision is around 100m@67%
features (moving/static and based on publicity in the early period
indoor/outdoor), and feature databases are (with grid granularity used as the
constructed in iterative mode to perform precision). The competitive and
feature matching location. comparison result in 2012 showed that
the precision is lower than Huawei.
Arieso A series of location technologies are It was declared that the precision had
provided based on different scenarios and reached 80m@67% in the early period.
RATs. The key location technologies The average precision in competitive
include: analysis materials is 100m@67%.
● Geometric location based on a field
strength path loss model
● Feature matching location based on
feature signatures iteratively generated
from terminal data
● Engineering parameter check and
iterative location based on confidence
probability density functions
Actix It is inferred from the location effect in No official publicity information is

2020-02-21 Huawei confidential. No spreading without permission. Page 13 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

publicity materials that a cell ID+delay obtained. The average precision in

technology is used. competitive analysis materials is
80m@67%.
Ericsson It is inferred from the location effect in According to the publicity materials in
publicity materials that a cell ID+delay the early period, the precision is around
technology is used. 150 m (unclear precision definition). It is
mentioned in official materials that the
more the location points are, the higher
the precision is.
Nokia A mixed location technology based on cell Urban area: 112-188m@67%
(NICE) ID, delay, and feature matching data is Suburban area: 188-284m@67%
used. Three-dimensional location is
supported.
ZTE TD-SCDMA: AOA+six-reference-point Indoor: cell ID (20 m to 1000 m)
location City: A-GPS (5 m to 30 m); cell
UMTS/LTE: Cell ID+delay+A-GPS mixed ID+delay (100 m to 1000 m)
location Suburban/Rural area: A-GPS (5 m to 10
m); cell ID+delay (200 m to 20000 m)

Every technology has its limitations, such as high costs, limited application scenarios, and technical
barriers. Location methods depending on terminal capabilities and auxiliary network devices can only
be used as auxiliary approaches due to their commercial use limitations. MR-based geo-location is
mainly used in network O&M optimization. A suitable location technology should be selected with the
precision, cost, and application scenario requirements taken into consideration. The mainstream location
technologies are feature matching location and geometric location based on a field strength path loss
model. Other technologies, such as the delay-based method, are auxiliary approaches for improving the
precision within a specific scope.

Figure 5.2 Precision analysis

2020-02-21 Huawei confidential. No spreading without permission. Page 14 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

2.5 Geo-location Technologies in the Discovery

2.5.1 Location Principle Overview
The Discovery uses three key processing modules to perform geo-location calculation:
 Calling position location
Based on the data resource input, perform mixed location by automatically degrading the location
method priorities in the following sequence: A-GPS location, indoor distribution location, RTT
location, feature matching location, and WCCL location. The system calculates the confidence of
each location method and selects the location result with a high confidence level.
A-GPS location and RTT location requires many external factors and can only be used as auxiliary
approaches. The Discovery mainly uses the following two location methods:
− Feature matching location: feature database model–based signal fingerprint matching technology
− WCCL location: topological-centroid geometrical location technology based on cell IDs and signal
strength weighting
 Subscriber environment feature analysis
The Discovery constructs feature databases based on indoor/outdoor calling environments. It
differentiates indoor and outdoor calls and estimates the moving speed of subscribers in calls based
on the wireless environment features, subscriber behavior features, and subscribers' geographic
positions.
 Map matching and calibration
Based on subscribers' calling behavior features, environment analysis results, and electronic map
information, the Discovery calibrates location results. For example, it matches indoor calls with
surrounding buildings and calls on roads with surrounding roads.
Geo-location principles have been briefly introduced in section 2.4"Main Geo-location
Technologies in Industry." The following sections make detailed technical clarification on each
location processing module of the Discovery.

2020-02-21 Huawei confidential. No spreading without permission. Page 15 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Figure 5.3 Discovery geo-location principle overview

2.5.2 A-GPS Location

Location Principles
A-GPS location in the Discovery is a generalized concept. High-precision longitude and latitude
location results contained in MR data sources reported by NEs are directly used.
 If the difference between the longitude and latitude location and the cell centroid based on MR
measurement exceeds a certain threshold, the A-GPS location result is considered to be abnormal
and is deleted.
 If there are multiple valid longitude and latitude records for the same call, the valid longitude and
latitude records are added to abnormal longitude and latitude records through interpolation based

2020-02-21 Huawei confidential. No spreading without permission. Page 16 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

on subscriber moving features. The abnormal records include abnormal-result deletion records and
records that are not reported to the A-GPS.

Figure 1.1 A-GPS MR flow

Table 1.1 Application of longitude and latitude information contained in MR data sources
for each RAT
RAT Application of Longitude And Latitude Information in MR Data Sources

GSM Not supported

UMTS 1. Only Huawei data sources are supported:
● Intra-frequency periodic or event MR blocks are mainly used. Location results of
LCS measurement blocks are combined.
2. LCS location methods based on protocols:
● Cell ID+RTT: The LCS provides only RTT measurement values. The Discovery
calculates the location results. For details, see section 2.5.4"RTT Location."
● OTDOA: The location depends on terminal capabilities. In this scenario, MR LCS
measurement blocks do not directly provide location results.
● A-GPS: The location depends on terminal capabilities. In some scenarios, location
results can be provided under certain conditions. For details, see Dependencies and
Restrictions.
● UTDOA: UE uplink signal receiving delays of receiving devices at difference
positions are used for location. Huawei NE devices are not supported.
LTE 1. Only Huawei data sources are supported:
● Longitude and latitude information in intra-frequency MDT data or event MRs
reported by MDT UEs is directly used.
● Intra-frequency periodic or event MR blocks are mainly used.
PrivateUEPositioning location results are combined.
2. MDT data is mainly used in GPS location.
3. PrivateUEPositioning location methods:
● E-CID: The location is based on cell IDs.
● OTDOA: The location depends on terminal capabilities. For details, see
Dependencies and Restrictions.
● A-GPS: The location depends on terminal capabilities. For details, see
Dependencies and Restrictions.

2020-02-21 Huawei confidential. No spreading without permission. Page 17 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

 Huawei RNC devices are deployed with two LCS location processing systems. One is for
commercial LCS service feature sales. The CN responds to location requests initiated based on LCS services in
real time and performs location as needed. The other is used by MR LCS measurement blocks. The RNC
simulates CN devices to periodically initiate location requests and uses an MR switch to control the period.
 When using the location function of MR LCS measurement blocks, the RNC reuses the computing
processing module of LCS service features. This location function only achieves the following location modes:
UE-based A-GPS location, standalone location (GPS), and UE-assisted cell ID+RTT location.

Dependencies and Restrictions

1. Dependencies of UMTS A-GPS location
(1) Enable and activate the A-GPS Based LCS feature on Huawei RNC devices.
On the RNC MML command execution interface, run the DSP LICUSAGE command to
check whether the Allocated value corresponding to the license ID LQW1AGPS01 is
greater than 0. If it is, the feature has been enabled. If it is not, you need to apply for the A-
GPS license.

For details about feature enabling, see section 9.3 "Configuring A-GPS Based LCS" in LCS
Feature Parameter Description for UMTS. You can obtain the document at the following
link (RAN19.0 used as an example):
http://support.huawei.com/hdx/hdx.do?
docid=DOC1000257970&lang=en&id=library_change_preview&from=HedExLite&client
Width=1904&browseTime=1495799845947
(2) Run MML commands to manually enable MR LCS measurement blocks.
a) Configure an RNC-level overall algorithm switch:
SET UMRCTRL:
MRTRAFFICTYPE=TRAFFIC_AMR-1&TRAFFIC_VP-1&TRAFFIC_BE-
1&TRAFFIC_OTHER-1, MRINTRAFREQPERIOD=D12,
MRRPRTCFGSWITCH=MR_PRD_INTRA_FREQ_MEAS_SWITCH_1-
1&MR_EVT_INTRA_FREQ_MEAS_SWITCH_2-
1&MR_PRD_INTER_FREQ_MEAS_SWITCH_3-
0&MR_PRD_INTER_RAT_MEAS_SWITCH_5-0&MR_DL_BLER_MEAS_SWITCH_7-
0&MR_UE_TX_POWER_MEAS_SWITCH_8-0&MR_UL_SIR_MEAS_SWITCH_9-
0&MR_LCS_MEAS_SWITCH_10-1&MR_RACH_MEAS_SWITCH_11-0,
MRLCSMETHODSWITCH=LCS_AGPS_UEBASED_SWITCH_1-
0&LCS_AGPS_STANDALONE_SWITCH_2-1&LCS_CELLID_RTT_SWITCH_3-0,
MRLCSPERIOD=D12;

2020-02-21 Huawei confidential. No spreading without permission. Page 18 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

 The commands above are used to show the settings of MML parameter switches related to A-GPS
MR enabling (the red parts). Measurement blocks enabled by the MR collection configuration switch
(MRRPRTCFGSWITCH) should be consistent with data source subscription on the Discovery.
 Up to RAN19.0, only one type of the location values between A-GPS location and RTT location can
be reported and output by using the MR LCS measurement block location switch of RNCs.
b) Set the MR scope control switch.
RNC-level and cell-level MR scope control switches are supported.
SET URNCMRSCOPE (this command is used to set the RNC-level MR scope control
switch):
 MRScopeCtrl: indicates whether the RNC-level MR function is supported.
− ON: the MR collection function is enabled.
− OFF: the MR collection function is disabled.
− DEPEND_CELL_SWITCH: the function depends on the MR switch of each cell.
SET UMRSCOPE (this command is used to set the RNC- or cell-level MR switch):
 If ScopeType is set to ByRnc and MRScopeCtrl is set to ON or OFF, the switch
specifying whether the MR reporting function of all cells under this RNC is set to ON
or OFF. If ScopeType is set to ByCellId, the value set for MRScopeCtrl is invalid.
 If ScopeType is set to ByRnc and MRScopeCtrl is set to
DEPEND_CELL_SWITCH, the value set for MRScopeCtrl is valid when
ScopeType is set to ByCellId.
RAN16.0 SPC620 and later:
 Turn on the switch for all cells under an RNC:
SET URNCMRSCOPE:MRSCOPECTRL=ON;
 Turn on the switch for certain cells under an RNC:
SET URNCMRSCOPE:MRSCOPECTRL=DEPEND_CELL_SWITCH;
SET UMRSCOPE: ScopeType = ByCellId, CellId = XXX, MRScopeCtrl = ON;

 The MR switch is off by default in a newly added cell. The switch needs to be turned on. Otherwise,
MRs generated within the cell will not be recorded.
 In RAN15.0 and later, MR switches within specific scopes for different application types are added
to NEs. The logic relationship between the new switches and the original switch is "or". The MR collection
function is enabled after any of the switches is turned on. The query command is LST UMMMRSCOPE.
(3) Run MML commands to manually set the A-GPS MR problem workaround algorithm
switch.
Refer to problem descriptions in UMTS A-GPS Problem and Maturity Summary. Terminal
compatibility problems are resolved mainly by setting the A-GPS MR whitelist. A-GPS MR
data is collected from only terminals without problems or with workaround measures.
The latest A-GPS MR whitelist needs to be obtained from the HQ, and internal algorithm
switch operations need to be performed on related devices. You are advised to contact
frontline device O&M engineers and the TMO of each region to conduct strict review to
reduce risks.
2. UMTS A-GPS Problem and Maturity Summary
(1) Terminal compatibility problems
 Increase of terminal power consumption

2020-02-21 Huawei confidential. No spreading without permission. Page 19 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

The GPS module enabled on a terminal in the connected state cannot be disabled after
the terminal accesses the idle state. Subscribers complain on high power consumption.
This problem mostly occurs on Samsung terminals.
 Terminal GPS icon display
When a terminal responds to an NI request, the GPS icon is turned on. Subscribers
complain on sensitive location information collection. This problem mostly occurs on
terminals integrated with MTK chips.
 GPS MR report failure
After A-GPS MR measurement control is issued, an A-GPS MR failure is directly
received. Whether position information is reported depends on the NI request
processing policies of each terminal manufacturer. Mainstream terminals, such as
iPhone 4S and later, and Samsung S4/S5/Note2/Note3 (LTE), do not report position
information.
 Abnormal terminal restart
QualComm MSM8X55 chips have a bug. When A-GPS MR data is cached on a UE in
non-CELL_DCH state, the UE will definitely be restarted after a CS/PS RAB setup
process, or DCCC-caused F2D(H) or P2D(H) process is started. As a result, KPIs such
as the PS R99 RAB setup success rate and PS R99 RAB call drop rate over the entire
network deteriorate.
Workaround measure on the NE side: In RAN16.0 SPC620 and later, add a whitelist
feature. It is recommended that all terminal compatibility problems be resolved by
setting the A-GPS MR whitelist.
− A-GPS MR blacklist and whitelist switch:
SET UALGORSVPARA:RsvSwitch13=RESERVED_SWITCH_13_BIT9-1
where
1: the whitelist function is enabled.
0: the whitelist function is disabled and the blacklist function is enabled.
− After the whitelist function is enabled, add whitelist terminals (for resolving the
icon problem):
ADD UIMEITAC:TAC_FUNC =
Special_User_Enhance,TAC=xx,Description="XX",RsvSwitch=
RESERVED_SWITCH_BIT20-1
− Enable single-time reporting to resolve the power consumption problem (only for
terminals with abnormal power consumption in the whitelist):
ADD
UIMEITAC:TAC_FUNC=Special_User_Enhance,TAC=xx,Description="XX",R
svSwitch=RESERVED_SWITCH_BIT21-1
where
1: When the switch is turned on, the RNC issues single-time MR A-GPS
measurement control messages. The purpose is to reduce power consumption.
0: When the switch is turned off, the RNC issues periodic MR A-GPS
measurement control messages.
(2) KPI deterioration due to heavy network load
 High RACH bit error rate
A-GPS MRs are reported through RACHs, resulting in increase of the RACH bit error
rate.
Workaround measure on the NE side: In RAN16.0 SPC620 and later, release A-GPS
measurement control messages before changing the state transition switch value to

2020-02-21 Huawei confidential. No spreading without permission. Page 20 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

D2F. In this way, no A-GPS MR messages are reported in the CELL_FACH state,
reducing the RACH load.
SET UCORRMALGOSWITCH:
CmpSwitch3=CMP_MR_AGPS_NOT_SUPP_STAT_TRANS_SWITCH-1 matching
bit30
SET UCORRMALGOSWITCH:
CmpSwitch3=CMP_REL_MR_AGPS_BF_D2F_SWITCH-0 matching bit32
 Abnormal group services
When UE_based A-GPS location is enabled, the size of UePos measurement control
messages is large, and a large number of measurement control signaling messages need
to be issued to NEs. As a result, in group service scenarios, the second RB SETUP
message cannot be issued to UEs, and UEs do not have sufficient time to respond to the
second service. Some UEs restart and the AMR completion rate decreases.
Workaround measure on the NE side: Do not use the UE_based A-GPS mode in LCS
location.
Summary: Based on optimization of several UMTS RNC versions and workaround measure
implementation on RAN16.0 and later, enabling A-GPS does not have significant impact on KPIs.
(1) The A-GPS MR whitelist solution is recommended. It can resolve the call drop, icon display,
and power consumption problems. In offices with this solution used, no subscribers
complain on the problems and no KPIs deteriorate.
(2) The number of whitelist solution reporting samples significantly reduces. In A-GPS feature
database construction scenarios, use the solution for at least two weeks.
(3) It is recommended that A-GPS MRs be collected in standalone mode to reduce impact on
the network side.
(4) The A-GPS MR whitelist solution is one of the main location solutions in the latest three
years.
 The solution requires terminal compatibility tests, and the IOT periodically updates the
test results.
 Mainstream terminals (iPhone and Samsung) do not report A-GPS MRs. Fewer and
fewer terminals will support this function in the future.
3. Dependencies of LTE A-GPS MR location
(1) MDT: The GPS location method is mainly used. The condition is that there are MDT
subscribers.
(2) A-GPS: No feature or switch is configured on Huawei eNodeBs to control the A-GPS
function. It is enabled by default, but the following UE and MME requirements must be met
before using the function:
 This function is supported only on Huawei CNs. The MME must be deployed with an E-
SMLC (mobile location center) and the location method configured on the E-SMLC
must be A-GPS.
 UEs support 3GPP R9, and GPS chips supporting A-GPS are installed in the UEs.
Refer to eRAN LCS Feature Parameter Description at the following link:
http://support.huawei.com/hdx/hdx.do?
docid=DOC1000119460&lang=en&id=library_change_preview&from=HedExLite&clientWidth=
1282&browseTime=1495847905307
4. LTE A-GPS MR Problem and Maturity Summary
(1) The Discovery already has software capabilities for supporting MDT location, but no delivery
experience is accumulated.

2020-02-21 Huawei confidential. No spreading without permission. Page 21 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

(2) The Discovery does not support A-GPS location based on PrivateUEPositioning event
because of its unavailable dependencies. The Discovery uses its own location results in this
scenario.

2.5.3 Indoor Distribution Location

Location Principles
The Discovery uses site type information in engineering parameters to detect whether there are indoor
distributed sites in the cell measured in an MR. If indoor distributed sites are detected in the primary
serving cell or a neighboring cell, the Discovery locates the MR using the weighting-based centroid
location method. The theoretical coverage scope of an indoor distributed site is smaller than that of a
micro site. The Discovery enhances the weight of indoor distributed sites to ensure that the location
deviation is within a reasonable scale.

MWC(x,y) is the weighting-based centroid located position, Wi is the weight, and BSi(x,y) is the site
address of the cell measured in the MR.

Figure 1.1 Indoor distribution location

Dependencies and Restrictions

The indoor/outdoor attribute of each site must be specified in engineering parameters.

2.5.4 RTT Location

Location Principles
RTT location of the Discovery is also called "cell ID+RTT location". The RNC requests all NodeBs
covering active set cells to finish RTT (RRT of measurement over antenna ports of NodeBs)
measurement and requests UEs to finish UE RxTx measurement (Rx/Tx time difference of
measurement over antenna ports of UEs) for related cells. The Discovery uses the measurement results
to calculate the TOA value.
TOA = (RTT – UE RxTx) / 2

2020-02-21 Huawei confidential. No spreading without permission. Page 22 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

TOA indicates the distance between a UE and a base station. The TOA track is a circle.
 If there is only one or two TOA values, a pure geometric calculation method is used, and the
intersection point is the UE position.
 If there are three or more TOA values, the RTT location calculation is converted to TDOA
calculation, and a hyperbolic calculation method is used.
TDOA = TOANbr – TOARef

Figure 1.1 RTT location principles

Dependencies and Restrictions

1. Dependencies of UMTS RTT location
(1) Enable and activate the Cell ID + RTT Function Based LCS feature on Huawei RNC
devices.
On the RNC MML command execution interface, run the DSP LICUSAGE command to check
whether the Allocated value corresponding to the license ID LQW1CIDLCS01 is greater than 0.
If it is, the feature has been enabled. If it is not, you need to apply for the license.
(2) Run MML commands to manually enable MR LCS measurement blocks.
Refer to Dependencies of UMTS A-GPS location. Turn on the RTT switch before configuring an
RNC-level overall algorithm switch. Other operations are consistent with A-GPS location
operations.
MRLCSMETHODSWITCH=LCS_AGPS_UEBASED_SWITCH_1-
0&LCS_AGPS_STANDALONE_SWITCH_2-0&LCS_CELLID_RTT_SWITCH_3-1,
2. UMTS RTT Problem and Maturity Summary
(1) The RTT location success rate and location precision are unstable:
 In RTT location, UEs must be in the CELL_DCH state.
 UE RxTx measurement values are classified into two types. The measurement
precision of type 1 is relatively low, but all UEs support the measurement. The
precision of type 2 is high, but few commercial UEs support the measurement.
Currently, the RTT location precision of the Discovery is mainly derived from
measurement of type 1.
 A large number of TOA values are required to ensure high location precision. To be
specific, a large number of neighboring cells need to be added in active sets and stably
stays there. The ratio of MRs with this soft handover status is not high in most cases.
The forcible soft handover algorithm of RNCs can increase the number of TOA values.
However, after the forcible soft handover switch is turned on, the number of signaling
processes increases. In an extremely severe wireless environment, UEs may fail to

2020-02-21 Huawei confidential. No spreading without permission. Page 23 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

respond ACTIVE SET UPDATE CONFIRM messages to the RNC in time. As a result,
the SRB resets over the Uu interface, and the RNC call drop rate slightly increases.
(2) Up to RAN19.0, only one type of the location values between A-GPS location and RTT
location can be reported and output by using the MR LCS measurement block location switch
of RNCs. A-GPS location precision is higher than RTT location precision, but A-GPS location
imposes a greater impact on the network than RTT location. It is recommended that RTT
location be enabled on networks where A-GPS is not used.

2.5.5 Feature Matching Location

Location Principles
The technical principles of feature matching location in Discovery are the same as those in industry. The
Discovery searches for the position with the most similar features with the current MR point in the
established feature database.
The key algorithm processing points are as follows:
 Accurate feature databases are constructed based on cell coverage information on the live network
 Feature matching scoring models are constructed based on neighbor relationships and neighboring
cell level values.
 The scope of the candidate grid set in matching calculation is determined based on delays and
angles.
 The optimal grid with the highest score is identified in the candidate grid set through iterative
operation.
N
M   k n  Cn  [CMatchCons tan t  abs ( Rxn  rxn )]
n 1

where
M is the matching value.
Kn is the cell matching weight.
Cn is the co-site attenuation coefficient.
CmatchConstant is the matching constant for adjusting the relationship between the neighboring cell
relationships and level matching degree.
Rxn is the level of cell n in the MR.
rxn is the level of cell n in the grid.
An M value is obtained for each grid. The larger the value is, the high the matching degree is. The
grid with the largest M value is the located point.

Figure 1.1 Feature matching location principles

2020-02-21 Huawei confidential. No spreading without permission. Page 24 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Dependencies and Restrictions

1. Dependencies of feature matching location
A feature database must be constructed in advance. The precision of the feature database has a
great impact on the location precision.
Currently, a feature database can be constructed based on coverage prediction or A-GPS.

Table 1.1 Feature database construction modes

Construction Description
Mode

Based on 1. Create a feature database file based on coverage prediction by using an external
coverage planning tool, such as U-Net, and then import the file into the Discovery.
prediction
2. The following diagram shows the data source preparation and application
process for generating a feature database using U-Net. For detailed operations,
see Feature Database Preparation Guide.

Based on A- 1. A-GPS or MDT data needs to be collected from NEs. Feature databases are
GPS constructed based on A-GPS data and MRs.
2. MRs and A-GPS data need to be collected. For details about A-GPS data

2020-02-21 Huawei confidential. No spreading without permission. Page 25 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

collection dependencies, see section 2.5.2"A-GPS Location."

2. Feature Matching Location Problem and Maturity Summary

(1) Feature matching applies to all RATs and most scenarios but problems may arise in the
following scenarios:
 Isolated-site and dual-site scenarios: Feature information about each position in the to-
be-searched area varies slightly because no network topology information and no
sufficient neighboring cell information are available for reference. As a result, the
number of matched positions is large and it is difficult to ensure the location precision.
 Mountain scenario: If the propagation prediction method is used to generate feature
databases, height calculations in the current propagation model may be incorrect during
the prediction. Inaccurate level prediction results in low location precision.
 Seaside and lakeside scenarios: the located positions of the location algorithm may be
in the sea or lake. If the seaside and lakeside are rarely visited, increase the penetration
loss of these areas during coverage prediction when simulation feature databases are
used. In this way, MR location results will not fall in these areas.
 Scenario where there are many repeaters: Repeaters change the coverage of donor
cells. In software planning, antenna and loss parameters need to be accurately
calculated and configured for repeater simulation, and it is difficult to ensure the
accuracy in actual operations. Currently, the location algorithm cannot accurately
distinguish subscriber locations between donor cell coverage areas and repeater-
covered areas. As a result, location precision in both donor cells and repeater-covered
areas is low.
 RRU scenarios: RRUs are used between different buildings. The Discovery does not
record RRU data. The location precision is affected by the RRU distance.
(2) Maturity summary
 The precision of feature databases constructed based on coverage prediction
completely depends on the feature database input and construction experience. DT data
is needed to correct feature databases to ensure high-precision gains. The construction
costs, skill requirements, and operation costs are all high. Due to the limitations, it is
recommended that this feature database construction mode be used only when A-GPS
is unavailable, high-precision is required, and the delivery scale is small.
 Sufficient MRs and A-GPS data should be collected. (For details, see section
3.2"Location Precision.") As most data sources are collected from the network side, the
Discovery can automatically construct feature databases, significantly improving the
delivery efficiency and location precision. Currently, only GENEX Discovery
V200R002C30 provides the LTE MDT-based feature database construction capability.
No delivery experience is accumulated for outputting a problem and maturity summary.

2.5.6 WCCL Location

Location Principles
The WCCL location algorithm of the Discovery combines ranging-based geometric location and
weighting-based centroid location.
Ranging-based geometric location uses MR-measured delays (pr disguised delays constructed using
propagation models) and intersection points of azimuths to determine the candidate positions of UEs.
Weighting-based centroid location constructs polygons based on cells measured in MRs and calculates
topological centroids based on weights to determine the candidate position of UEs.

2020-02-21 Huawei confidential. No spreading without permission. Page 26 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

The Discovery determines the final UE positions based on the results of these two location methods.

Figure 1.1 WCCL location principles

Dependencies and Restrictions

MRs and engineering parameters are the main data resources for WCCL location. This is a light and
low-cost geo-location technology. The precision is strongly related to the inter-site distance and is
greater than 100 m in average (1/2 to 1/3 of the inter-site distance). Therefore, this location method is
suitable only in urban areas (the inter-site distance is less than 800 m). If the inter-site distance is greater
than 800 m, the precision will be low.
1. The topology structure between networks is roughly considered in the current algorithm.
Information such as the actual propagation environment and antenna information for accurate
analysis is not required. However, the site address and azimuth information in engineering
parameters must be accurate.
2. WCCL location is the last phase of the mixed location calculation of the Discovery. It ensures that
a position based on geometric location can be obtained when conditions of other location methods
are not met. Therefore, the precision is not even. In the following scenarios, the location deviation
is large:
 Cells sharing the same direction
When cells measured in an MR share the same direction, the location deviation is large. An
enhanced algorithm will be provided and verified in later versions.
 Dual-site scenario
For an MR that records only the signals of two base stations (especially when there are also only
two cells), there is a high probability that the position is located in the connection line between two
base stations. In the case of an excessive amount of data, straight lines are displayed on the traffic
map, affecting traffic analysis.
 Isolated site scenario
In this scenario, no additional topology is available for reference. Most isolated sites are
omnidirectional sites in rural areas. The location in this scenario may decrease the average location
precision on the entire network. The site location results are controlled by switches.

2020-02-21 Huawei confidential. No spreading without permission. Page 27 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

2.5.7 Subscriber Environment Feature Analysis

Location Principles
The subscriber environment feature analysis of the Discovery divides MRs of a single call into multiple
call fragments and analyzes the feature attributes of each call fragment. The feature attributes include
where a call is made (indoor or outdoor), in which status a call is made (static or mobile), and
estimation of the current moving speed of subscribers.
The key action for this algorithm is to construct indoor and outdoor call feature models, identify indoor
or outdoor calls based on wireless network features, subscriber behavior features, and subscriber
geographic locations extracted from CHRs and MRs, and estimate the moving speed of subscribers
during calls.

Figure 1.1 Indoor and outdoor call feature model

Extract certain fields from the MRs and CHRs to form features that adapt to the following dimensions
and identify the tendency for a call to occur in an indoor or outdoor area. If any of the following feature
dimensions is not met, the indoor and outdoor identification is not performed.
The following describes related feature dimensions and calculation principles:
2. Level feature
(1) Level of signals received during a call: Around a position in a macro site area, signals of an
outdoor call are stronger than signals of an indoor call.
(2) Stability of signals received during a call: Signals of an indoor call are more stable than
signals of an outdoor call because the indoor transmission environment is more stable and
indoor signals have fewer changes.
3. Call feature
(1) Call connection duration: A long call is more likely to occur in an indoor area.
(2) Call period: A call at night is more likely to occur in an indoor area.
(3) Call inertia: When a subscriber makes calls at a specific position for a long time, the
subscriber's common position can be identified based on the behavior. The call is more likely
to occur in an indoor area.
4. Service feature

2020-02-21 Huawei confidential. No spreading without permission. Page 28 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

A service feature distinguishes voice services, data services. Data services are more likely to occur
in an indoor area based on subscriber behaviors.
5. Moving feature
Moving features mainly refer to the subscriber moving distance and speed. Moving feature data is
obtained based on geo-location. If a subscriber is moving at a high speed or the subscriber's
position changes quickly during a call, the call is more likely to occur in an outdoor area.
6. Cell feature
(1) Type of serving cells: cells under a macro site or indoor distributed site. The purpose of
building indoor distributed sites is to serve indoor areas. Calls are more likely to occur in an
indoor area served by an indoor distributed site.
(2) Number of serving cells: If the number of serving cells is large during a call, the call is more
likely to occur in an outdoor area and subscribers in the call may move. If only one serving
cell is involved during a call, the call is more likely to occur in an indoor area because the
indoor cell is more stable.
7. Handover feature
(1) The handover feature can distinguish ping-pong handovers between two cells or among three
cells. If a ping-pong handover occurs during a call, there is a high probability that the
subscriber does not move during the handover. This feature can be used with other dimensions
to identify whether a call occurs in an indoor or outdoor area.
(2) Handover among multiple cells: If a handover occurs, a call is more likely to occur in an
outdoor area.
8. Neighboring cell feature
The number of cells whose signals are received by a UE during an indoor call is less than that of
cells whose signals are received by a UE during an outdoor call. A cell's signals can be received in
the outdoor area but may exceed the demodulation threshold in the indoor area and cannot be
received by a UE, which reduces the number of cells whose signals are received by a UE during an
indoor call.
The confidence level of each feature dimension is unique. A single dimension cannot ensure that
the identification is correct, and some dimensions even conflict with others. The conflict must be
considered and each dimension should have a proper weight to obtain a comprehensive
identification result.

Dependencies and Restrictions

This algorithm is used to determine whether a call occurs in an indoor area or an outdoor area based on
subscriber behaviors and features extracted from MRs and correlate the determined attributes with each
MR. This algorithm does not modify subscriber locations calculated in the location technologies
previously described. That is, the algorithm does not improve location precision directly.
With the existing average location precision, the Discovery cannot ensure all indoor calls are made in
buildings on the GIS map nor ensures outdoor calls are not made in buildings on the GIS map. If users
expect indoor calls to be made and outdoor calls not to be made in buildings on the GIS map, the map
matching and location calibration technologies must be used.

2.5.8 Map Matching and Location Calibration

Location Principles
The map matching and location calibration algorithms of the Discovery calibrate the following location
results based on subscriber behavior analysis results and geographic information of electronic maps:

2020-02-21 Huawei confidential. No spreading without permission. Page 29 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

1. Building matching: If determining that an MR is reported by a UE that is making an indoor call

based on the subscriber behavior analysis, the Discovery matches the MR to a building based on an
electronic map. If determining that an MR is reported by a UE that is making an outdoor call, the
Discovery moves the MR location to an adjacent building based on the Euclidean distance-based
grid index algorithm.

Figure 1.1 Building matching (left) and Euclidean distance-based grid index (right)

2. Road matching: For MRs of outdoor calls that meet high speed and long call conditions, the
Discovery projects and calibrates the MRs based on electronic map indexes and road curves which
are mostly similar to MR samples.
 High moving speed: Outdoor calls whose moving speed is greater than 30 km/h in more than
2/3 of the call duration (an outdoor call can be identified based on call feature)
 Long call: A long call is an MR sequence segment that meets the following conditions:
 MR identification interval is shorter than 42s.
 Call duration is longer than 2 minutes.
 The number of MRs is greater than 30.

Figure 2.1 Road matching

The formula for calculating the best road is as follows:

p1 p2 pq
Rmatching  Max{WRi , WRi ,. ., WRi }
i1 i1 i1

where, WR i indicates the weight of road Ri , Rmatching indicates the best matched road, q
indicates the total number of roads between two points, and p1 , p2 ,... pq indicate the numbers
of road sections on each road.

2020-02-21 Huawei confidential. No spreading without permission. Page 30 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

3. Outdoor non-road areas: For MRs of outdoor calls that do not meet high speed and long call
conditions (including MRs for calls with failed road matching), if these calls are made in buildings
or waters within the map scope, the Discovery calibrates MR locations of these calls to the
adjacent outdoor clutter type that can receive traffic.

Dependencies and Restrictions

The map matching algorithm performs additional correction or deviation on some call samples based on
MR location calculation and makes distribution of indoor and outdoor calls more appropriate by using
high-precision digitized electronic maps. This algorithm theoretically improves location precision with
reference to matched electronic maps.
This algorithm requires high-precision electronic maps (only Planet maps are supported now).
Otherwise, the calibrated location result may exert adverse effect on location precision. In addition,
location post-processing is added, and therefore the end-to-end processing delay is prolonged. The
details on dependencies are as follows:
1. Building matching and outdoor non-road matching
 Depend on precision of indoor and outdoor MR location analyzed by the subscriber
environment feature analysis module.
 Require the Planet map to contain the clutter layer and building vector layer. Map quality
determines building matching effects.
2. Road matching
 Depend on speed estimation by the subscriber environment feature analysis module. Road
matching has higher requirements on cross-NE CHR splicing for long call determination.
 Require vector (road) layers. Map quality determines road matching effects.
 Since high-speed moving long calls on roads are extracted, you are advised to only configure
roads (unidirectional double-lane roads) of a higher level than 3. Otherwise, the sample
analysis result will be affected.
 The existing location technology cannot distinguish road overlap scenarios, such as viaducts.
When the average road distance is less than the location tolerance, the road matching
precision does not increase significantly.
 In scenarios with great location errors or road loss, the Discovery cannot find appropriate
roads based on some MRs and gives up matching for these MRs.

3 Geo-location Technology Specifications

3.1 Location Delay
In the scenario where DPU and CMA clusters are deployed, cross-NE records generated by a call are
out of order during collection. Because some location algorithm processing modules (filter, subscriber
call behavior characteristics analysis, and GIS map matching calibration) process data based on
continuous MRs generated during a call, the RTL locates a UE after a delay (current cache threshold: 6
minutes). An RH2288 server processes 300,000 records per second. With the appropriate cluster
deployment, the data processing delay of the RTL is less than 10 minutes.

3.2 Location Precision

Error: Reference source not found lists the location precision specifications for each location method.

2020-02-21 Huawei confidential. No spreading without permission. Page 31 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Table 1.1 Location precision for each location method

Scenario Dense Urban Urban Suburban Rural

Inter-site distance <400m 400m~800m 800m~1500m >1500m

Assisted location AGPS/MDT 20m~25m 20m~25m 20m~25m 20m~25m

based on high-
precision data OTT 20m~40m 20m~40m 20m~40m 20m~40m

high-precision data
Feature matching >80% >80% >80% >80%
coverage rate
location based on
High-precision Location deviation of
database 1/2 Inter-site
traffic hotspots 50m~80m 50m~80m 80m~140m
distance

Number of non-co-site
>2 >2 >2 N/A
cells measured in MRs
WCCL location
Location deviation of
traffic hotspots 100m~150m 120m~200m 200m~300m N/A

Restrictions:
1) Minimum number of MRs to be collected at each position >=5000
2) Maximum site position deviation <15m
3) Maximum antenna azimuth deviation <10°
4) Not applicable for SFN network scenario
5) OTT is only applicable for China area

3.3 Indoor and Outdoor Differentiation Accuracy

When the average location precision reaches 80 to 120 meters, the indoor and outdoor differentiation
accuracy is 85%.

 LTE network without applicable IMSI will lead to an accuracy decrease by 5%~10%

 Outdoor strong occlusion and indoor without occlusion scenario will lead to an accuracy decrease
by 10% approximately

3.4 Road Subscriber Identification Accuracy

When the average location precision reaches 80 to 120 meters, subscriber identification accuracy for
long calls with high moving speeds on roads is 85%.

2020-02-21 Huawei confidential. No spreading without permission. Page 32 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

3.5 Location Success Rate

When no engineering parameters are missing and the exception deletion switch is disabled, the MR
location success rate is up to 95%.
3. MR samples involved in MR location success rate
 GSM: common MRs (excluding EMRs, namely, MRLog[message_type=40,L3.Pdormsg=6])
 UMTS: intra-frequency periodic MRs and intra-frequency event MRs. If MRs are invalid,
they are not counted in success rate calculation.
 LTE: intra-frequency periodic MRs/MDTs, and event MRs
4. Exception deletion switch
 Some abnormal points may exist because the input measurement information is insufficient
for some scenarios (such as the isolated-site scenario) or the input engineering parameters is
incorrect during MR calculation. The average location precision decreases due to these
abnormal points. Subscribers can still use these points or do not use them to locate UEs.
 This abnormal-location scenario relies on network features and the accuracy of subscriber
input information, such as engineering parameters, feature databases, and maps. Specific
standards cannot be provided. Therefore, you are advised to turn off the exception deletion
switch during location success rate verification.
5. Location failure scenarios
 Engineering parameter loss: When engineering parameters of all cells in MRs are lost, MR
location (including MR correlation) fails. The engineering parameter loss possibility varies
depending on the input engineering parameter integrity. Therefore, MRs for which the
location fails should not be counted in location success rate verification.
 Overlapping coverage: When the primary cell measured in an MR is far away from the
neighboring cells (for example, the distance between a primary cell and a neighboring cell is
three times greater than the average inter-site distance of the primary cell) and neighboring
cell levels are invalid or unavailable in engineering parameters, MR location fails. This issue
occurs at a low probability. (In most cases, the location succeeds but the results are abnormal.)
 Delay disguise failure: When only the WCCL fast location method can be used, delay
information is required for topological calculation. If no delay information is available in
MRs, the system needs to generate a disguised delay based on engineering parameters. If the
site height in the input engineering parameters is 0, delay calculation fails. This issue occurs
at a low probability. (You can modify the site height in engineering parameters to solve this
issue.)
 LTS SFN scenario: In this scenario, a cell has multiple site addresses. If the corresponding
cell sector information is invalid, you cannot obtain the site addresses and the location fails.
This issue occurs at a low probability. (Most location failures are caused by NE measurement
problems.) Multi-site cells support only the LTE SFN scenario. If MRs in other scenario are
collected, location failures are not counted in success rate calculation. (The other scenarios
here include the remote GSM repeater scenario, UMTS RRU or multi-antenna scenario, and
other non-SFN LTE multi-sector combination scenarios.)

4 Terms
Use the Huawei standard terms in the following table uniformly in software development,
documentation, and technical communication.

Huawei Standard Term Other Version in Industry

geo-location

2020-02-21 Huawei confidential. No spreading without permission. Page 33 of 34

 GENEX Discovery Geo-location Technical Paper INTERNAL

Huawei Standard Term Other Version in Industry

position location
accuracy precision
deviation
probability
confidence level
accuracy of single geo-location
traffic accuracy
(feature) pattern matching Database Correlation Method (DCM)
feature fingerprint
feature database 1. PSD-predicted signature database
2. fingerprint database
3. property matrix
4. character database
5. geographic property database

grid BIN
lattice
grid granularity
create (feature database)
update (feature database)
predict
propagation model
(propagation model) calibration
engineering parameter

References
1. GUL Geo-location Technological Algorithm Specifications V4.0 by Li Xi, 2015-11-09
2. Indoor and Outdoor Differentiation Algorithm Design Specifications by Huang Xiecheng, 2015-
11-01
3. WCDMA RNC SMLC TDOA Geo-location Calculation Algorithm Analysis Report by Xu Bin,
2006-10-30
4. UMTS_AGPS_MR Geo-location Solution Technical White Paper by Li Chunguang, 2015-07-05

2020-02-21 Huawei confidential. No spreading without permission. Page 34 of 34