AMR

Adaptive Multi Rate Training

Bangkok July 14-15/1

1 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Course Objectives
• After this course, participants are able to :

• Know how AMR works

• The purpose and planning aspects of the feature

• Understanding of AMR parameters

• Inter-working of AMR with other features

• HW/SW requirements for AMR

• Impact of AMR on network performance

2 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Table of Contents

• AMR Introduction
• AMR Benefits
• AMR Codecs
• Nokia AMR Link Adaptation and codec mode adaptation
• Nokia AMR interaction with other Nokia features
• AMR support in Nokia system
• Nokia AMR parameter
• Nokia AMR KPI
• Nokia AMR planning aspects
• AMR implementation

3 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Introduction :
• Name :
• Company :
• How long :
• Position :
• Past experience :
• Expectation from AMR
training:
•List of questions

4 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Introduction

5 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Hard/Soft Blocking

Hard
Hardblocking
blocking
The
Thewhole
wholeradio
radioresource
resourceisisin
inuse
use--no
nomore
morecalls
callscan
canbe
beestablished
established
due to lack of free radio timeslots.
due to lack of free radio timeslots.

Dominates with large reuse factors = Wideband deployment

Soft
Softblocking
blocking
The
Thecapacity
capacityof
ofindividual
individualcells
cellsis
islimited
limitedby
bythe
thelevel
levelof
ofthe
theinterference
interference
rather than the number of TRXs available
rather than the number of TRXs available

Is dominating with tight reuse patterns = Narrowband

deployments

6 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Spectral Efficiency & Performance
• Standard Measure: Erl/km²/MHz
• Nokia Measure: Effective Frequency Load ( Erl/MHz)
• Spectral Efficiency is equivalent to performance
• Assuming no lack of radio resources or HW blocking
Performance is Key Performance Two alternative solutions
a trade-off Indicator – CDR, BQS
between
capacity AND Operating
quality Point Targeted quality level
Quality Enhancement
is measured in terms of
increased quality Dropped calls due
for the same load to coverage gaps
Increased
Increasedperformance
performance Traffic
(spectral
(spectralefficiency)
efficiency) Load
delivers improved
delivers improved Capacity Increase is measured
quality
qualityand/or
and/orhigher
higher in terms of additional load
capacity
capacity for thesame
for the same at the same quality level
quality criteria
quality criteria

7 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Effective Frequency Load Defined
• EFL is a measure of the average frequency utilization in the area
 Represents how loaded each frequency can be across the
system
• EFL is proportional to spectral efficiency
• EFL is directly proportional to the carried traffic  x % higher EFL
= x % more carried traffic
Busy hour area
level average
Erlangs/cell

ErlBH 1
EFL  
Tot # freq Ave # ( TCH )
TRX
Total number of Average
frequencies number of
used to carry timeslots/TRX
the traffic

8 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Effective Frequency Load Explained
• EFL is a measure of the average frequency utilization in the area  Represents how
loaded each frequency can be across the system
• Assume 1.2 Mhz (6 x 200 kHz carriers) of hopping frequencies in addition to the
BCCH carrier
• Assume in each cell 5 simultaneous voice users on the average
• In this case the Effective frequency load is ~ 5 Erlangs / 48 timeslots = 10.4%

r
e
m
r r ie Ti
ca
er
p
ots
e sl
t im
5

Frequency
200 kHz 200 kHz 200 kHz 200 kHz 200 kHz 200 kHz

6 frequencies @ 200 kHz each

9 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Adaptive Multi-Rate Codec (1/2)
• Adaptive Multi-Rate (AMR) codec consists of a family of codecs
(source and channel codecs with different trade-off bit-rates)
operating in the GSM FR and HR channels modes
• The AMR system exploits the channel performance and robustness
added by the coding rates by adapting the speech and channel
coding rates according to the quality of the radio channel
• AMR adapts its error protection level (select its optimum
channel mode and codec mode) to the local radio channel and
traffic load conditions to deliver the best possible combination of
speech quality and system capacity
• Codec mode adaptation for AMR is based on received channel
quality estimation in both MS and BTS, followed by a decision on
the most appropriate speech and channel codec mode to apply at a
given time
• The basic AMR codec mode sets for MS and BTS are provided by
BSC via layer 3 signaling
• MS shall support all speech codec modes, although only a set of up
to 4 speech codec modes is used during a call
10 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Adaptive Multi Rate Codec (2/2)
• GSM FR/EFR channel gross bit-rate is 22.8 kbit/s in GSM FR/EFR:
13 kbit/s speech coding and 9.8 kbit/channel coding (HR channel
gross bit rate 11.4 kbit/s)
• For AMR case, different codecs use different bit rate to encode
speech (source coding). The rest of the gross bit-rate is used for
channel protection
25
Channel coding
Channel bit-rate (kbit/s)

20 Speech coding
Robustness
15

10

5
Speech Qual
0
FR FR FR FR 7.4 FR 6.7 FR 5.9 FR FR HR HR 7.4 HR 6.7 HR 5.9 HR HR
12.2 10.2 7.95 5.15 4.75 7.95 5.15 4.75

AMR codec mode

11 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Algorithms Related to AMR
• In order to select the codec, MS and infrastructure
vendors implement the Link Adaptation algorithm or
Codec Mode Adaptation

• Additionally, there is another algorithm to change the

channel rate between FR and HR codecs, which is
called Channel Mode Adaptation

Set of Codecs Codec Mode Adapt.

Channel Mode Adaptation AMR

12 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Benefits

13 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Capacity and Coverage Gain
• Link level results show very high improvement in the terms of TCH
FER when robust AMR modes are used
• As high as 6 dB improvement at 1% FER in C/I can be achieved 
Therefore, high capacity gain can be expected when robust AMR
modes are utilized
• In addition, increased robustness to channel errors can be utilized
in the cell coverage, i.e. lower C/I can be allowed at the cell edge
• However, in the mixed traffic case the cell coverage has to be
planned according to EFR mobiles
• With respect to signaling channels, the retransmissions schemes
used by SACCH and FACCH channels maintain the probability of
signalling success even for very degraded conditions

14 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Capacity Increase with AMR
• Due to robust AMR codec modes, very low TCH FER compared to EFR
• In 850 MHz case all mobiles are AMR capable, but this comparison
illustrates
the capacity gain AMR provides when it is introduced in a typical network
10
AMR MS penetration: 0%
9 AMR MS penetration: 25%
TCH FER > 5.4 % (%)
Relative Frequency

8 AMR MS penetration: 50%

7 AMR MS penetration: 75%
AMR MS penetration: 100%
6 ~150%
gain
5 relative
to EFR
4
3
2
Capacity gain 1
based on the
2% outage of 0
the bad TCH 5 7.5 10 12.5 15
FER samples
Effective Frequency load (%)
ONE-LAYER
ONE-LAYER(RF-hopping
(RF-hopping2/2,
2/2,no
noBCCH
BCCH
included)
included)
15 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Improved BCCH Plan
• Since the average C/I found in a cell area can be measurably less
than that used in a non-AMR network and still provide comparable
quality to EFR, the existing clean BCCH layer can be tightened,
potentially releasing frequencies to be used on the non-BCCH
layer
• This offers improved speech quality and extra capacity for TCH,
especially in the narrow band deployment (frequency band less
than 5 MHz)
• However, if EFR roaming mobiles are to be taken care of, the
BCCH will have to be planned accordingly
• How to plan networks to ensure the quality for the old EFR
mobiles?
• One method is to use more aggressive power adjustment for
AMR mobiles in order to decrease the average interference
level in the network
• Therefore, the overall interference decreases in the network
(smaller average transmission power) and thus the quality of
the existing EFR connections increase
16 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Half-Rate Utilization in AMR Codec
• Half-rate is an efficient way to increase capacity in the case of
limited number of TRXs per cell
• AMR HR codec obtains remarkable better speech quality than
previous GSM EFR HR codec
• AMR FR obtains better quality than AMR HR only when higher FR
modes than 7.4 are used (due to higher number of speech coding
bits)
• AMR FR 7.4 kbit/s mode and AMR HR 7.4 kbit/s mode have the
same speech quality when the C/I is high (error free case)
• AMR HR channels can be then used in high C/I conditions
without noticeably speech quality loss
• In theory for ideal frequency hopping about 11-12 dB C/I is
required for AMR HR to obtain the evaluated good speech quality
limit (in real networks, depending on the BTS configuration and on
FH mode used, it might be necessary 1-4 dB higher)
• Based on this, all connections having at least 12 dB C/I could be
handed over to HR channel remaining the good speech quality

17 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR FR+HR vs. AMR FR Usage
3.0%
4

AMRFR +HR 3.5 AMR FR

2.5% 22.98%

Bad TCH FER samples [%]

AMRFR AMR FR+HR
3
2.0%
2.5
TCH FER > 4%

??
HR Usage 36.23% 31.49%
1.5% 2
44.83%
54.48% 1.5

?
1.0%

1
0.5%
0.5

0.0%
0
17% 19% 21% 23% 25% 27% 29% 31% 0 5 10 15 20 25 30 35
Effective Frequency Load (%) Effective Frequency Load [%]

• The performance degradation in • The FER performance of AMR

FER between the usage of AMR FR+HR and AMR FR are about the
FR+HR and AMR FR is equivalent same
to the quality loss of 0.2 in the Capacity
MOS between AMR FR and HR Capacitygain
gainin
inFH
FH
carrier when AMR FR+HR
carrier when AMR FR+HR
codec is
isused
usedcontinue
continuetotobe
be
~150%
~150%

18 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Benefits For End User
MOS vs.
4.5
CIR
4
3.5
3 AMR FR
EFR
2.5
MOS

FR 12.2 MOS
2 FR 7.4 MOS
FR 5.9 MOS
1.5 FR 4.75 MOS
1 HR 7.4 MOS
HR 5.9 MOS
0.5 HR 4.75 MOS
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
C/I (dB)

• A user in good radio conditions perceives

the same quality as EFR.
•However, a user in bad radio conditions still
receives acceptable speech quality while
with EFR it would not received satisfactory
speech quality.

19 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Benefits For Operator
100%
fs475iFH

Coverage Gains
fs515iFH
fs590iFH
Capacity / fs670iFH
fs740iFH 10%
fs795iFH

TCH FER
fs102iFH
fs122iFH

1%

0%
10 8 6 4 2 0
C/I [dB]
• Approx. 5.5 dB link level gain in hopping layer
• This turns into approx. 140% capacity gain for
AMR-FR
• Coverage enhancement (>4dB)
• Tighter BCCH reuse schemes.
• Saving of resources by deploying AMR-HR

20 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Benefits of AMR
Summary
• Speech quality enhancement: AMR maintains good speech quality
in the situation where the connection faces low C/I or low signal
level
• Capacity and coverage gain: Link level simulation results
illustrated improvement in terms of TCH FER (up to 5.5dB at 1%
FER in C/I)
• Signalling channel performance: due to retransmissions schemes
used by these channels the probability of signalling success
maintain very high even for very degraded conditions
• Improved BCCH plan: tighter frequency reuse or better quality
with same frequency reuse, potentially releasing frequencies to
be used on the non-BCCH layer.
• HR utilisation increases the hardware capacity of the cell since
two half-rate connections can be allocated to fill only one timeslot.
• When compare AMR HR to previous GSM HR codec, it is noticed
that AMR HR obtains remarkable better speech quality

21 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Codecs

22 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Codecs
AMR Full Rate performance compared to
Full Rate Half Full Rate EFR in Clean Speech
rate
EFR MOS (Mean Opinion Score)
5.0
12.2
10.2
4.0
IS-136 7.95 7.95
6 dB gain in
7.4 7.4 3.0 performance
6.7 6.7
2.0 EFR
5.9 5.9 AMR FR
5.15 5.15 1.0
No Errors 16 dB C/I 13 dB C/I 10 dB C/I 7 dB C/I 4 dB C/I
4.75 4.75
Speech bit rate AMR Half Rate performance compared to
Full Rate in Clean Speech
Quality loss
MOS (Mean Opinion Score) of ~ 0.2
5.0
• New AMR family of codecs between
AMR HR
tolerates 6 dB higher interference 4.0 and FR
than current GSM EFR codec
3.0
• Can be directly utilized for higher
capacity with Frequency Hopping AMR HR
2.0
• Higher interference tolerance AMR FR
• Reduced time slot occupancy
1.0
No Errors 19 dB C/I16 dB C/I 13 dB C/I10 dB C/I 7 dB C/I 4 dB C/I

23 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel and Speech Codecs for AMR
Channel Channel Source coding Net bit-rate, Channel Channel
mode codec bit-rate, speech in-band coding coding
Mode channel bit-rate, bit-rate, in-
speech band
CH0-FS 12.20kbit/s (GSMEFR) 0.10 kbit/s 10.20 kbit/s 0.30 kbit/s
CH1-FS 10.20 kbit/s 0.10 kbit/s 12.20 kbit/s 0.30 kbit/s
CH2-FS 7.95 kbit/s 0.10 kbit/s 14.45 kbit/s 0.30 kbit/s
TCH/FR CH3-FS 7.40 kbit/s (IS-641) 0.10 kbit/s 15.00 kbit/s 0.30 kbit/s
CH4-FS 6.70 kbit/s 0.10 kbit/s 15.70 kbit/s 0.30 kbit/s
CH5-FS 5.90 kbit/s 0.10 kbit/s 16.50 kbit/s 0.30 kbit/s
CH6-FS 5.15 kbit/s 0.10 kbit/s 17.25 kbit/s 0.30 kbit/s
CH7-FS 4.75 kbit/s 0.10 kbit/s 17.65 kbit/s 0.30 kbit/s
CH8-HS 7.95 kbit/s (*) 0.10 kbit/s 3.25 kbit/s 0.10 kbit/s
TCH/HR CH9-HS 7.40 kbit/s (IS-641) 0.10 kbit/s 3.80 kbit/s 0.10 kbit/s
CH10-HS 6.70 kbit/s 0.10 kbit/s 4.50 kbit/s 0.10 kbit/s
CH11-HS 5.90 kbit/s 0.10 kbit/s 5.30 kbit/s 0.10 kbit/s
CH12-HS 5.15 kbit/s 0.10 kbit/s 6.05 kbit/s 0.10 kbit/s
CH13-HS 4.75 kbit/s 0.10 kbit/s 6.45 kbit/s 0.10 kbit/s
(*) Requires 16 kbit/s TRAU. Therefore it is not seen as a feasible codec mode and will not be
supported by Nokia BSS10.

In high-error conditions more bits are used for error correction to obtain
error robust coding, while in good transmission conditions a lower
amount of bits is needed for sufficient error protection and more bits can
therefore be allocated for source coding
24 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Benefits of AMR 1/2
• Scenario MOS

• Highest MOS 5.0

• Highest HR usage then

highest MOS but MOS 4.0

>= 3.8
• Highest quality 3.0

• What is quality? EFR

12.2
• Anything else? 10.2
7.95
2.0
7.4
6.7
5.9
5.15
4.75 Conditions
1.0
No Errors C/I=16 dB C/I=13 dB C/I=10 dB C/I= 7 dB C/I= 4 dB C/I= 1 dB

EFR 4.01 4.01 3.65 3.05 1.53

12.2 4.01 4.06 4.13 3.93 3.44 1.46

10.2 4.06 3.96 4.05 3.80 2.04

7.95 3.91 4.01 4.08 3.96 3.26 1.43

7.4 3.83 3.94 3.98 3.84 3.11 1.39

6.7 3.77 3.80 3.86 3.29 1.87

5.9 3.72 3.69 3.59 2.20

5.15 3.50 3.58 3.44 2.43

4.75 3.50 3.52 3.43 2.66

25 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Benefits of AMR 2/2
• HR utilisation
doubles the M OS Experiment 1b - Test Results
capacity of the 5.0

cell
4.0

• When compare
AMR HR to 3.0 EFR

previous GSM HR 7.95

7.4
6.7
codec, it is 2.0
5.9
5.15

noticed that AMR 4.75

FR
HR
HR obtains 1.0
Condi ti ons

remarkable
No Errors C/I=19 dB C/I=16 dB C/I=13 dB C/I=10 dB C/I= 7 dB C/I= 4 dB

EFR 4.21 4.21 3.74 3.34 1.58

better speech 7.95 4.11 4.04 3.96 3.37 2.53 1.60

quality 7.4
6.7
3.93
3.94
3.93 3.95
3.90
3.52
3.53
2.74
3.10
1.78
2.22 1.21
5.9 3.68 3.82 3.72 3.19 2.57 1.33
5.15 3.70 3.60 3.60 3.38 2.85 1.84
4.75 3.59 3.46 3.42 3.30 3.10 2.00
FR 3.50 3.50 3.14 2.74 1.50
HR 3.35 3.24 2.80 1.92

26 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Demo
• Speech samples

27 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Nokia AMR Link Adaptation

28 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Link Adaptation in AMR Codec (1/2)
• Link Adaptation is the capability of AMR feature to vary the codec used
according to the link conditions
• Both network, for uplink, and MS, for downlink, measure the radio
conditions in each link and take decisions on which codec should be
applied to each way  AMR codec mode adaptation is done
independently in UL and DL
• There are two link adaptation (LA) modes; the ETSI specified fast LA and
the Nokia proprietary slow LA
• slowAmrLaEnabled: if it is set to "N" (default) it is used ETSI fast LA;
if it is set to "Y" it used Nokia slow LA
• With slow LA, BTS allows in-band codec mode changes only on the
SACCH frame interval of 480 ms
• Two different types of link adaptation algorithms are defined: Codec
Mode Adaptation and Channel Mode Adaptation
• AMR codec mode adaptation algorithm adapts the bit-rate partitioning
between the speech and channel coding for a given channel mode to
track changes in the radio link and to account for specific input conditions
(speech signal characteristics, acoustic environmental characteristics,
etc.)
• AMR channel mode adaptation algorithm allocates a half-rate or full-rate
channel according to channel quality and the traffic load on the cell in
order to obtain the best balance between quality and capacity
29 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Link Adaptation in AMR Codec (2/2)

C/I EFR operation AMR mode

C/I AMR
30 Mode AMR
FR HR
25 12.2 kbit/s
12.2
10.2
20 7.95 kbit/s

7.95
7.4 7.4
[dB]

15
6.70 kbit/s

6.7 6.7
10 5.90 kbit/s
5.9 5.9
5.15 5.15
5
4.75 4.75
0
0 5 10 15 20 25 30

Time[s]

30 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Codec Mode Adaptation

31 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Codec Mode Adaptation
• Codec Mode Adaptation or Link Adaptation (LA) is the algorithm that selects
which codec has to be used each moment by the MS (in UL) or by the
network (in DL direction).

• The basic AMR codec mode sets for MS and BTS are provided by BSC via
layer 3 signalling

• Both the MS and the network implement their own C/I measurement
algorithms

• C/I measuremnt algorithms are vendor dependant / proprietary

• Nokia has common UL/DL link adaptation thresholds

32 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Objective of Codec Mode Adaptation
•Select the codec that provides the best speech
quality depending on radio conditions
4.5

3.5

2.5
FR 12.2 MOS
MOS

2 FR 7.4 MOS
FR 5.9 MOS
1.5 FR 4.75 MOS
HR 7.4 MOS
1
HR 5.9 MOS
0.5 HR 4.75 MOS

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

C/I (dB)

33 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Parameters for Link Adaptation
• When deploying AMR the following parameters are
important for the Link Adaptation:
• ACS (Active Codec Set) which defines the codecs
that can be used in a BTS during a call.
• Thresholds used: Defines the CIR value to change
the codec from a less robust codec to the
immediate more robust one in the ACS
• Hysteresis: the values in dB to add to the
thresholds in order to go from a robust codec to
the immediate less robust one in the ACS.
• For instance: ACS= [AFS12.2, AFS7.9, AFS4.75],
Thresholds: 12dB, 8dB, Hysteresis: 1dB, 1dB
• With these settings the change from codec AFS7.9
to AFS4.75 will happen when the CIR is below 8dB,
while from AFS4.75 up to AFS7.9 it will be with 9dB.

34 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Procedure in MS for AMR Link
Adaptation
CIR FadingProfile
Estimator

Meas i Measi+1 Meas i+2 Meas i+3 ...

t

Filter

Conversion to CIR and

Comparison with CIR norm compensation for
CMR Thresholds (LQE) Channel Profile

35 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Link adaptation
Codec mode adaptation, BTS Channel mode adaptation
level • FR <-> HR changed by handover
• Codec mode changed (packing and unpacking)
according to channel
• Based on BTS load (BSC level) and
conditions (UL/DL C/I)
channel condition (RxQual)
• Only up to four codecs can be
used during a call
• Goal—the highest MOS (Mean
Opinion Score)
• Mode indications inform the
receiver about the currently
applied codec mode
• Mode Command informs MS
about the codec mode to be
applied on the uplink

36 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 In band signalling

CMI,
CMR C MC
,
CMI

SF 2 SF 4CMR SF 6CMR CMR CMR CMR CMR CMR CMR

UL SF 1 SF 3
CMI SF 5CMI SF 7 CMI CMI CMI CMI CMI CMI CMI
8
TDMA

frames
SF 2 CMI SF 4 CMI SF 6 CMISF 8 CMI CMI CMI CMI CMI CMI

DL SF 1 SF 3
CMCSF 5 CMCSF 7 CMCSF 9CMC CMC CMC CMC CMC CMC

SF= Speech Frame CMC = Codec Mode Command time

CMI = Codec Mode Indicator CMR= Codec ModeRequest
37 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Impact of wrong LA
•Due to wrong LA threshold selection, or wrong estimation of
radio conditions, the codec used under certain conditions might
not be the best performing one, reducing Speech Quality.

non-ideal LA 1
(Lower MOS &
HIGH FER)

non-ideal LA 2
(Lower MOS &
low FER)
AFS475

AFS7.90

AFS12.2

Actual CIR for codec Actual CIR for codec

changes for LA 1 Ideal CIR for codec changes for LA 2
changes
38 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Channel Mode Adaptation

39 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel Mode Adaptation
• Channel Mode Adaptation is an intra BTS HO
algorithm that aims at select the correct channel rate
(FR or HR).
•The selection of the channel rate depends on 2 main
factors: load and quality

load Good
Quality
FR packing HR
FR unpacking HR
Bad
Qualit
y
40 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Packing Procedure

41 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel mode adaptation: Packing
• Handover between AMR FR and AMR HR is intra-cell (intra BTS) handover
• Spontaneous packing of FR AMR calls to HR AMR calls is triggered when the cell
load is high enough, the number of free full rate resources reduces below the
value of the parameter btsLoadDepTCHRate (HRL).
• Packing continues until the cell load is low enough, the number of free full rate
resources increases above the value of the parameter btsLoadDepTCHRate (HRU).

Free FR TCHs

Upper limit for free FR TCHs

btsLoadDepTCHRate(HRU)

Lower limit for free FR TCHs

btsLoadDepTCHRate(HRL)

Time
No packing of Packing of No packing of
AMR FR calls AMR FR calls AMR FR calls

42 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Packing from AMR FR to AMR HR (1/2)
• Spontaneous Packing of AMR FR to AMR HR call is triggered when
• free full rate resources reduces below the value of the parameter
btsLoadDepTCHRate(HRL) or btsSpLoadDepTCHRate (FRL)
• HRL is a BSC level parameter
• FRL is a BTS level parameter, once defined, it can overwrite
HRL
• AND FR calls which quality is better than amrHandoverFr(IHRF) for
both UL and DL
– Px: The Px of Threshold quality DL Px (QDP) is used.
– Nx: The Nx of Threshold quality DL Nx (QDN) is used.
• Note:
• Packing quality threshold is for both UL/DL threshold
• Packing quality does not have its own Px/Nx
• Packing quality does not have its own averaging windows and
weighting

43 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Packing from AMR FR to AMR HR (2/2)
• Packing happens to permanent HR channels
• Packing happens to DR channels which half has been
occupied
• Packing happens when there are even number of FR calls
to DR channels.
• Packing continues until the number of free full rate resources
increases above the value of the parameter
btsLoadDepTCHRate (HRU) or btsSpLoadDepTCHRate (FRU)
• HRU is a BSC level parameter
• FRU is a BTS level parameter, once defined, it can
overwrite HRU

• Packing is triggered by new TCH allocation

• Queueing is not allowed for packing procedure

44 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Un-Packing Procedure

45 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Unpacking from AMR HR to FR
• Spontaneous unpacking of AMR HR calls to AMR FR
calls is triggered when the quality of a AMR HR call
degrades below the amrHandoverHr(IHRH) for
either UL or DL
• Px: The Px of Threshold quality DL Px (QDP) is
used.
• Nx: The Nx of Threshold quality DL Nx (QDN) is
used.
• Queuing is allowed for unpacking procedure

46 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Nokia AMR Interaction
with Other Nokia Features

47 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Interaction with other features
• DADL/B

• New adjacent cell parameter to handover AMR calls from non-

AMR-capable cells to co-located AMR-capable cells during call
set-up phase

• Handover

• Prioritization of AMR capable cells during internal and external

handovers (AMR capable cells which load is low (BTS load
threshold (BLT) parameter), are on the top of the handover
target cell list)
• New RxQual thresholds for AMR FR and AMR HR
• New RxQual thresholds for HOs between AMR channel rates
(relates to AMR FR call packing and AMR HR call unpacking)

48 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Interaction with other features
• Power Control

• New RxQual thresholds for AMR FR and AMR HR

• IFH and IUO

• New good and bad C/I thresholds for AMR FR and AMR HR

49 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Radio Link Timeout
Background
• 3GPP 05.08 states that Radio Link Failure (RLF) in the MS is
determined by the success rate of decoding messages on the
downlink SACCH
• The aim of determining RLF in the MS is to ensure that calls with
unacceptable voice/data quality, which cannot be improved either
by RF power control or handover, are either re-established or
released in a defined manner
• The Radio Link Timeout (RLT) parameter controls that a forced
release (drop) will not normally occur until the call has degraded
to a quality below that at which the majority of subscribers would
have manually released it
• The RLF procedure is implemented in the RRM at the BSC and is
as follows:
• After the assignment of a dedicated channel a counter is
initialized to RLT
• When a SACCH message is unsuccessfully decoded the counter
is decreased by 1
• When a SACCH message is successfully decoded the counter is
increased by 2
• If the counter reaches 0 a RLF is declared  Call is released
50 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Radio Link Timeout
AMR FR vs. EFR - Test 1
EFR AMR

• The RLT is based on SACCH Number of

Number of unaccepta Number of
Number of
unaccepta

erased frames, which are Time RLT FER

una cceptable
samples
ble
sa mples
(FER>5 0%) ( FER>25
Time RLT FER
unacceptable
samples
RLT to
select
ble
samples

independent of speech frames

(FER>50%) (FER>25
%) %)
00:25:55:82 64 13
00:26:01:79 64 29 0 0 1
00:25:56:30 64 42
00:26:02:27 63 43 0 1 2
• The tests were aimed to find RLT 00:25:56:78
00:25:57:26
00:25:57:74
63
62
64
54
25
50
00:26:02:76
00:26:03:24
62
61
75
50
1
1
2
3
3
4

value producing the same speech 00:25:58:23

00:25:58:71
63
62
42
54
00:26:03:72
00:26:04:20
00:26:04:68
60
59
58
29
50
29
1
1
1
4
5
6
5
6
7
degradation in AMR as EFR would
00:25:59:20 64 38
00:25:59:68 64 50 00:26:05:16 57 70 2 7 8
00:26:00:16 63 54 00:26:05:65 56 21 2 8 8

suffer with default RLT value for 00:26:00:64

00:26:01:12
00:26:01:60
62
64
64
46
58
58
00:26:06:13
00:26:06:61
00:26:07:09
55
54
53
63
63
82
3
4
5
9
10
11
9
10
11
this traffic (i.e. 20) 00:26:02:08
00:26:02:56
00:26:03:05
63
62
61
46
67
75
00:26:07:58
00:26:08:06
52
51
92
78
6
7
12
13
12
13
00:26:08:54 50 58 8 14 14
00:26:03:53 60 58

• The indicator used was number of 00:26:04:01

00:26:04:49
59
61
67
58
00:26:09:02
00:26:09:50
00:26:09:99
49
48
47
71
92
91
9
10
11
15
16
17
15
16
17
BQS-FER (with FER>50%/25%)
00:26:04:97 60 75
00:26:05:45 59 54 00:26:10:47 46 67 12 18 18
00:26:05:93 61 58 00:26:10:95 45 88 13 19 19

between the time when the 00:26:06:41

00:26:06:89
00:26:07:37
60
59
58
79
58
88
00:26:11:43
00:26:11:91
00:26:12:40
44
43
42
88
83
100
14
15
16
20
21
22
20
21
22
counter starts decreasing from its 00:26:07:85
00:26:08:34
57
56
92
63
00:26:12:88
00:26:13:36
41
40
63
70
17
18
23
24
23
24

top value (64) to the point where

00:26:08:82 55 88 00:26:13:84 39 96 19 25 25
00:26:09:30 54 83
rivada: 00:26:14:32 38 88 20 26 26
00:26:09:78
RLT = 44 is equivalent to 53 88
00:26:14:80 37 80 21 27 27
the link would be released (i.e., 00:26:10:26
0 when using FI XED RLT 52
equal to 20 (DEFAULT
00:26:10:74
NOKIA VALUE)
00:26:11:23
51
50
88
88
75
00:26:15:28
00:26:15:76
39
38
88
82
22
23
25
26
28
29

when the counter is decreased by 00:26:11:71

00:26:12:19
49
48
75
83
00:26:16:25
00:26:16:73
00:26:17:21
37
36
35
95
78
92
24
25
26
27
28
29
30
31
32
the RLT parameter)
00:26:12:67 47 83
00:26:13:15 46 96
00:26:13:63 45 92
00:26:14:11 44 83 26 27

RLT has very high impact on

DCR
51 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Radio Link Timeout
AMR-FR vs. EFR – Test 2
• Aim is to evaluate when AMR-FR is used which RLT value will
result in comparable performance (point at which call is released)
to the recommended RLT for EFR
• The RLT is based on SACCH erased frames, which are
independent of speech frames. The principle of the tests is to find
RLT value producing the same speech degradation (FER > 15 % 
MOS < 1.5  no audible speech during 30 sec before dropping) in
AMR as EFR would suffer with default RLT value for this traffic (i.e.
20)

• The driving route started at a good coverage location and ended

at a bad coverage area

52 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR vs EFR: FER Before Dropping
RLT 20 RLT 28

100 100
90 90
80 80
70 70
60 60
FER (%)

FER (%)
50 50
40 40
30 30
20 20
10 10
0 0
30 to 25 25 to 20 20 to 15 15 to 10 10 to 5 5 to 0 30 to 25 25 to 20 20 to 15 15 to 10 10 to 5 5 to 0
seconds before dropping seconds before dropping

EFR AMR EFR AMR

RLT 32 RLT 36

100 100
90 90
80 80
70 70
60 60
FER (%)

FER (%)
50 50
40 40
30 30
20 20
10 10
0 0
30 to 25 25 to 20 20 to 15 15 to 10 10 to 5 5 to 0 30 to 25 25 to 20 20 to 15 15 to 10 10 to 5 5 to 0
seconds before dropping seconds before dropping

EFR AMR EFR AMR

FER average every 5 seconds, during last 30 seconds before

dropping for: RLT = 20, 28, 32, 36

53 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Dropped Call Experience
• Dropped call experience: how long terminal does not receive any
audible speech (MOS<1.5) before it drops
• During 30 seconds before dropping, FER AMR is lower than FEREFR
• FEREFR > 15 % during last 30 seconds before dropping, for RLT =
20 AMR, FER > 15 % EFR, FER > 15%
RLT = 10-15 seconds before dropping 25-30 s. before drop
20
RLT = 15-20 seconds before dropping
28
RLT = 18-22 seconds before dropping
32
RLT = 25-30 seconds before dropping
36
RLT = 35-40 s. before drop
xx

• Conclusion:
• AMR =36 has a similar dropped call experience to EFR
RLT RLT=20
54 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 AMR Support in Nokia System

55 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Elements Needed for AMR Feature
• Following network elements and mobile phone are needed to get
AMR system feature to work:
• BTS: Nokia Talk-Family DF6, Nokia Prime Site DF6, Nokia
MetroSite CXM3.0-2, Nokia Ultra Site CX3, Connect Site.
• S10.5
• TCSM: TCSM2 with an AMR capable pool (23)
• MSC: M10
• NMS: NetAct OSS3.1
• MS: AMR capable phones

56 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Support in Nokia BTS's

 Nokia 2nd Generation BTS (DE21 BTS:)

 Nokia's 2
nd
generation BTS will not support AMR.
 Nokia Talk-family BTS (DF34 BTS): TALK-family
Nokia's Talk-family BTS will have AMR support for FR BTS:
modes 4.75, 5.9, 7.4 and 12.2 as well as for HR modes Full Rate Half
rate
4.75, 5.9 and 7.4 (*). With this approach, the link 12.2
adaptation between full scale of FR modes and almost full 10.2
scale of HR can be achieved (note that only 4 codecs can 7.95 7.95
be selected to be used during a call). 7.4 7.4
 Nokia PrimeSite BTS: 6.7 6.7
NokiaPrimeSite will have same AMR support as Talk. The 5.9 5.9
5.15 5.15
frequency hopping functionality will be removed from
PrimeSite BTSs to enable this modification. Last PrimeSite 4.75 4.75
SW release supporting frequency hopping will be DF5.0.
 Nokia InSite BTS:
 Nokia InSite BTS will not support AMR
 Nokia MetroSite and UltraSite BTS:
 Nokia's MetroSite and UltraSite will have full AMR
support.

57 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Nokia AMR Parameters

58 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Link Adaptation Related Parameters

59 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Steps to configure Link Adaptation

• First, selection of the Active Codec Set (codecs to be used

during the calls for both FR and HR). FRC & HRC parameter
• Then, selection of CIR thresholds to change codecs and
hysteresis
• Selection of the initial codec to be used at the beginning of the
allocation (ICMI- FRI & FRS for FR and ICMI-HRI and HRS for
HR)
• Selection of standard Link Adaptation (every 40 ms) or Slow
Link Adaptation (every 480ms) (SLA)

60 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Link Adaptation Related Parameters
• BTS level parameters
Q3 NAME Acronymn RANGE UNIT MML MML Default

amrConfigurationFr: codecModeSet FRC kbit/s EQ 12.2 7.40 5.9 4.75

amrConfigurationFr: hysteresis1 FRH1 0 ... 15 0-7.5dB EQ 2
amrConfigurationFr: hysteresis2 FRH2 0 ... 15 0-7.5dB EQ 2
amrConfigurationFr: hysteresis3 FRH3 0 ... 15 0-7.5dB EQ 2
amrConfigurationFr: initCodecMode ICMI(FRI) 0 / 1 EQ 0
amrConfigurationFr: startMode FRS 00, 01, 10,11 EQ 0
amrConfigurationFr: threshold1 FRT1 0 ... 63 0-31.5dB EQ 8
amrConfigurationFr: threshold2 FRT2 0 ... 63 0-31.5dB EQ 14
amrConfigurationFr: threshold3 FRT3 0 ... 63 0-31.5dB EQ 22
amrConfigurationHr: codecModeSet HRC kbit/s EQ 7.40 5.90 4.75
amrConfigurationHr: hysteresis1 HRH1 0 ... 15 0-7.5dB EQ 2
amrConfigurationHr: hysteresis2 HRH2 0 ... 15 0-7.5dB EQ 2
amrConfigurationHr: hysteresis3 HRH3 0 ... 15 0-7.5dB EQ 0
amrConfigurationHr: initCodecMode ICMI(HRI) 0 / 1 EQ 0
amrConfigurationHr: startMode HRS 00, 01, 10,11 EQ 0
amrConfigurationHr: threshold1 HRT1 0 ... 63 dB EQ 22
amrConfigurationHr: threshold2 HRT2 0 ... 63 dB EQ 28
amrConfigurationHr: threshold3 HRT3 0 ... 63 dB EQ 0

61 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Definition of the Active Codec Set
• amrConfigurationFr: codecModeSet (FRC)
• & amrConfigurationHr: codecModeSet (HRC)

• Maximum of 4 codecs can be included in ACS, although it can be less (or even disable)

Codecs supported in different BTS models

BTS AMR FR AMR-HR

2nd Generat - -
Talk Family AFS 475, AFS590, AFS740, AFS122 AHS475, AHS 590, AHS 740
Metro & Ultrasite All codecs All exepct AHS790

Values
Range: 0..240 (0 or 1-4 values Range: 0..30 (0 or 1-4 values from
from these: 1, 2, 4, 8, 16, 32, 64 & 128) these: 1, 2, 4, 8 & 16)

0 (0000 0000) = disabled 0 (0000 0000) = disabled

1 (0000 0001) = 4.75 kbit/s 1 (0000 0001) = 4.75 kbit/s
2 (0000 0010) = 5.15 kbit/s 2 (0000 0010) = 5.15 kbit/s
4 (0000 0100) = 5.90 kbit/s 4 (0000 0100) = 5.90 kbit/s
8 (0000 1000) = 6.70 kbit/s 8 (0000 1000) = 6.70 kbit/s
16 (0001 0000) = 7.40 kbit/s 16 (0001 0000) = 7.40 kbit/s
32 (0010 0000) = 7.95 kbit/s
64 (0100 0000) = 10.2 kbit/s
128 (1000 0000) = 12.2 kbit/s

Example (default) 1001 0101 = (4.75, 5.90, 7.40 & 12.2) 0001 0101 = (4.75, 5.90 & 7.40)

62 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Codec mode adaptation: Threshold and
hysteresis
• Both Threshold and hysteresis has 0.5 dB step
• One threshold to go from one codec to the closest
higher or lower one
• FR Example:
• AFS122, C/I < 11 dB, AFS122  AFS740
• AFS740, C/I > 11+1 dB, AFS740  AFS122

Codec Mode Threshold (C/I) Hysteresis (C/I)

(kbit/s)
4 (12.2) TH3 (11 dB)
3 (7.4) TH2 (7 dB) H3 (1 dB)
2 (5.9) TH1 (4 dB) H2 (1 dB)
1 (4.75) H1 (1 dB)

63 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Codec mode adaptation: Threshold and
hysteresis
Codec
mode

FR12.2
(codec 4)
1dB
FRH3

FR7.4
(codec 3) 1dB
FRH2

FR5.9 (codec 2)
1dB
FRH1

C/I
FR4.75
(codec 1) 4dB 5dB 7dB 8dB 11dB 12dB
estimati
FRT1 FRT2 FRT3 on

64 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Initial Codec Mode Indicator
• Separate parameters for FR and HR:
FR:
• amrConfigurationFr: initCodecMode (ICMI) (FRI)
• Initial codec mode for call set-up and HO
• ”0” = Initial codec mode is defined by the implicit rule
provided in GSM 05.09
• ”1” = Initial codec mode is defined by amrConfigurationFr:
startMode (FRS) 00: Codec mode 1 (most robust within ACS)
01: Codec mode 2
10: Codec mode 3
HR: 11: Codec mode 4

• amrConfigurationHr: initCodecMode (ICMI) (HRI)

• amrConfHrStartMode (HRS)
• HR 7.4 cannot be used as start mode

65 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Slow Link Adaptation
• slowAmrLaEnabled (SAL): ”Y/N”
• enable slow link adaptation. This is a proprietary
algorithm where codec mode changes happen every
SACCH period (480ms) instead of as fast as 40ms.

66 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Report 245:
Distribution of call samples UL/DL by FER
classes
The reported codec
is the last used
ND 245 codec of reported
SACCH period.

DL codec and FER

distribution is
avaibable from
Drive Test tool.

67 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel allocation Parameters
(Call set-up and Packing/Unpacking)

68 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 IAC (Initial AMR Channel Rate)
• initAmrChannelRate (IAC):
• ”1” = Any rate. Channel type allocation depends
on further network parameters/settings
• ”2” = AMR FR
• AMR FR is preferred over AMR HR and
allocated despite of the values of the currently
used information for channel allocation
• IAC=2 overrides tchRateInternalHo (HRI)
parameter

Q3 NAME Acronymn RANGE UNIT MML MML Default

initAmrChannelRate IAC 1 / 2 EE 1

69 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Packing/Unpacking Parameters and Channel Rate
allocation
• BSC level parameters

Q3 NAME Acronymn RANGE UNIT MML MML Default

btsLoadDepTCHRate HRL 0 ... 100 % EE 100

btsLoadDepTCHRate HRU 0 ... 100 % EE 0

• BTS level parameters

Q3 NAME Acronymn RANGE UNIT MML MML Default

btsSpLoadDepTCHRate (BTS level) FRL 0 ... 100 % EQ 100

btsSpLoadDepTCHRate (BTS level) FRU 0 ... 100 % EQ 0
amrHandoverFr IHRF 0 ... 7 EH 0
amrHandoverHr IHRH 0 ... 7 EH 4

70 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel allocation: Call set-up and
handovers
• btsLoadDepTCHRate (HRL) and btsLoadDepTCHRate (HRU)
are considered in call set-up and handovers only when IAC=1
• HR is to be assigned if free resources go below HRL
• FR is to be assigned if free resources go above HRU

• In channel allocation for Handovers, there is an additional

parameter that can be used to set further control
• tchRateInternalHo (HRI) is used to control the speech
and channel type changes in handover when IAC=1
• If set HRI=1, channel type and speech codec used in
source BTS are primarily allocated in the target BTS

71 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel mode adaptation: Packing

1. New channel allocation

2. BSC make Intra-cell HO

Case 1. (Packing TS4&5 calls into HR calls TS7)
Free FR resources increased by one

1. New channel allocation

2. Packing TS4 FR call into HR call TS7

Case 2. Free FR resources increased by one

TCHF FR call
MBCCHC TCHD HR call

72 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel mode adaptation: Packing

1. New channel allocation

2. No packing
Case 3. (due to lack of TCHD resources)

• BSC performs packing for FR AMR calls only when Rxqual

is above the amrHandoverFr (IHRF)

73 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel mode adaptation: Unpacking

1. Unpacking due to Rx qual

Case 4.

2. Packing due to load

• Unpacking of HR AMR calls to FR AMR calls is triggered when the

quality of a HR AMR call degrades below the amrHandoverHr
(IHRH).
• IHRH = 3, RxQual 3 to 7 will be unpacked
• Cell load does not have an effect

74 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Packing/Unpacking RXQUAL Threshold
Selection from Field Tests

75 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 RXQUAL vs. aveFER in 2sec PERIODs

Averaged RXQual distribution in ROUTE 1 (TEMS) Average FER in 2sec periods

50% AMR-HR AHS4.75-AHS7.4 16dB Thr & 3dB Hyst
45%

40% 25

35%
20
30%

Average FER
25%
%

15
20%

15% 10
10%

5% 5

0%
RXQUAL 0 RXQUAL 1 RXQUAL 2 RXQUAL 4 RXQUAL 4 RXQUAL 5 RXQUAL 6 RXQUAL 7 0
RXQUAL 1 RXQUAL 2 RXQUAL 4 RXQUAL 4 RXQUAL 5 RXQUAL 6 RXQUAL 7
Averaged RXQual distribution in ROUTE 1
Average FER in 2sec

According to the average FER, RXQual 5 can be considered still as

providing enough quality to serve AMR-HR. Also note that in such
conditions, most robust HR codecs will be used but for high
capacity networks this would be enough

76 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Correlation RXQUAL vs FER in 2sec
PERIOD
With RXQual 5, still AMR-HR can provide enough quality to serve
traffic if high capacity numbers are required.
(TEMS) Distribution of FER in 2sec per RXQUAL class for AMR-HR AHS4.75-AHS7.4 16dB and 3dB Hyst

100.00%

90.00%

80.00%

70.00% RXQual 1
RXQual 2
60.00%
RXQual 3
50.00% RXQual 4
40.00% RXQual 5
RXQual 6
30.00%
RXQual 7
20.00%
10.00%
.00%
RXQ
ual 7
RXQ
ual 6
%
FE
R
ER RXQ
RXQ
ual 5 10% of
>0 %F
ual 4
>2
4 %
FE
R
R RXQ
RXQ
ual 3 samples
> FE ual 2
>8
%
FE
R RXQ
ual 1 having worse
2%
>1 than 4% FER
WITH RXQual
5
77 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 TEMS Snapshot

• In an
environment
around RXQual
5, less than 2
frames erased
per SACCH
period.
• AMR-HR with
RXQual 5 could
be used when
trying to have
agressive HR
penetration

78 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Packing/Unpacking threshold selection
• From the performance of AMR-HR it can be concluded that RXQual 4
might be a safe value for packing from FR to HR in order to ensure
proper quality of AMR-HR.
• If more agressive deployment is required and more penetration of
AMR-HR is aimed at, then RXQual 5 threshold can be used still safely.

(NOTE: In the current implementation of BSS10.5 calls pack when

RXQual is < IHRF and unpack when RXQual is >= IHRH.)

79 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Channel mode adaptation: Additional
notes
• HRL and HRU are set on BSC level but load evaluation is based on
working FR TSLs for CS in individual BTS.
• Lower limit for FR TCH resources (FRL) and upper limit for FR TCH
resources (FRU) are BTS specific parameters. They have priority
over (HRL) and (HRU)
• The feature is disable when set HRU <= HRL or FRU <= FRL
• Requirements for maximum usage of HR
• HRU = 99
• HRL = 98
• IAC = 1
• IHRF = 3, RX Qual 0, 1, 2, 3 are required for packing
• IHRH = 4, Rx Qual 4, 5, 6, 7 triggers unpacking
• TRIH = 0, no restriction, channel rate selection based on HRI

80 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Other parameters affecting channel
allocation in HO´s

81 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 amrConf lnHandover
• amrConf lnHandover
• ”1” = the currently used multirate configuration is preferred
• ”2” = the multirate configuration of target BTS is preferred

• amrSetGradesEnabl ”Y/N”
• Y = downgrades and upgrades are applied
• N = downgrades and upgrades are not applied

• If multirate configuration of source and target BTS are the same,

these 2 parameters has no impact.

82 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 amrConflnHandover
• If multirate configuration of source and target BTS are difference
(e.g. Talk family BTS supports less codecs than UltraSite and
MetroSite), the multirate configuration can be aligned before or
after HO.
• Recommendation
• amrConflnHandover = ”2”
• amrSetGradesEnabl = ”Y”

• Mode modify is triggered for BTS and MS on source side before HO

if target BTS support less codec (downgraded, UltraSite  Talk
family)
• Mode modify is triggered for BTS and MS on target side after HO if
target BTS support more codec (upgrade, Talk family  UltraSite)
• In order to make it possible to connect unidirectional speech path
on target side, the multirate configuration on both sides should be
the same (reduce muting period during HO).

83 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Report 248: Codec set modification failure
ratio

ND 248

84 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 TCH in handover (HRI)
With this parameter you define the traffic channel
allocation during BSS internal or external handovers. The
parameter controls the target cell selection and the TCH
channel rate and speech codec determination in traffic
channel allocation.

• HRI = 1 Call serving type TCH and speech codec are

preferred. The call serving type of speech codec inside the call
serving
•type of TCH can change.
HRI = 2
• Call serving type of TCH and speech codec are
preferred for speech
• Channel rate change is possible for data if the radio
interface data rate allows it

85 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 TCH rate intra-cell handover (HRI)
• HRI = 3 Channel rate and speech codec changes are totally denied

• RHI = 4 Preferred channel rate of TCH and

preferred speech codec have to be primarily
allocated
• RHI = 5 ... TCH has to be primarily allocated from
the best BTS of thehandover candidate list.

Recommended value = 1 (default) OR 4

86 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 TCH rate intra-cell handover (TRIH)
• With this parameter you control the TCH channel
rate determination in TCH allocation and the TCH
speech codec to be allocated during internal intra-
cell handover.
•TRIH = 0 No Constraints. Follows HRI settings
•TRIH = 1 Call serving type TCH and speech codec are
preferred Preferred over different type TCH and speech codec

Only when same type TCH and speech codec is not avaialble
• TRIH = 2
• Call serving type of TCH and speech codec are
preferred for speech
• Channel rate change is possible for data if the radio
interface data rate allows it

87 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 TCH rate intra-cell handover (TRIH)
• TRIH = 3 Channel rate and speech codec changes are totally denied

Not allowed
x

Allowed

• TRIH = 4 Preferred channel rate of TCH and preferred speech

codec have to be primarily allocated

Recommended value = 0 (default)

88 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR specific Hand-Over and Power Control
Parameters

89 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Power Control Parameters

• Separate thresholds for AMR-HR

• Different PC thresholds for AMR than for
EFR
• AMR allows to use more aggressive PC
thresholds (1-2 classes lower)
• Same voting used as for EFR (px/nx)

90 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Parameters
• AMR PC Control Thresholds - BTS level parameters

Q3 NAME Acronymn RANGE UNIT MML MML Default

amrPowerControlFr *lower threshold dl Rx Qual* LDRF 0 ... 7 EU 4

amrPowerControlHr *lower threshold dl Rx Qual* LDRH 0 ... 7 EU 3
amrPowerControlFr *lower threshold ul Rx Qual* LURF 0 ... 7 EU 4
amrPowerControlHr *lower threshold ul Rx Qual* LURH 0 ... 7 EU 3
amrPowerControlFr *upper threshold dl Rx Qual* UDRF 0 ... 7 EU 0
amrPowerControlHr *upper threshold dl Rx Qual* UDRH 0 ... 7 EU 0
amrPowerControlFr *upper threshold ul Rx Qual* UURF 0 ... 7 EU 0
amrPowerControlHr *upper threshold ul Rx Qual* UURH 0 ... 7 EU 0

91 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Impact at system level of Power Control
Parameters

EFR vs AMR
Performance vs. PC settings
4.5%

4.5% 4.0%
100% EFR PC=2/3
4.0%
100% AMR 3.5%
% Bad Quality S am ples

3.5%
3.0%
3.0%

T CH fER > 4%
2.5%
2.5%
2.0% 2.0% 140 %
1.5% 1.5%

1.0%
1.0%
0.5%
0.5%
100% AMR PC=3/5
0.0%
P C=5/3 PC=5/4 PC=4/3 PC= 3/2 0.0%
PC se ttings 5% 10% 15% 20% 25% 30% 35%
EFL( %)

TCH FE R > 4% % ended call avg estimated MOS < 3.5 Codec MA EFR

92 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 RXQual Handover Parameters

• Separate thresholds for AMR-FR than for AMR-

HR
• Different HO thresholds for AMR than for EFR
• Same px/nx values used for both AMR and
EFR
• AMR allows to use more aggressive HO
thresholds (1-2 classes lower)
Q3 NAME • BTS level parameters
Acronymn RANGE UNIT MML MML Default

Threshold dl Rx qual AMR HR QDRH 0 ... 7 EH 4

Threshold dl Rx qual AMR FR QDRF 0 ... 7 EH 5
Threshold ul Rx qual AMR FR QURF 0 ... 7 EH 5
Threshold ul Rx qual AMR HR QURH 0 ... 7 EH 4

93 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Thresholds Overview

• 7
• 6
• 5 Qual_reason_HO for FR UDRF, UURF

• 4 unpack HR -> FR
• 3 Qual_reason_HO for HR; LDRF, LURF
• 2 pack FR -> HR UDRH, UURH
• 1
• 0 LDRH, LURH

94 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Handover Prioritization
When there are cells
• In order to facilitate the • Call set-up with AMR and
continuation of an AMR call… • without
In AMR environment, AMR
DADL/B is used
• HO target cell list is sorted so that to handover AMR calls from non-AMR-
lightly-loaded AMR-capable cells capable cell to co-located AMR capable
are more attractive. cells during call set-up phase.
• AMR-capable cells can be defined • Handover
by the AMR target cell of direct • AMR cells which load below
access to desired layer (DADLA) BTSLoadThreshold and meet
• Prioritized by BTS Load Threshold hoMarginPBGT are prioritised for AMR
(BLT) call.

• Prioritization only used for an on-

going AMR-mode call
1) DADL/B used to direct AMR 2) Prioritisation of AMR capable
mobiles to AMR capable cells cells in handovers

No-AMR SDCCH
2nd gen. BTS
capable
AMR capable
UltraSite TCH
(co-located)

95 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Relation with other HOs: Priority
1. Interference (UL or DL) 8. Downlink level
2. NonBCCHLayerExit 9. MS-BS Distance
3. Uplink quality 10. Turn-around-corner MS
4. Downlink quality 11. Rapid field drop
5. AMR unpacking due to UL 12. Fast/Slow moving MS
level HO threshold and UL
quality (unpacking is started 13. Better cell (Power budget HO or
instead of UL level based HO if Umbrella HO)
both, UL level and UL quality 14. Load based HO in Common
for unpacking triggers) BCCH from BCCH/non-BCCH
6. Uplink level layer to non-BCCH layer
7. AMR unpacking due to DL 15. AMR packing
level HO threshold and DL 16. AMR unpacking
quality (unpacking is started
instead of DL level based HO if
both, DL level and DL quality
for unpacking triggers)

* Priority applies when criteria are fulfilled at the same time

96 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Nokia AMR KPI

97 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Performance Monitoring
• AMR brings new speech codecs and channel encoders into air interface
• Speech Codecs impact would be reflected in the speech quality
perceived
• Channel encoding enhancement is translated into better error
correction capabilities (lower FER for same RawBER), which, in turn,
enhances the speech quality
• Traditional performance monitoring indicators based on RXQUAL
distribution, or Drop Call Rate, etc. does not reflect clearly the
perceived speech quality by the end user.
• New methods to monitor the performance of AMR closer to speech
quality perceived by the user. 2 Main indicators are proposed: FER and
O-MOS (Objective MOS).
• O-MOS is not simple to measure
• (BSS 10.5) FER is available in UL, but DL FER will be reported when R
´99 MS come. In the meantime DL FER is just estimated from
RXQUAL values

RXQUAL FER O-MOS

98 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR KPIs

• RXQUAL: Reported raw bit error rate

• RXLEV: Reported received power
• FER: Frame Erasure Rate (after decoding)
• Codec distribution
• MOS - speech quality
• FER per codec
• Call Drop Rate and other standard benchmarking measures

Microsoft Word
• Network collected measurements: Document

99 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 MOS & FER
Example of relation MOS-FER:
4-5% FER seems to be acceptable for almost all codecs
 M OS Perceived quality (MOS) degradation as a function of the FER  M OS Perceived quality (MOS) degradation as a function of the FER
(FR Tests in Clean Speech) (HR Tests in Clean Speech)
0.50 0.50

0.00
GSM 0.00
06.7
-0.50
5 -0.50

-1.00 -1.00
7.95 HR

-1.50 -1.50 7.4 HR

12.2
10.2 6.7 HR
7.95 FR
-2.00 -2.00 5.9 HR
7.4 FR
6.7 FR
5.9 FR 5.15 HR
-2.50 5.15 FR -2.50
4.75 FR 4.75 HR
FER FER
-3.00 -3.00
0.001% 0.010% 0.100% 1.000% 10.000% 100.000% 0.001% 0.010% 0.100% 1.000% 10.000% 100.000%

MOS degradation for FR MOS degradation for HR

100 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Effect on HO_QUALITY
• RXLEV and Power Budget HO parameters identical for AMR
and EFR
• AMR call would handover at the same point as an EFR call.

• Separate RXQUAL threshold settings for AMR

• Default set to “worse” values than EFR. (e.g. EFR =4,
AMR = 5)
• With these default settings AMR calls would be expected
to have fewer HO due to quality
• No difference in RXQUAL measurement method between EFR
and AMR
• EFR call and AMR call in identical location should show
identical RXQUAL measurements
• Packing/Unpacking
• Unpacking from HR to FR is always based on RX quality
• In congested cell with no available TS for unpacking, Inter-
cell HO required based on RXQUAL.

101 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Effect on HO_Failures
• Improved robustness in AMR over EFR
• AMR better able to handle poor radio conditions - low RXLEV,
poor RXQUAL, low C/I
• Separate RXQUAL HO threshold for AMR
• Effort to “squeeze” more performance from AMR could have
negative impact in case parameters are not properly set up
and/or traffic is low
• Packing/Unpacking
• Congestion could cause negative impact to HO_Fail
• No available TS for unpacking within cell. Inter-cell HO
required

• Conclusion
• Optimization of separate AMR parameters is important to
ensure no negative impact to HO_Failures. Different
environments will need different parameter settings to
optimise the performance.
• Unpacking algorithm under congested conditions may
negatively impact HO_Failures

102 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Effect on DCR
• The AMR feature itself will not impact the individual
connections DCR, but it will affect the overall system
DCR since the interference generated in the network is
lower due to the AMR power control settings.

• Radio Link Timeout can be adapted to AMR in order for

dropped calls to maintain the same correlation with
voice quality degradation as with EFR (RLT value could
be moved from 20 to 36, for instance)

103 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Effect on BH_Congestion
• Use of AMR HR will decrease BH_Cong
• Special dimmensioning techniques are required:
• % of HR capable terminals -> system level
• Load in busy hour (C/I distribution) determines % of HR
terminals which can use HR -> cluster level
• HR Dimmensioning tables determine the traffic (Erlangs) to be
carried for a certain TSL configuration, a certain grade of
service and % of HR traffic -> cell level

104 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Network Doctor Reports related to AMR
• REPORTS:
• Report 246:
• Percentage of call time using non-AMR, AMR-FR and AMR-HR
• UL/DL RxQual classes for non-AMR, AMR-FR and AMR-HR
• UL/DL FER classes for non-AMR, AMR-FR and AMR-HR ND 246

• Report 244:
• Distribution of call samples UL/DL by codecs and RxQual
classes
• In FLA the codec mode reported is the last used in 480ms
measurement interval (statistics will be fully accurate for SLA)ND 244
• Report 245:
• Distribution of call samples UL/DL by FER classes
ND 245

• Report 247: Call Failure rate per codec type

ND 447

• Report 248: Codec set modification failure ratio ND 248

105 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Nokia AMR Planning Aspects

106 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Link Level Performance of AMR

107 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 C/I vs. FER performance, AMR FR
Frame Error Rates (FER) for ARM, EFR and FR Codecs
on Different C/I conditions (FR -Channel)
C/I [dB]
16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
0.00%

2.00%

GSM EFR
4.00%
GSM FR
5- AMR 12.2 kbit/s
6dB 6.00% AMR 10.2 kbit/s

FER [%]
AMR 7.95 kbit/s
8.00% AMR 7.4 kbit/s
AMR 6.7 kbit/s
AMR 5.9 kbit/s
10.00% AMR 5.15 kbit/s
AMR 4.75 kbit/s
12.00% GSM HR
TU3-iFH
14.00%

108 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 C/I vs. FER performance, AMR HR
Frame Error Rates (FER) for ARM HR Codecs
on Different C/I conditions (HR -Channel)

C/I [dB] Better

16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
Performance
0.00%
than GSM-HR
2.00%
and GSM-FR
(previous slide)
4.00%
GSM HR
6.00% AMR 7.95 kbit/s

FER [%]
AMR 7.4 kbit/s
8.00% AMR 6.7 kbit/s
AMR 5.9 kbit/s
10.00% AMR 5.15 kbit/s
AMR 4.75 kbit/s

TU3-iFH 12.00%

14.00%

109 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 UL lab results for fixed codecs and LA
TU50 non hopping, UL: Speech Quality TU3 RH5, UL: Speech Quality TU3 non hopping, UL: Speech Quality

4.0000 4.0 4.0

3.5000 3.5 3.5

3.0000 3.0 3.0

2.5
2.5000 2.5

PESQ
2.0
PESQ

PESQ
2.0000
•Note that only for 2.0

1.5
the case of TU3-rfh5
1.5000

the real EFR codec 1.0 Speech Quality 1.5

(PESQ-MOS)
1.0000 1.0
has been used to 0.5
0.5000
assess PESQ-MOS 0.0
0.5

0.0000 and FER. For other 0 2 4 6 8 10 12 14 16 0.0

0 2 4 6
cases,
8 10
it 12has14been
16 CIR(dB) 0 2 4 6 8 10 12 14 16
CIR(dB) CIR(dB)
only AFS12.2 AFS475(1.C.1), PESQ AFS122(1.C.2), PESQ AFS740(1.C.3), PESQ
AFS475(1.A.1), PESQ AFS122(1.A.2), PESQ AFS740(1.A.3), PESQ AFS590(1.C.7), PESQ LAdef(1.C.4),PESQ LAopt1(1.C.5),PESQ AFS475(1.B.1), PESQ AFS122(1.B.2), PESQ AFS740(1.B.3), PESQ
AFS590(1.A.7), PESQ LAdef(1.A.4),PESQ LAopt2(1.A.6) EFR(1.C.8) AFS590(1.B.7), PESQ LAdef(1.B.4),PESQ LAopt2(1.B.6),PESQ

TU50 non hopping, UL: FER(%) TU3 RH5, UL ( lab measurements) TU3 non hopping, UL: FER(%)

100.00% 100.00% 100.00%

10.00% 10.00% 10.00%

FER(%)
FER(%)
FER(%)

1.00% 1.00% 1.00%

0.10% 0.10% 0.10%

FER
0.01% 0.01% 0.01%
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16

CIR(dB) CIR(dB) CIR(dB)

AFS475(1.A.1) AFS122(1.A.2) AFS740(1.A.3) AFS475(1.C.1) AFS122(1.C.2) AFS740(1.C.3) AFS590(1.C.7) AFS475(1.B.1) AFS122(1.B.2) AFS740(1.B.3)
AFS590(1.A.7) LAopt1(1.A.5) LAopt2(1.C.6) Ldef(1.C.4) LAopt1(1.C.5) EFR AFS590(1.B.7) Ldef(1.B.4) LAopt2(1.B.6)

110 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Comparison UL lab results vs.
Simulations
TU50 non hopping, UL: FER(%) TU3 RH5, UL ( lab measurements) TU3 non hopping, UL: FER(%)

100.00% 100.00% 100.00%

10.00% 10.00% 10.00%

FER(%)
FER(%)
FER(%)

1.00% 1.00% 1.00%

0.10% 0.10% 0.10%

Lab
0.01%
0 2 4 6 8 10 12 14 16
0.01%
0 2
Measurements
4 6 8 10 12 14 16
0.01%
0 2 4 6 8 10 12 14 16

CIR(dB) CIR(dB) CIR(dB)

AFS475(1.A.1) AFS122(1.A.2) AFS740(1.A.3) AFS475(1.C.1) AFS122(1.C.2) AFS740(1.C.3) AFS590(1.C.7) AFS475(1.B.1) AFS122(1.B.2) AFS740(1.B.3)
AFS590(1.A.7) LAopt1(1.A.5) LAopt2(1.C.6) Ldef(1.C.4) LAopt1(1.C.5) EFR AFS590(1.B.7) Ldef(1.B.4) LAopt2(1.B.6)

TU50 no-hop, DL ( Simulations) TU3 RH5, DL ( Simulations) TU3 no-hop, DL ( Simulations)

100.00% 100.00% 100.000%

10.00% 10.00% 10.000%

TCH FER
TCH FER

TCH FER
1.00% 1.00% 1.000%

0.10% 0.10% 0.100%

Simulations
0.01% 0.01% 0.010%

0 5 10 15 0 5 10 15 0 5 10 15
C/I [dB] C/I [dB] C/I [dB]

AFS4.75 AFS5.90 AFS7.40 AFS12.2

fs475_RH5 fs590_RH5 fs740_RH5 fs122_RH5 AFS4.75 AFS5.90 AFS7.40 AFS12.2

111 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Coverage Enhancement

112 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR Coverage Improvement (link
budget)
AMR 12.2 AMR 5.15
Uplink Downlink Uplink Downlink

Transmitter

TX power1 30 dBm 45 dBm 30 dBm 45 dBm

Combiner losses 0 3.4 dB 0 3.4 dB

Cable and 0 2 dB 0 2 dB
connector losses

Body loss 3 dB 0 3 dB 0

TX antenna gain 0 dBi 15 dBi 0 dBi 15 dBi

Receiver:

Antenna gain 15 dBi 0 dBi 15 dBi 0 dBi

Body losses2 0 dB 3 dB 0 dB 3 dB

Link:

SNR requirement 9 dB4 9 dB4 4 dB 4 dB

Receiver sensitivity -107.4 -103.4 -112.4 -106.4

Max. allowed path 149.4 154 155.4 159

loss

Fading marging 7.4 7.4 7.4 7.4

Range [km]

4) Estimated performance for 1% FER in TU3 multipath channel with frequency

hopping
113 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
AMR FR: DL TU50 non-hopping – MOS & FER
MOS/FER vs C/I TU50 NO HOP DL

4 16

3.5 14

3 12

~6 dB
2.5 10
EFR MOS
AMR FR MOS
MOS

FER
2 8
12.2 FER
AMR FR FER
1.5 6

1 4

0.5 2

~6 dB
0 0
2 4 6 8 10 12 14 16 18
C/I

114 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR FR & HR: DL TU3 FH5 – MOS & FER
MOS/FER vs C/I TU3 FH5 DL

4 50

45
3.5

40
3
35
~4 dB
2.5 EFR MOS
30
AMR FR MOS
AMR HR MOS
MOS

FER
2 25
12.2 FER
AMR FR FER
20
1.5 AMR HR FER

15
1 ~4 dB
10

0.5 ~1 dB
5

0 0
2 4 6 8 10 12 14 16 18
C/I

115 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Field Tests:Sentivitiy Enhancement of
AMR
AMR
EFR

100 100

90 90

80 80

70 70 RXLEV -110 to -108

60 60 RXLEV -108 to -106

50 RXLEV -106 to -104

50
RXLEV -104 to -102
40 40
RXLEV -102 to -100
30 30
RXLEV -100 to -98
20 20
to -4
7 o -47
V -96 -96 t RXLEV -98 to -96
XLE -96 R10XLE V o -96 10
R
V -98 to
8 E V -98 t 8
RX LE to -9 0 RX
L to -9 RXLEV -96 to -47
V -100 00 E V -100 00
0
RX LE to -1 RX L to 1
-
V -102 02 FE E V -102 02 FE
RX LE to -1 FE R RX L to -1 FE R
V -104 04 FE R 10 E V -104 04 FE R 10
RX L E to -1 FE R 12 0% RX L to -1 FE R 12 0%
V -106 06 FE R 8- -1 E V -106 06 FE R 8- -1
RX L E to -1 5- 12 00 RX L to -1 5- 12 00
V -108 FE R 8% % % V -108 FE R
4- 8% % %
E 8 4 LE 08 FE R
RX L o -10 FE R -5 RX to -1 5%
V - 110 t FE R 2-
4
%
E V -110 FE R 2-
4%
RXLE R 0- % RXL R 0-
<= 2% <= 2%
0 0

The higher robustness of AMR is clear, finding similar behaviour of

AMR in [–104,-106] dBm as with EFR in the margin [–100,102], so,
at least around 4dB coverage gain.

116 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Dimensioning with AMR-HR

117 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Dimensioning with AMR-HR
• Erlang B table can not be used for dimensioning when AMR-HR
penetration is foreseen
• That is, if all TSL are configured as DR, then the maximum
traffic that can be served with less than 2% GoS is not the one
provided by Erlang-B table when doubling the amount of
channels.
• The reason is because now AMR-HR is not used in all TSL but
only in those that has good radio conditions.
• New tables have been produced taking the percentage of traffic
that could be handled by HR.
• These tables have been produced using a mathematical model
based on Markov processes (see reference at the end of the
material).

118 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Example of reduction of TSL required
with AMR-HR
Saving Factor = % of resources (TS) saved by using AMR-HR

2% GoS % of users with good conditions to use AMR-HR (for example C/I > 12dB)
# Time Slots 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1 0.0% 0.2% 0.5% 1.2% 2.3% 3.7% 6.5% 7.5% 17.2% 34.0% 50.0%
2 0.0% 0.8% 2.5% 5.2% 9.2% 13.9% 19.7% 25.7% 32.5% 41.4% 50.0%
3
4
0.0%
0.0%
1.0%
1.3%
3.2%
4.1%
By using AMR-HR,
6.6%
8.1%
11.1%
12.9%
16.2%
18.3%
22.1%
24.1%
28.3%
30.0%
35.3%
36.5%
42.6%
43.2%
50.0%
50.0%
5 0.0% 1.6% 4.8% when 70% of the
9.2% 14.3% 19.6% 25.3% 31.0% 37.2% 43.6% 50.0%
resources
Saving in

6 0.0% 2.0% 5.4% 10.0% 15.2% 20.5% 26.1% 31.7% 37.6% 43.8% 50.0%
7 0.0% 2.2% 5.9% network has
10.6% 15.9% 21.1% 26.6% 32.1% 38.0% 44.0% 50.0%
8
9
0.0%
0.0%
2.5%
2.8%
6.4%
6.7%
conditions for AMR-
11.1%
11.5%
16.4%
16.7%
21.6%
21.9%
27.0%
27.3%
32.5%
32.7%
38.2%
38.4%
44.1%
44.2%
50.0%
50.0%
10 0.0% 3.0% 7.0% HR, we can save
11.8% 17.1% 22.2% 27.6% 33.0% 38.5% 44.3% 50.0%
11
12
0.0%
0.0%
3.1%
3.2%
7.3%
7.4%
34% if 24 AMR-HR
12.1%
12.3%
17.3%
17.5%
22.5%
22.7%
27.8%
28.0%
33.1%
33.3%
38.7%
38.8%
44.3%
44.4%
50.0%
50.0%
13 0.0% 3.3% 7.6% capable TS are
12.5% 17.7% 22.9% 28.1% 33.4% 38.9% 44.4% 50.0%
14 0.0% 3.4% 7.7% 12.6% 17.9% 23.0% 28.2% 33.5% 38.9% 44.5% 50.0%
15 0.0% 3.5% 7.9% available (we would
12.8% 18.0% 23.1% 28.4% 33.6% 39.0% 44.5% 50.0%
16 0.0% 3.6% 8.0% need 36 TS with only
12.9% 18.1% 23.2% 28.4% 33.7% 39.1% 44.5% 50.0%
24 0.0% 4.0% 8.6% 13.6% 18.7% 23.8% 28.9% 34.1% 39.4% 44.7% 50.0%
32 0.0% 4.2% 8.8% FR to serve the same
13.8% 18.9% 24.0% 29.1% 34.3% 39.5% 44.8% 50.0%
40
48
0.0%
0.0%
4.2%
4.2%
8.9%
9.0%
13.9%
14.0%
traffic)
19.1%
19.2%
24.2%
24.3%
29.3%
29.4%
34.4%
34.5%
39.6%
39.7%
44.9%
44.9%
50.0%
50.0%
56 0.0% 4.2% 8.9% 14.0% 19.3% 24.3% 29.5% 34.6% 39.8% 44.9% 50.0%

119 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 AMR-HR DIMENSIONING ISSUES
• Erlang B table can not be used for dimensioning when AMR-HR
penetration is foreseen
• this table displays the traffic that can be served for different number of
Time Slots (TS) available and different penetrations of AMR-HR
• for example, 70% HR penetration the number of TS required to serve
16.7 Erlangs is just 16, while for pure AMR-FR it would require around
24 time slots to serve the same traffic (around 33% saving in resources)
2% GoS % of users with good conditions to use AMR-HR
# Time Slots 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1 0.0204 0.02 0.02 0.02 0.02 0.03 0.03 0.03 0.06 0.11 0.22
2 0.2236 0.23 0.24 0.25 0.28 0.32 0.37 0.45 0.58 0.79 1.09
3 0.6024 0.62 0.65 0.70 0.77 0.87 1.01 1.19 1.44 1.80 2.28
4 1.0927 1.12 1.18 1.28 1.42 1.59 1.82 2.10 2.47 2.97 3.63
5 1.6578 1.71 1.81 1.96 2.17 2.42 2.73 3.11 3.60 4.25 5.08
6 2.2769 2.35 2.50 2.71 2.99 3.31 3.71 4.19 4.80 5.60 6.61
7 2.9367 3.04 3.24 3.51 3.85 4.25 4.74 5.32 6.06 7.00 8.20
8 3.6287 3.77 4.01 4.35 4.76 5.24 5.81 6.50 7.36 8.45 9.83
9 4.3468 4.53 4.82 5.21 5.70 6.25 6.92 7.71 8.69 9.93 11.49
10 5.0864 5.31 5.65 6.11 6.66 7.29 8.05 8.94 10.05 11.45 13.18
11 5.8443 6.11 6.51 7.02 7.64 8.35 9.20 10.20 11.44 12.98 14.90
12 6.6178 6.92 7.37 7.94 8.65 9.43 10.37 11.48 12.84 14.53 16.63
13 7.405 7.75 8.25 8.89 9.66 10.53 11.56 12.77 14.26 16.11 18.38
14 8.204 8.60 9.15 9.85 10.69 11.64 12.76 14.08 15.70 17.69 20.15
15 9.0137 9.45 10.06 10.82 11.74 12.77 13.98 15.41 17.15 19.29 21.93
16 9.8328 10.32 10.98 11.81 12.79 13.90 15.21 16.74 18.61 20.91 23.73
24 16.636 17.50 18.60 19.94 21.52 23.27 25.33 27.72 30.62 34.13 38.39
32 23.729 24.98 26.52 28.37 30.52 32.93 35.75 39.04 42.99 47.73 53.43
40 30.998 32.62 34.61 36.98 39.74 42.82 46.41 50.59 55.58 61.55 68.69
48 38.387 40.36 42.80 45.71 49.08 52.84 57.20 62.28 68.32 75.52 84.10
56 45.863 48.16 51.05 54.51 58.52 62.96 68.09 74.07 81.16 89.60 99.62

120 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Introducing AMR HR (e.g. 2 TRX Cell)
Phase 0: Existing EFR FR:
• 16 Ch/Cell, 13 Voice Ch/Cell, Max. 7 Erl/Cell*
TSL 0 1 2 3 4 5 6 7
TRX1 S S F F F F F G
TRX2 F F F F F F F F

Phase 1: Adding 1 DR TSL per TRX:

• 18 Ch/Cell, 15 Voice Ch/Cell, Max. 9 Erl/Cell*
TSL 0 1 2 3 4 5 6 7
TRX1 S S D F F F F G
TRX2 D F F F F F F F

Phase 2: All Voice TSL are DR:

• 29 Ch/Cell, 26 Voice Ch/Cell, Max. 18 Erl/Cell*
TSL 0 1 2 3 4 5 6 7
TRX1 S S D D D D D G
TRX2 D D D D D D D D

 Phase 2 essentially DOUBLES the voice capacity compared to phase 0. Signaling load between BSC and MSC has to be considered
as well

* at 2% Blocking rate, using Erlang B considering HR is used without Radio Link Constrains

121 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 System Level Performance of AMR-FR
Simulations

122 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Simulation Scenario
Regular
RegularHexagonal
Hexagonalgrid
gridwith
with3-
3-
sector sites
sector sites

Interference
Limited
Scenario

123 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Simulation parameters
Simulated network structure:
- Number of base stations 25 3-sector BTSs (75 cells)
- Antennas 65 sectorized
- Intersite distance 1.5 km
- Number of TRXs 4-6
- Frequency Reuse 1/1 (FH)
- BCCH TRX Reuse 12 - 9 (not
included in hop.)
- Frequency spectrum 5 MHz (12 hopping
freqs) Mobile velocity 3 km/h
Call mean hold time 120s (exp. distribution)
Type of frequency hopping Random hopping
DTX factor 0.5
Power Control On
AMR codec mode adaptation On
AMR full-rate modes in use: 12.2 kbit/s 7.4
kbit/s
1 Sample = 2 secs (96 Speech
5.9Frames);
kbit/s 4.75
kbit/s
1 Bad Quality Sample(BQS) = Sample with
FER>4.2%
X-axis for hopping plots = EFL = Effective Frequency
Load (how much each frequency is on the air)  Traffic
124 © NOKIA
(E)/(8*#hop.freqs)
Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 CIR vs FER link level in simulator
TU3, no Hop TU3, RH12

100.00% TU 100.000%

3
10.00% 10.000%

FER(%)
FER(%)

1.00% 1.000%

0.10% 0.100%

0.01% 0.010%
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
CIR(dB) CIR(dB)

AFS122, no Hop AFS475, no Hop LA, no Hop AFS122, no Hop AFS475, no Hop LA, no Hop

• Link Level performance in simulator for non-hopping seems to be

quite similar to the one measured in Nokia lab for UL
• Link Level Performance in simulator for hoppig over 12
frequencies seems to be aligned with Nokia lab results for UL
(between performance for TU3 rfh5 and TU50 non-hop)

125 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential

 Performance of AMR-FR
(100% AMR penetration vs 100% EFR)
BCCH layer (no- Hopping layer (12
hop) freqs.) EFR vs AMR
PERFORMANCE OF TRAFFIC CHANNELS OF BCCH
FOR EFR AND AMR-FR
AND BOTH 9-BCCH AND 12-BCCH REUSES
TU 4.5%

3 4.0%

16%
3.5%
14%
% Bad Quality Samples

3.0%
12%

Quality
fER > 4%
10% 2.5%

8%
2.0%
140

% BadTCH
Samples
6%
4% 1.5% %
2% 1.0%
0%
0.5%
1.0 1.5 2.0 2.5 3.0
Erlangs per cell in BCCH 0.0%
5% 10% 15% 20% 25% 30% 35%
AMR traffic; 12 BCCH EFR traffic; 12 BCCH
EFL(%)
AMR traffic; 9 BCCH EFR traffic; 9 BCCH
CodecMA EFR

AMR
AMRwith
withBCCH
BCCHreuse
reuse99performs
performs At
At2%
2%outage
outageof
ofBQS,
BQS,AMRAMR
similar
similar to EFR with BCCHreuse
to EFR with BCCH reuse allows
allows 140% more trafficthan
140% more traffic than
12
12 EFR.
EFR.
126 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 AMR Penetration
EFR vs AMR penetration

4.5%
• Most of the practical cases there
4.0% will be mixed AMR and EFR
3.5% mobiles in the network at the
3.0%
same time
TCH FER > 4%
% Bad Quality

2.5%

2.0% • Therefore, very tight frequency

plan may not be feasible in order
Samples

1.5%

1.0%
to maintain good speech quality
0.5%
with existing EFR calls
0.0%
0% 5% 10% 15% 20% 25% 30% 35%
EFL(%)
AM R capacity gain (%)

0 % AMR / 100% EFR 25% AMR / 75% EFR 63% AMR / 37 % EFR
160%
100 % AMR / 0% EFR Poly. (100 % AMR / 0% EFR)
140%

(%)
• TCH FER decreases considerably 120%

Gain (%)
when AMR penetration increases

CapacityGain
100%

80%
• Increased TCH quality can be
Capacity
60%

turned into capacity by allowing 40%

more traffic to the network 20%

0%
• About 140% traffic increase is 25% A MR 63%A MR 100%A MR

attained with 100% AMR A MR penetration

penetration
127 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Effect of AMR LA at network level
From FER point of view, the However, the gain of AMR LA is
usage of the most robust codec that the Speech Quality (SQ) is
(AFS475) improves the improved as the codec providing
performance of the network vs LA best SQ should be always used

FR AMR Codecs FER performance FR AMR Codecs MOS performance

4.0% 2.5%

FER
3.5%

3.0%
Hopping 2.0% MOS Hoppin
g layer

% ended calls avg estimated MOS<3

layer
2.5%
TCH fER > 4%

1.5%
% Bad Quality

2.0%

1.5% 1.0%
Samples

1.0%
0.5%
0.5%

0.0% 0.0%
0% 5% 10% 15% 20% 25% 0% 5% 10% 15% 20% 25%
EFL(%) EFL(%)

A FS122 A FS740 A FS590 A FS475 CodecMA A FS122 A FS740 A FS590 A FS475 CodecMA

POWER CONTROL OFF IN THESE

128 © NOKIA SIMULATIONS
Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Impact of LA Thresholds
C_th1 C_th2 C_th3 C_th4 C_th5
th1 9 11 13 15 17
th2 5 7 9 11 13
FER CodecMA vs AFS475 th3 2 4 6 8 10 MOS PER FORMAN C E

3.86 8%
0.8%

3.85 7%
0.7%

0.6%
3.84 6%

BQC (MOS <3.5)

0.5% 3.83 5%

AVG. MOS
FER(%)

0.4% 3.82 4%

0.3% 3.81 3%

0.2% CODEC USAGE 3.80 2%

0.1% 100% 3.79 1%

0.0% 90% 3.78 0%

C_Th1 C_Th2 C_Th3 C_Th4 C_Th5 C_Th1 C_Th2 C_Th3 C_Th4 C_Th5
80%
THRESHOLD SET THRESHOLD SET
% SPEECH FRAME USAGE

70%
% FER (CodecMA ) % FER samples > 4.2% Average MOS (1) % ended c alls avg MOS < 3.5
60%

•• The
The higher
higher the the
50%

40%

thresholds
thresholds the lower
the lower 30%

the
the FER,
FER, but but the the 20%

worse
worse the the Speech
Speech 10%

Quality as
Quality as codecs codecs 0%
C_Th1 C_Th2 C_Th3 C_Th4 C_Th5

more
more robust
robust (4.75, (4.75, THRESHOLD SET

5.90)
5.90) are
are used
used more more % A FS122 %A FS740 % A FS590 % A FS475

often.
129 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential
 Codec Usage for different loads
Codec usage in 100% AMR case

100%

90%

80%

70%
% codec usage

60%

50%

40%

30%

20%

10%

0%
21% 32%
EFL(%)

AFS475 AFS590 AFS740 AFS122

• The higher the load the worse radio conditions and therefore
higher usage of more robust codec, impacting in Speech Quality

130 © NOKIA Presentation_Name.PPT / DD-MM-YYYY / Initials Company Confidential