WCDMA RNO RF Optimisation Guidance-20040719-A-1.0.pdf

Huawei Technologies Co. Ltd.

Product Version Confidential level

V100R001 For internal use only

Product name: WCDMA RNP Total 27 pages

WCDMA RNO RF Optimisation Guidance

For internal use only

Prepared by URNP -SANA Date 2004-04-13

Reviewed by Date

Reviewed by Date

Granted by Date

Huawei Technologies Co., Ltd.

All rights reserved

WCDMA RNO RF Optimisation Guidance For internal use only

10-4-30 Confidential , Total 27

Revision Record

Date Revision version

Change Description Author

2004-04-13 1.00 Initial release Jamal



Contents

1 INTRODUCTION .................................................................................................................................... 6

2 RF OPTIMISATION OVERVIEW ....................................................................................................... 6

CLUSTERS ............................................................................................................................................................ 6 DRIVE ROUTES .................................................................................................................................................... 7 TOOLS ................................................................................................................................................................. 7 OPTIMISATION TARGETS ..................................................................................................................................... 7 OPTIMISATION SOLUTIONS .................................................................................................................................. 8 OVERALL PROCESS .............................................................................................................................................. 8

3 RF ANALYSIS APPROACH .................................................................................................................. 9

CELL DOMINANCE ............................................................................................................................................. 10 CPICH COVERAGE (RSCP) ............................................................................................................................... 11 INTERFERENCE (CPICH EC/IO) ......................................................................................................................... 12 UPLINK COVERAGE ........................................................................................................................................... 13 PILOT POLLUTION.............................................................................................................................................. 14 ESTIMATED ACTIVE SET SIZE ............................................................................................................................ 15 NEIGHBOUR LIST VERIFICATION ........................................................................................................................ 16 UE SHO PERFORMANCE ................................................................................................................................... 17 DROP CALLS ...................................................................................................................................................... 17

4 ANALYSIS SUMMARY ....................................................................................................................... 19

5 PROBLEMS & SOLUTIONS (TBA) ................................................................................................... 25

6 SUMMARY ............................................................................................................................................ 25

7 APPENDIX A: DEFAULT COLOUR SCHEMES IN ACTIX (TBA) ............................................. 25

8 APPENDIX B: ACTIX PREFERENCES (TBA) ............................................................................... 26

9 APPENDIX C: ACTIX THRESHOLDS (TBA) ................................................................................. 26

10 APPENDIX D: EXAMPLE ACTIX CELLREF FILE (TBA) .......................................................... 26



Tables

TABLE 1: CURRENT OPTIMISATION TARGETS. ......................................................................................................... 7 TABLE 2: PILOT POLLUTION RESULTS. ................................................................................................................... 15 TABLE 3: EXAMPLE OUTPUT OF THE NEIGHBOUR LIST VERIFICATION. .................................................................. 17 TABLE 4: SOFT HANDOFF SUCCESS RATE. .............................................................................................................. 17



Figures

FIGURE 1: HIGH LEVEL RF OPTIMISATION STEPS. .................................................................................................... 9 FIGURE 2: SCANNER SCRAMBLING CODE PLOT. .................................................................................................... 11 FIGURE 3: SCANNER CPICH RSCP PLOT. ............................................................................................................. 12 FIGURE 4: SCANNER CPICH EC/IO. ....................................................................................................................... 13 FIGURE 5: UE TX POWER. ...................................................................................................................................... 14 FIGURE 6: ESTIMATED ACTIVE SET FROM SCANNER DATA. ................................................................................... 16 FIGURE 7: DROP CALLS AND SCRAMBLING CODE PLOT. ......................................................................................... 20 FIGURE 9: DROP CALL 1 (UE VS. SCANNER BEST SERVER). ................................................................................... 23 FIGURE 10: DROP CALL NUMBER 1 (ACTIVE & MONITORED SETS AT TIME OF DROP). ........................................... 24 FIGURE 11: RSCP COVERAGE FROM SC018. ......................................................................................................... 24 FIGURE 12: DROP CALL NUMBER 2 (DL SIR, EC/IO, UE TX POWER & DL BLER AT TIME OF DROP) ................... 25



1 Introduction

This document provides a detailed discussion of the RF (Cluster) optimisation phase of the

3G radio network.

It is expected that all integrated sites will undergo the Single Site Verification process as

outlined in [Single Site Guidelines]. The objectives of the single site verification are to ensure

there is no installation or parameters’ related faults with any of the sites.

Once all the sites in a given area are integrated and verified, RF (or Cluster) optimisation

could begin. This refers to the main phase of optimisation which aims at optimising coverage

while in the same time keeping interference and pilot pollution under control over the target

area.

This document presents a step-by-step approach for the analysis of drive survey data

collected using Agilent Scanner and Qualcomm test UE. The analysis is being done using

Actix Analyzer.

It should be emphasised that the RF optimisation will be an ongoing activity and will need to

be revisited as traffic increases in the network and as new sites are deployed.

In addition, as the network matures, the optimisation process should take into account

statistical data and key performance indicators collected throughout the network.

In this document, sample data from one part of the network is used to illustrate the various

analysis techniques. Although this data is from an incomplete cluster of sites, it is sufficient

for the purpose of this document.

The layout of this document is as follows: Section 2 provides an overview of the RF

optimisation process while section 3 outlines the analysis steps need to identify various RF

issues. A summary of the analysis is provided in section 4. The appendixes at the end of the

document contain various Actix related configuration data that should be used when

performing data analysis to ensure consistency throughout the network.

2 RF Optimisation Overview

Clusters

Due to the nature of UMTS (i.e. the inter-dependence of coverage and capacity and the

frequency reuse factor of one), it is crucial that the RF optimisation is carried out for groups

or clusters of sites rather than on single site basis. This will ensure that the impact of all the

sites in a given area on coverage as well as interference is taken into account.

Prior to any changes to a specific site, detailed analysis of the impact of such change on the

adjacent sites must be considered to ensure there could be no adverse effects on the area.



Drive Routes

The cluster drive surveys should include the coverage areas of each cell and all the major

roads and streets as well as any other important locations.

It is essential to use identical drive routes prior and post any optimisation changes in order to

accurately quantify the impact of such changes.

Tools

The drive surveys should be conducted using the Agilent scanner and Qualcomm UE in

continuous AMR call.

The use of the UE data is important to the RF optimisation as it provides additional

information that could help identify certain issues, such as: Uplink coverage problems,

missing neighbours, too many soft handoff events, etc.

The scanner will be using an externally mounted antenna while the UE will be kept inside the

car in the same location for each drive test.

For post-processing the data, Actix Analyzer will be used as outlined in this document. For

consistency, Actix should be configured as outlined in Appendixes A-D of this document.

Optimisation Targets

The targets and thresholds in Table 1 below are for use in the early phase of network

optimisation and are applicable to the scanner measurements unless otherwise stated.

For description of terms such as pilot pollution, please refer to the next section.

Item Requirements Comments

CPICH RSCP Target ≥ - 85 dBm Corresponds to outdoor

measurements. Minimum -95 dBm

CPICH Ec/Io Target ≥ -8 dB

Applicable for unloaded

network.

Minimum - 14 dB

Active Set size

(estimated) Target ≤ 3 Based on scanner data.

Pilot pollution

Max % < 10 % % of time a cell is seen as a

pilot polluter.

Threshold 8 dB Relative to best server when

cell in not in Active Set.

UE Tx power Max <15 dBm Assuming 21 dBm max.

SHO Success

rate Target >95 % For e1a, e1b & e1c

Table 1: Current Optimisation Targets.



Optimisation Solutions

Most of the coverage and interference issues could be resolved through adjusting site’s

parameters, such as (in order of priority):

Antenna tilt

Antenna azimuth

Antenna location

Antenna height

Antenna type

Site location

New site

Detailed discussion of the different optimisation problems and solutions is provided in

section 5.

Overall Process

The high-level Cluster process is depicted in Figure 1 below. As can be seen, the process can

be quite iterative and therefore careful analysis is required to ensure the optimum solution is

achieved with the minimum number of iterations.



Figure 1: High level RF optimisation steps.

3 RF Analysis Approach

This section presents various plots produced using Actix along with a description of the

analysis approach.

It should be noted that the effect of an RF problem will typically be seen in a number of these

plots and therefore a summary of the analysis is needed to conclude on the fundamental

causes of any failures. This summary will be provided in the next section.

Yes

No

Drive Test

Start

Identify any

RF Issues

Identify candidate

cells for changes

Identify nature of

required changes

Determine amount of

change

Implement changes

Repeat Drive Test

End

Problem

Resolved?



During the analysis of the individual plots, any observed issues should be marked to facilitate

further investigations and comparisons with other plots.

Cell Dominance

One of the first plots that should be analysed is the scrambling code plot as shown in Error!

Reference source not found..

The plot should visually be checked for:

Cells with no dominance at all:

This could indicate that a site was not radiating during the drive survey (this should be

confirmed from the network stats).

If a cell is suspected to not have been radiating during the test, the problem must be

confirmed before proceeding with the rest of the analysis. (The drive survey will need to be

repeated if not all the cells were radiating).

Very poor dominance can also be caused by blocking of the antenna. If such a problem is

suspected, a site visit must be made to verify the antenna clearance.

Cells with either excessive or poor dominance:

This could be due to a high site or non-optimum antenna tilts.

Cells with too large dominance will be causing interference to adjacent cells resulting in

poorer capacity.

Areas of non-dominance:

This refers to areas where there is not a single clear dominant cell and where the best server

changes too frequently. Such conditions will result in excessive number of soft hand off

events reducing the system efficiency and increasing the probability of call drops.

UE vs. Scanner scrambling codes:

It is also useful to perform visual comparison between the UE and scanner SC plots.

Significant differences between the plots may indicate a missing neighbour or failed soft

handoff problem.

Any observed issues should be marked on the plot for further investigation and correlation

with other plots.



Figure 2: Scanner Scrambling Code plot.

CPICH Coverage (RSCP)

The RSCP plot should be analysed based on the thresholds presented in Table 1 which are

summarised below:

Good: RSCP ≥ -85 dBm

Fair: -95 dBm ≤ RSCP < -85 dBm

Poor: RSCP < - 95 dBm

The above levels are applicable for outdoor scanner measurements.

Areas of poor coverage as well as significant areas of fair coverage should be highlighted for

further investigation.

Example of too

many best

server changes

Too many pilots

leading to excessive

soft handoff



Figure 3: Scanner CPICH RSCP plot.

It is also useful to examine the RSCP coverage on per cell bases in order to highlight any

cells that have too large a footprint. An example can be seen in Figure 11 in section 4.

When comparing RSCP coverage from scanner and UE, it should be noted that the UE will

have lower levels as a result of the in-car penetration loss and differences of antenna gain.

Interference (CPICH Ec/Io)

In parallel with the analysis of RSCP coverage, the Ec/Io plot should also be analysed based

on the thresholds presented in Table 1, as follows:

Good: Ec/Io ≥ -8 dB

Fair: -14 dB ≤ Ec/Io < -8 dB)

Poor: Ec/Io < - 14 dB

The -8 dB threshold takes into account the expected future interference increase as a result of

increased traffic.

Areas of poor Ec/Io should be checked against RSCP levels as follows:

If RSCP levels are also POOR, then the fundamental cause of low is Ec/Io is poor

coverage

If RSCP levels are GOOD, this will imply strong system interference. Such scenario

could arise when two sectors are pointing at each other.

Example of

poor coverage



Areas of poor Ec/Io should be highlighted for further investigation. An example Ec/Io plot is

shown in Figure 4 below.

Comparisons of the Ec/Io plots from the scanner and UE should be made. Areas where UE

Ec/Io is significantly lower than that of the scanner may imply a problem of missing

neighbour or delayed soft handoff which can be associated with call drops.

Figure 4: Scanner CPICH Ec/Io.

Uplink Coverage

Error! Reference source not found. shows an example of UE Tx power. Any areas where

the UE Tx power is high should be highlighted as areas of possible poor uplink coverage that

require further investigation.

Areas of high Tx power should be compared to the CPICH plots to verify if the problem only

exists on the uplink.

An example

of poor Ec/Io

as a result of

poor coverage



Figure 5: UE Tx power.

Pilot Pollution

Within Actix, the Pilot Pollution Set includes all pilots that are not in the active set BUT are

within a certain margin of the best server (the margin is set to 8dB as listed in Table 1 above).

An example of pilot pollution results is provided in Table 2. This shows the % of time each

cell was seen as a pilot polluter. Cells which are frequently seen as polluters (e.g. >10)

should be marked and investigated.

These results should be used in conjunction with the Estimated Active Set Size, Figure 6,

which shows the locations that have too many pilots.

Example

of high UE

Tx power



SC Count % in Pollution Set

8 206 12.9%

9 165 10.3%

10 157 9.8%

11 156 9.7%

12 148 9.2%

13 135 8.4%

16 95 5.9%

17 94 5.9%

18 76 4.7%

19 74 4.6%

20 73 4.6%

21 56 3.5%

32 54 3.4%

35 39 2.4%

37 33 2.1%

43 19 1.2%

48 8 0.5%

53 7 0.4%

67 3 0.2%

80 2 0.1%

130 2 0.1%

Table 2: Pilot pollution results.

Estimated Active Set Size

Another useful measure of pilot pollution is by looking at the estimated active set based on

the scanner data. This plot is obtained by modelling the network soft handoff parameters

within Actix.

In order to see areas of excessive SHO candidates, the estimated active set size is allowed to

exceed maximum of 3.

Locations where there are more than 3 pilots in the active set should be marked and sources

of the interference should be identified. This can be done in conjunction with the Pilot

pollution analysis.



Figure 6: Estimated Active Set from scanner data.

Neighbour List verification

The neighbour list could be verified and optimised using the Neighbour List Verification tool

within Actix.

Prior to performing this analysis, the neighbour lists of each cell must be included in the

CellRef file used by Actix.

The tool compares the drive survey data against the neighbour list in the CellReff and then

provides the following recommendations for each cell:

Retain: This indicates that those neighbours have been confirmed from the drive

survey data.

Add: Missing neighbours (that’s neighbours seen in the drive test but not included

in the neighbour list)

Remove: These neighbours that were not measured but are in the neighbour list.

Table 3 below shows a typical example from running the Neighbour List Verification for one

cell (SC: 009).

It should be noted that careful consideration is needed prior to removing neighbours since the

Actix results are drive route dependent.

An example of

too many pilots

(SHO

candidates)



009 576 Retain 018 82 14.2%

Retain 010 46 8.0%

Retain 016 31 5.4%

Retain 032 20 3.5%

Retain 011 18 3.1%

Add 130 17 3.0%

Retain 021 17 3.0%

Retain 008 12 2.1%

Retain 020 6 1.0%

Retain 012 5 0.9%

Retain 017 2 0.3%

Remove 053 0 0.0%

Remove 019 0 0.0%

Remove 034 0 0.0%

Remove 037 0 0.0%

Remove 013 0 0.0%

Remove 051 0 0.0%

70548 Ajman Central 25.41204 55.447

Nbr SCSample

Count%Latitude Longitude

Sample

CountActionSC Cell Site

Table 3: Example output of the Neighbour list verification.

UE SHO Performance

The success rates for event 1a, 1b & 1c and can be obtained from Actix as shown in the

example below.

Number of Active Set Updates

Event Count

Event 1a - Cell Addition328

Event 1b - Cell Removal306

Event 1c - Cell Replacement64

Number of Active Set Update Completes

Event Count

Event 1a - Cell Addition326

Event 1b - Cell Removal305

Event 1c - Cell Replacement62

Soft-Handover Success Rate

Event Rate

Event 1a - Cell Addition99.4

Event 1b - Cell Removal99.7

Event 1c - Cell Replacement96.9

Table 4: Soft handoff success rate.

Drop Calls

All drop calls which are due to RF issues must be analysed and the appropriate steps taken to

avoid such drops from reoccurring.



RF related issues that may result in drop calls may include:

Poor coverage (RSCP & Ec/Io)

High interference and hence poor Ec/Io

Poor uplink coverage (insufficient UE Tx power)

Poor dominance (best cell changes too frequently resulting in too many SHO events)

Pilot pollution (too many cells present)

Missing neighbours

Fast change of RF conditions (e.g. turning a corner)

If none of the above are the causes of the call drop and the RF conditions are evidently good

at the location of the drop call, then the failure should be reported as a system fault for further

analysis. (Such analysis would also require collecting network traces).

Note that drop calls that repeatedly occur in same locations must be analysed in detail in

order to determine the exact causes.

Drop Call Analysis

There are a number of approaches for drop call analysis and the steps below are designed to

assist in quickly identifying RF related failures:

1. If RSCP & Ec/Io degrades before drop for BOTH scanner and UE then check for

coverage problems

2. If prior to the drop, the Ec/Io (and RSCP) degrades for UE ONLY while scanner

shows no degradation, then the following checks should be made:

a. Is the best server for the UE is the same as that of the scanner? (If not, its

possible that the UE failed to perform soft handoff)

b. Does UE camp on new cell immediately after drop?

c. If the UE camps on a new cell after the drop, was that cell neighboured to the

previous cell? (if not, consider adding this neighbour)

d. Was the UE measuring this neighbour?

e. Were there too many and too quick changes of best server making it difficult

for UE to perform measurements and SHO in time. (if this is the case: improve

cell dominance through antenna optimisation)

3. Does the UE Tx power increase to max prior to dropping call while Ec/Io level

remains good?

a. If the Tx power increase is gradual and UE is far from site the failure is due to

uplink coverage limitation

b. If the increase is sudden and UE may be not be too far from site - Check

uplink load from SIB7 following drop call – is it unusually high?



i. If uplink load is reported to be high, confirm from network stats that

the high load is due to genuine traffic – otherwise check for a possible

site fault

ii. If uplink load is not high, problem could be due to possible power

control failure.

If the above steps do not reveal the causes of the drop calls then analysis of the messages

should be carried out to determine the sequence of events prior to the drop call.

If the drop call does not appear to be RE related and the RF conditions at the location of the

drop appear to be goods then no further work is needed as part of the RF Optimisation.

However, exact location of the drop should be marked for later comparisons with future drive

surveys (if un-explained drops keep occurring at the same location, more detailed

investigation will be required to establish the exact causes).

4 Analysis Summary

In the previous section analysis of various Actix plots were discussed as well as steps for

identifying RF related drop calls.

In this section a summary of the analysis is provided as well as possible causes of two drop

calls labelled (1) and (2) in Figure 7 below.



Figure 7: Drop calls and scrambling code plot.

Drop Call 1:

Figure 7 shows that drop call 1 occurred at an area of frequent change of best server as shown

by the scanner scrambling code plot and was highlighted in Error! Reference source not

found. above.

It is interesting to compare Ec/Io from both scanner and UE at the time of the drop as shown

in Error! Reference source not found.. This clearly shows the UE Ec/Io to drop to < -21

dB while the scanner remained above -11 dB.

Comparing the best servers from the UE and the scanner at the time of drop, Figure 9: Drop

Call 1 (UE vs. scanner best server).Figure 9, shows that for the scanner and UE SC008 is the

best server prior to the drop. However, about 30 seconds before the drop, the scanner selected

SC018 as the best server while the UE continued to have only SC009 in its active set

resulting in the drop call. Immediately after the drop, the UE camps on SC018.

Drop 1

Drop 2



Examining the UE Active and Monitored set, Figure 10, does not show SC018 to be

measured by the UE prior to the drop. This scenario resembles a missing neighbour problem,

although in this case the two cells in question are neighboured.

It seems that the best server’s changes from SC009 to SC011 and then to SC018 were too fast

for this UE to perform soft handoff on time. Although, other UEs may have succeeded in

performing soft handoff in such conditions, it is important to improve the cell dominance in

the affected area.

Looking at Figure 7 clearly shows that at the location of the drop, SC018 should not be the

best server. Cell SC018 clearly requires some down tilting to control its interference into the

area of Drop 1. To illustrate this, RSCP coverage of SC018 (Figure 11) shows clearly that

the cell’s is extending into a large area. E.g. around the location of drop call, SC018 RSCP

is > -75dBm.

Drop Call 1 Summary:

The drop call appears to be associated with unnecessary change of best servers caused by

excessive dominance of SC018. To improve the dominance in the affected area, SC018

should be considered for downtilting.



Drop Call 2:

This drop occurred close to SC020 where RSCP and Ec/Io are better than -65 dBm and -8 dB,

respectively (based on scanner measurements). Unlike Drop 1, in this case:

- there were no changes of best server following the drop call

- Ec/Io & RSCP did not degrade prior to drop

- Both UE and scanner were camped on same cell (SC020)

It can be seen from Error! Reference source not found. that the UE was running at high Tx

power (> 20 dBm) at the time of drop. In fact, Figure 12 clearly shows that just before the

drop:

- Ec/Io was as high as -5 dB

- Target DL SIR starts increases up to 15 dB

- UE Tx power suddenly increases to the maximum (24 dBm)

- DL BLER increases to 100 %

It appears that the UE ran out of Tx power which is unlikely to be a coverage problem since

the UE is very close to the site and the degradation was too sudden.

In addition a check using Actix of the uplink load from SIB 7 following the drop did not

reveal any abnormal noise rise.

Drop Call 2 Summary:

The drop occurred in an area of good coverage and was preceded by a sudden degradation of

quality and an increase of UE Tx power. The cause could be related to issues with UE power

control performance.



Figure 9: Drop Call 1 (UE vs. scanner best server).

Figure 8: Drop Call Number 1 (with Ec/Io from scanner & UE at time of drop).

UE Ec/Io

degrades

prior to drop

Scanner best

servers NOT in

UE active set

UE camps

on SC018

after drop

UE Ec/Io

degrades before

the drop



Figure 10: Drop Call Number 1 (Active & monitored sets at time of drop).

Figure 11: RSCP coverage from SC018.

SC018 was not

measured by UE

prior to drop

SC018

RSCP > -75

dBm



Figure 12: Drop Call Number 2 (DL SIR, Ec/Io, UE Tx power & DL BLER at time of drop)

5 Problems & Solutions (TBA)

6 Summary

This document outlined analysis approach for RF drive survey data as well as analysis of

some drop calls. Throughout the report, example Actix plots were used for illustration

purposes only as the data was collected from an incomplete cluster of sites.

This document will be kept updated with additional information and optimisation cases

studies.

7 Appendix A: Default Colour Schemes in Actix (TBA)

UE Tx power

reaches max

Quality degrades

to100 % BLER



8 Appendix B: Actix Preferences (TBA)

9 Appendix C: Actix Thresholds (TBA)

10 Appendix D: Example Actix CellRef file (TBA)



Lists of References：

WCDMA RNO RF Optimisation Guidance-20040719-A-1.0.pdf

Documents

Transcript of WCDMA RNO RF Optimisation Guidance-20040719-A-1.0.pdf