Network Optimization & Tuning:

Feature impacts analysis

FRS 89934 & FRS 89933
Improved Uplink Interference Cancellation (I-UIC)
&
Improved Uplink Receiver Sensitivity (IUSR)
BR 9.0 Basic Packet

Version 1.0
Document status: IUS

Author: Krystian Majchrowicz

Date:

SN SDC NA B1

Tel.: +48 71 777-74441

E-mail: krystian.majchrowicz@siemens.com

21.03.2007

Issued by the
SIEMENS AG 2006
The reproduction, transmission or use of this document or its contents is not permitted without express written
authority. Offenders will be liable for damages. All rights, including rights created by patent grant or registration
of a utility model or design, are reserved. Technical modifications possible. Technical specifications and
features are binding only in so far as they are specifically and expressly agreed upon in a written contract.

This page is intentionally left blank

Table of contents
TABLE OF CONTENTS ............................................................................................. 3
1

HISTORY ..................................................................................................... 6

LITERATURE REFERENCES ..................................................................... 7

INTRODUCTION .......................................................................................... 8

3.1

Prerequisites for applicability ........................................................................................................... 9

3.2

When the IUIC and when IURS is applied to an on-going call...................................................... 9

NETWORK INTRODUCTION AND PARAMETERIZATION ASPECTS .... 10

CONTEXT OF APPLICATION AND EXPECTED BENEFIT ...................... 11

5.1

Application context .......................................................................................................................... 11

5.2

Expected influence of IUIC on the network performance............................................................ 11

5.3

Expected influence of IURS on the network performance ........................................................... 12

5.4
Total impact of both BR 9.0 LY 1 improvements (IUIC and IUSR) on the network
performance ........................................................................................................................................................ 13
5.5

Possible side effect ............................................................................................................................ 13

5.6

Cell context proposals for analysis of IURS and IUIC effects ..................................................... 13

RESULTS OF SIMULATIONS OR COLLECTED FROM FIELD ............... 15

7
VERIFICATION OF THE NETWORK IMPROVEMENT AFTER FEATURE
INTRODUCTION ...................................................................................................... 16
8

EVALUATION OF THE FEATURE IMPACT WITH SAG TOOLS ............. 17

8.1

Feature impact on AnatomN........................................................................................................... 17

8.2

Feature impact on Radio Network Analyser (RNA)..................................................................... 18

8.3

Feature impact on Siemens PM Counters...................................................................................... 18

ABBREVIATIONS
AFS

AMR Fullrate Speech

AHS

AMR Halfrate Speech

AMR

Adaptive Multi Rate

BER

Bit Error Rate

BTS

Base Transceiver Station

CDR

Call Drop Rate

CGI

Cell Global Identifier

CM

Configuration Management

DL

Down Link

DTM

Dual Transfer Mode

ECU

EDGE Carrier Unit

EDGE

Enhanced Data rates for GSM Evolution

EFL

Effective Fractional Load

EFR

Enhanced Full Rate

FACCH

Fast Associated Control Channel

FCU

Flexible Carrier Unit

FR

Full Rate

HO

Handover

HO ACCESS

Handover Access

HO COMPLETE

Handover Complete

HSCSD

High Speed Circuit Switch Data

HR

Half Rate

IUIC

Improved Uplink Interference Cancellation

IURS

Improved Uplink Receiver Sensitivity

LMS

Least Mean Squares

LY

Layer

MOC

Mobile Originating Call

MTC

Mobile Terminating Call

MS

Mobile Station

NOT

Network Optimization and Tuning

RACH

Random Access Channel

RLF

Radio Link Failure

SABM

Set Asynchronous Balance Mode

SACCH

Slow Associated Control Channel

SAIC

Single Antenna Interference Cancellation

SDCCH

Standalone Dedicated Control Channel

SMS

Short Message Service

TA

Timing Advance

TCH

Traffic Channel

UIC

Uplink Interference Cancellation

UL

Up Link

History

Version

Date

Chapter

Changes / Reasons

0.1

22.12.2006

DRAFT_AFI

1.0

21.03.2007

Set to IUS after internal review

Literature references
1. [AEK89933] FRS Improved Uplink Receiver Sensitivity by averaging over Multiple
bursts,
Dr. Konrad Joos, 14.02.2005,
2. [AEK89934] FRS Improved UIC, Dr. Konrad Joos, 14.02.2004,
3. Network Optimization & Tuning: Feature impacts analysis Uplink Interference
Cancellation Sebastian Lysiak 09.01.2006
4. PLR 89933 "Improved Uplink Receiver Sensitivity Andrzej Macioek, 17.03.2006
5. BR 9.0 PM: SBS counters

Introduction

Traditionally the BTS receiver has been optimized for signal reception in noise limited
scenarios. Especially with the introduction of tight reuse and increasing fractional
load in some customer networks the situation changes towards an interference
limited scenario. To react on this, single antenna interference cancellation (SAIC)
techniques are currently being introduced into mobiles. This improves the downlink
performance in such situations with high fractional load. Note that SAIC gain in DL
depends on the penetration rate of SAIC capable MS. To achieve a symmetric
improvement, the Siemens BTS release BR 8.0 contains yet a first version of Uplink
Interference Cancellation (UIC). This technique has been implemented only on the
ECU and FCU. In BR9.0 further UIC improvements were implemented. In
interference limited scenarios the improvements are enhanced by 1 dB in comparison
to UIC (BR 8.0).
In case of noise limited scenario BR 9.0 basic packet offers enhanced gain in
receiver sensitivity what improves performance in uplink. This feature is called
Improved Uplink Receiver Sensitivity by averaging over Multiple bursts. As both
IUIC and IURS are the basic features and are activated as soon as the software
upgrade is performed (no dedicated CM parameters) there is no possibility to
evaluate feature benefit separately for IUIC and IURS. For this reason this NO&T
document treats these features jointly.
IURS employs sophisticated mathematical algorithm aimed to improve sensitivity in
UL. This method performs best in stationary and low mobility environments where
good averaging is possible and in environments with long power delay profile. It
works only on circuit switched timeslots where the same user is transmitting with an
interleaving of a certain number of bursts per frame. The feature is applied to voice
and circuit switch data (HSCSD). It is not applicable to packet data services (GPRS,
E-GPRS) since different user can be multiplexed on the same TS, where the BTS
has no knowledge about which user sends a specific burst. In this case an averaging
over following bursts makes no sense as these bursts need not come from the same
user. For a user in DTM mode only the CS channel is improved.
IUIC is the enhancement of the BR 8.0 solution (UIC) which aims to suppress
interference in uplink. The UIC algorithm is implemented as a burst data preprocessing step (pre-filter) applied before channel equalization. The algorithm is
trying to find the optimum parameters for pre-processing by maximizing the SNR in
the receiver during the midamble period. An error term is calculated from the known
midamble symbols and the received symbol sequence. This error term is fed into an
LMS algorithm, as the optimum parameters are found when the error term is
minimized.
Both features: IURS and IUIC are included in basic packet of BR 9.0 what means
that once the upgrade to BR 9.0 is performed they are enabled and the operator
does not have to pay for them extra (the price for the feature is included in the price
for the whole basic packet).

3.1

Prerequisites for applicability

The IURS feature is valid for the uplink and depends on the scenario the mobile is
moving about. The best performance is expected for cases of a discrete power delay
profile compared to a continuous one. The best candidates are the rural and hilly
terrain scenarios. For urban scenarios the effect is expected to be less as these are
mostly interference limited. This means that the gain will be highest for operators with
large radio cells.
The UIC feature can brings the benefit only in typical urban environments with
interference scenarios of several co- and adjacent channel interferers and TU3 or
TU50 channel models. It is not applicable in case of sensitivity limited scenarios.
Both features were implemented for BTS plus family only.

3.2

When the IUIC and when IURS is applied to an on-going call

In BR 9.0 it is possible that only one of these two features works at the same time. In
noise limited scenario IUSR is applied while in interference limited scenario IUS is
activated. In BTS there is special algorithm which evaluates interference and decides
whether it is sensitivity or interference limited scenario.

Network introduction and parameterization aspects

Being both IUIC and IURS part of the basic package of BR 9.0 features, they do not
need to be enabled and managed by any O&M. As soon as the BSS software is
upgraded to BR 9.0 these features are activated and can not be disabled. As there is
no single parameter related to this feature it can not be parameterized either.

Context of application and expected benefit

5.1

Application context

IUSR and IUIC are designed to improve performance of uplink but each of them is
dedicated to different scenario. In noise limited cluster IUSR will be in use, while in
interference limited networks IUIC will be applied.
Because both feature improves only uplink, the balance (or imbalance) between upand downlink in the analyzed network has a significant meaning for expected gain. It
is recommended to analyze the data concerning AMR channel type usage to get
information about uplink and downlink performance and to determine which link is the
weaker one. For this purposes Online RF Loopback measurements can be used as
well. Both features can bring significant benefit only in case of uplink limited
networks.

5.2

Expected influence of IUIC on the network performance

This subchapter shows the field of potential improvements due to feature activation.
However it is important to remember that these improvements are strictly dependant
on network parameterization and can become unnoticeable due to inappropriate
tuning of CM parameters (for instance a wrong power control settings will not allow to
see impact on RxLev).
Impact on RxQual
The IUIC algorithms suppress uplink interference and therefore it improves C/I what
is mirrored in UL RxQual. This has the impact on AMR codec distribution (less robust
codecs will be used more often) and UL FER as well. UL RxQual is one of the criteria
for handover and therefore with IUIC feature the number of handovers (intracell and
intercell) triggered due to bad uplink quality will decrease.
Impact on RxLev
Better UL RxQual will have the influence on power control mechanism in this way
that it will decrease the power level necessary to maintain to maintain the acceptable
quality. This will cause battery save and less interference in the network.
Impact on call drops
IUIC improves C/I conditions by 1 dB what impacts not only quality of the traffic
channels but signalling as well. Signalling performance has the major impact on call
drops. With better C/I in uplink (on SACCH as well) the probability that S counter
reach 0 (radio link failure will be generated) will be decreased. It decreases also
drops due to handover access failure (HOAF) and T8 and T10 expiry as these are
impacted also by performance of the messages sent in uplink during HO procedure
like SABM and HO COMPLETE (HO ACCESS is not impacted as these is send on
access burst and IUIC improves only performance on normal burst).

Impact on throughput
Generally the gain from IUIC implementation allows using the weaker channel coding
schemes more often (because the BER value decreases), so the throughput in the
uplink can increase. Since the uplink data rate is not a strong limiting factor of packet
data transmission, the influence of the feature is not so significant, however there are
some applications requiring high throughput in uplink in such a case the QoS will
be improved (because of throughput increase).

5.3

Expected influence of IURS on the network performance

IURS feature extends cell range due to the better BTS receiver sensitivity. The
experienced gain is up to 0.8 dB on fading channels and approx. 0.5 dB in average in
the UL link budget.
Impact on call drops
In noise limited scenarios the majority of drops are caused by the Radio Link Failure.
If uplink is the limiting factor (S counter in BTS expiries) the feature IURS will reduce
call drops. As call which was dropped prematurely now will last longer (as IURS can
prevent against the drop or delays it), the mean holding time will increase.
Impact on RxLev and RxQual
As the better BTS sensitivity prevents some calls against the drop in noise limited
scenario, now the call can last longer (is not dropped that early due to RLF in spite of
low UL RxLev and RxQual). For these reason the distribution of RxLev in UL should
change in this way that with IURS there will be more samples with low RxLev. These
samples will come from the calls which were saved against the drops (thanks to
IURS). Increased number of samples with low UL RxLev will imply increased number
of samples with bad quality (UL RxQual >4).
Impact on TA
As IURS improves BTS receiver sensitivity the number of calls served on the cell
edge can be increased and therefore the distribution of the timing advance in uplink
will be changed in this way that it will be moved into higher TA ranges.
Impact on call setup
With the better sensitivity the ratio between Attempted Immediate Assignment
Procedure and paging messages will increase. Before the feature activation in noise
limited scenario - MS sent in answer to paging CHANNEL REQUEST, which could
not be received by the BTS. With improved uplink sensitivity now it can be received
what should be illustrated in number of attempted immediate assignment in answer to
paging.

5.4
Total impact of both BR 9.0 LY 1 improvements (IUIC and IUSR) on
the network performance
As activation of the both features is scenario dependant the total impact of IUIC and
IUSR will be the combination of impacts described in the two previous chapters. In
case of contrary impacts the one coming from this feature which is used most often
(determined by the scenario) will be the dominant one.
Here below the total impact of layer 1 improvements in BR 9.0 basic packet is given.
Impact on call drops
The most significant call drop reduction will be for drops due to RLF, however these
drops due to HOAF, expiry of T8 or T10 will decrease as well.
Impact on RxQual
On one hand IUIC improves quality in uplink, on the other IURS allows lower RxLev
in UL which is most of the times correlated with low quality. With respect to RxQual
there might have contradictory impacts. Which one will be the dominant one depends
whether, in the cluster under analysis, noise or interference is the most limiting
factor.
Impact on RxLev
UL RxLev will decrease as the effect of IURS. Also IUIC will decrease UL RxLev but
this effect will be indirect. With UIC RxQual values will decrease (which means better
quality) and due to this power control will reduce UL RxLev.
Impact on TA
The distribution of the uplink timing advance will be changed in this way that it will be
moved into higher TA ranges as the effect of IURS.
Impact on call setup
The ratio between Attempted Immediate Assignment Procedure and paging
messages will increase Attempted Immediate Assignment Procedure (CHANNEL
REQUIRED messages) will increase as the effect of IURS.
Impact on throughput
In uplink throughput will increase as the gain from IUIC implementation allows using
the weaker coding schemes more often (because the BER value decreases). This
gain will only be noticeable in interference limited scenario. Also as the uplink data
rate is not a strong limiting factor of packet data transmission, the influence of the
feature is not considerable.
5.5

Possible side effect

There is no side effect brought by activation of these features.

5.6

Cell context proposals for analysis of IURS and IUIC effects

Both IURS and IUIC improve uplink and therefore features benefit can only be
noticeable in these cells where uplink is the limitation. As it was mentioned before
both features are enabled as soon as BSS software is upgraded to BR 9.0 and can

not be deactivated. Depending on the scenario IURS or IUIC will be in use and
therefore the expected benefit is determined by the cell cluster (coverage or
interference limited).

Results of simulations or collected from field

According to link level simulation results the IUSR gives improvement of up to 1 dB

on fading channels (and 0,8 dB on average) for the sensitivity. It was ensured that
the feature does not cause any degradation, especially in case of high velocity.
The same simulations for IUIC showed that the receiver performance in C/I limited
channels improved by about 1 dB in interference limited scenarios.

7
Verification
introduction

of

the

network

improvement

after

feature

In order to estimate benefit provided by LY 1 improvements in BR 9.0 basic packet it

is suggested to observe counters and KPI related to call drops (and call drops in
relation to traffic load) before and after software upgrade. It is expected that these
counters should generally indicate a decrease of the call drop figures.
Applying the BR 9.0 LY 1 improvements shall result in the following changes
(towards network with BR 8.0) of the certain KPIs (for definitions of specific KPIs
refer to chapter 8):

Drops due to RLF per Erlh will decrease;

Drops due to HOAF per Erlh will decrease;

Intercell HO Drops (T8 expiries) per Erlh will decrease;

Intracell HO Drops (T10 expiries) per Erlh will decrease;

Call drop rate, TCH drop rate, TCH Drop Erlh will decrease;

mean UL RxLev will get lower;

number of samples within TA7 range will increase;

the ratio of attempted immediate assignment procedure to the paging

messages will increase;

Evaluation of the feature impact with SAG tools

8.1

Feature impact on AnatomN

The idea of this document is to present how to measure expected benefit coming
from BR 9.0 LY 1 improvements. In order to do this the results of BR 8.0 PM
counters (before upgrade) have to be collected and then compared to BR 9.0. This
can be done in AnatomN with the usage of attached predefined reports which can be
simply saved and imported in AnatomN.

Impact of LY1
improvements on netw

Figure 1 AnatomN predefined report presenting KPIs which should be observed in

order to verify benefit from LY1 improvements in BR 9.0.
This AnatomN report includes the most important KPIs which should be observed
before and after upgrade to BR 9.0. The report contains Traffic Carried as well since
the cluster should be analysed in busy hour. The example in the figure visualizes the
results.

8.2

Feature impact on Radio Network Analyser (RNA)

In Radio Network Analyzer there exists History on Dropped Call functionality which
allows to trace the number of calls which were dropped with cause of drop and
additional information as RxQual, RxLev, TA and CGI of the target cell (if drop
occurred during handover). Therefore this tool can be used as the crosscheck option
for the analysis performed in post processing offline PM tool (like AnatomN).
8.3

Feature impact on Siemens PM Counters

The implementation of IUIC and IURS will have impact on the Performance
Measurement counters in Siemens SBS. Here below are listed all counters which will
be influenced by BR 9.0 LY1 improvements (which stay behind the formulas of the
KPIs included in AnatomN report enclosed in chapter 8.1).
Impact on call drop statistics
NRFLTCH [7,16] Number of lost TCH connection - this measurement determines

the total number of lost connections (while using TCH/F or TCH/H) due to Radio Link
Failure (S - counter reaches value 0). Based on this counter, KPI which gives the
relation between drops due to RLF and carried traffic can be calculated.
Drops_RLF_Erlh = NRFLTCH[7,16] / MEBUSTCH[1..4]
where MEBUSTCH = Mean Number of Busy TCHs counters;
After upgrade to BR 9.0 it is expected that the value of this KPI will significantly
decrease.
NRFLTCH [6,15] Number of lost TCH connection - this measurement determines the

total number of lost connections (while using TCH/F or TCH/H) due to handover
access failure (T3105 expires NY+1 times and MS does not succeed to return to the
old cell with HO FAIL).
Based on this counter, KPI which gives the relation between HOAF drops and carried
traffic can be calculated.
Drops_HOAF_Erlh = NRFLTCH[6, 15] / MEBUSTCH[1..4]
With LY 1 improvements in BR 9.0 it is expected that the value of this KPI will
decrease.
UNIHIALC - Unsuccessful Internal Handovers Intracell with Loss of MS - this

measurement determines the number of intracell handover failures where connection

with the MS was lost and T8 expired.
From this counter the rate between intercell HO drops and the traffic can be
estimated.

Drops_T8_Erlh = UNIHIALC / MEBUSTCH[1..4]

After IUIC and IURS activation it is expected that the value of this KPI will decrease.
UNIHIRLC - Unsuccessful Internal Handovers Intercell with Loss of MS - this

measurement determines the number of intercell handover failures where connection

with the MS was lost and BSC timer T10 expired.
From this counter the rate between intercell HO drops and the traffic can be
estimated.
Drops_T10_Erlh = UNIHIRLC / MEBUSTCH[1..4]
After IUIC and IURS activation it is expected that the value of this KPI will decrease.
Impact on RxLev distribution
CRXLVQUU [1..64] - Correlated RXLEV to RXQUAL measurements in uplink this
measurement provides the RX level correlated to RX quality for uplink direction.
Each counter is related to the RXLEV and RXQUAL value of the measurement report
retrieved from the internal BTS uplink measurements (RXLEV, RXQUAL). Based on
this counter MeanULRxLev can be calculated according to the formula:
MeanULRxLev =

0 * CRXLVQUU[1..8] + 1* CRXLVQUU[9..17] + 2 * CRXLVQUU[18..26] + .. + 7 * CRXLVQUU[ 57..64]

CRXLVQUU[1..64]

After IUIC and IURS activation it is expected that the value of this KPI will decrease.
PWRUPDW [1..64] Power and Quality Measurements on Uplink Busy TCHs - this
measurement provides the number of times a certain received field strength (RXLEV
value) was reported for a specified TCH in UL.
From this counter mean UL signal level can be computed:
MeanULLev =

0 * PWRUPDW[1] + 1 * PWRUPDW[2] + 2 * PWRUPDW[ 3] + .. + 63 * PWRUPDW[64 ]

PWRUPDW[1] + .. + PWRUPDW[64 ]

As the meaning of this KPI is the same as the MeanULRxLev, thus its value should
be reduced after BR 9.0 upgrade.
The difference between this KPI and the one above is that MeanUL Lev is based on
PWRUPDW which gives RxLev results in 64 bands (from 0 to 63), while
MeanULRxLev gives RxLev in 8 reduced bands and therefore it is less accurate.
Remark: Most of the time measurements on the channel object (e.g.PWRUPDW) are
not activated as the maximum number of simultaneously observed channels per BSC
is restricted to 80. Then this KPI can not be validated.
Impact on RxQual distribution
From the measurement CRXLVQUU presented above MeanULRxQual can be calculated
according to the formula:

MeanULRxQaul =

0 * CRXLVQU[1 ,9,17,25,33,41,49,57] + 1 * CRXLVQUU[2,10,18,26,34,42,50,58] + .. + 7 * CRXLVQUU[ 8,16,24,32,40,48,56,64]

CRXLVQUU[1] + .. + CRXLVQUU[64]

Both IUIC and IURS influence the quality, however this impact is opposite and
therefore the total impact of LY1 improvements depends on the network scenario. In
interference limited scenario IUIC is applied and therefore quality will be improved,
while in coverage limited cluster IURS will be enabled what will cause that more
samples with low level (and probably low quality) will be received.
Impact on TA distribution
CRXLVTAU [1..64] - Correlated RXLEV to Time Advance Measurements in uplink this measurement provides the RX Level correlated to time advance for uplink
direction. Each counter is related to the RXLEV and Timing Advance (TA) value of
the measurement report retrieved from the internal BTS uplink measurements
(RXLEV, TA). From this counter the rate of received samples having timing advance
in TA Band equal to 7 can be calculated:
TA7 Dist =

CRXLVTAU[8,16,32,24,32,40,56,64]
CRXLVTAU [1..64]

With the active IURS it is expected that the number of measurement samples within
TA7 band will increase as IURS allows the call to be served on the cell edge what
can lead to cell dragging (when HO can not be performed). It is very important to
remember that indeed there are 64 TA ranges, however CRXLVTAU considers only 8
so called reduced TA ranges. The mapping of 64 TA ranges into 8 reduced bands
strictly depends on settings of the parameters TIMADVR in the object SCANCTRX.
Impact on immediate assignment procedure
ATIMASCA [1..6] - Attempted Immediate Assignment Procedure - this measurement
gives the number of times an immediate assignment was requested for SDCCH
channel. Different counters distinguish between the assignment causes (MOC, MTC,
location update, emergency call. call reestablishment, mobile terminating SMS,
mobile originating SMS). With IURS it is expected that the number of Attempted
Immediate Assignment Procedure (CHANNEL REQUIRED messages) will increase.
Before feature activation in the noise limited scenario - MS accesses RACH and
sends CHANNEL REQUEST. It could happen that this message is not reached as
the received level is below BTS receiver sensitivity. With IURS the probability that the
message send on RACH is lost decrease and therefore the number of immediate
assignment attempts should increase. This is the true under condition that the
number CHANNEL REQUEST messages before and after feature activation is more
or less the same.
Based on this counter KPI which gives the number of times immediate assignment
procedure was requested on SDCCH channel due to any reason can be calculated
according to the formula:
ImmAss = ATIMASCA [1..6]

Since IURS should increase the number of attempted immediate assignments it is

expected that the value of this KPI will increase as well.