173312178 IRAT Handover Optimization Project

IRAT Handover Optimization Project

Title of presentation (Menu View, Master, Slide Master)

dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

2

Agenda

Background

– P3 IRAT Handover Strategy

– P4 IRAT Handover

P3 IRAT Handover Strategy analysis

Test cases

Test Results

Additional Analysis and the way forward


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

3

-100

-101

Background – P3 IRAT Handover Strategy IRAT based on RSCP only

Perform IRAT as soon as 3G gets below -101 dBm

-98

-97

-80UMTS

Coverage

GSM Coverage

-95

CM

UMTS Satisfied

RS

CP

UMTS and GSM Satisfied

GSM not ready

e2d

e2f

e3a

e3a


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

4

Background – P3 IRAT Handover

IRAT based on RSCP only and Immediate handover

– Almost 30% of all active connections end in 2G

– Total Drop rate = 0.5%

– Total drop rate due to IRAT Handover failure = 0.05%


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

5

Background – P4 IRAT Handover IRAT based on Ec/No or RSCP or UeTx Power (Uplink and Downlink

Monitoring)

downlink

uplink Ue Tx Power

quality = good if all measurement items are

good

quality = bad if at least one measurement is bad

CONNECTION QUALITY MONITORING

CPICH RSCPCPICH Ec/No

Ec/No

bad good

RSCP

bad good

UeTxPwr

badgood

Configurable thresholds for event triggering


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

6

Background – P4 IRAT Handover IRAT based on RSCP or Ec/No or UeTx Power

RSCP: Typically Indoor or coverage border area

Ec/No: Typically Polluted or high traffic area

UeTx Power: Typically high Ul Interference

RSSI (Interference)

RSCP (Signal)Thermal Noise

Better Ec/No but poor RSCP

RSSI (Interferen

ce)

Thermal Noise

RSCP (Signal)Good RSCP

everywhere but bad Ec/No

P4: 384/HSDPA

P5: E-UL

P4: HSDPA -> check maxTxPowerDl and cpich settings


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

7

Check to be performed on all P4 cells

Total TX Power

Common Channel Power

Power available for traffic

Ec/No=CPICH - L - (IN + OUT + TH)

Ec/No = CPICH - L - IN = CPICH - L - (TotTxPower - L) = CPICH - TotTxPower

Admission Margin (R99)

HSDPA Margin

Ec/No = CPICH - TotTxPower + HsdpaMargin

Ec/No = CPICH - TotTxPower + AdmissionMargin

(primaryCpichPower-maximumTransmissionPower+hsPowerMargin)/10

>

usedFreqThresh2dEcNo-hysteresis2d/4


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

8

P3 IRAT Handover Strategy Analysis IRAT based on RSCP only and

“immediate handover” policy

Ec/No power triggering is almost disabled by setting the 2d threshold to the maximum

UeTx power triggering is disabled by setting the relative filter coefficient to its maximum

gsmThresh3a -95hysteresis2d 4hysteresis2f 4hysteresis3a 0usedFreqThresh2dRscp -100usedFreqThresh2dRscpDrnc -100usedFreqRelThresh2fRscp 2utranRelThresh3aRscp 20TimeToTrigger2dRscp 200TimeToTrigger2fRscp 1280TimeToTrigger3a 4 (60)timeTrigg6a 320timeTrigg6b 1280utranFilterCoefficient3 1gsmFilterCoefficient3 1filterCoefficient1 3filterCoefficient2 3filterCoeff6 19ueTxP owerThresh6a 21ueTxP owerThresh6b 18usedFreqThresh2dEcNo -24usedFreqThresh2dEcNoDrnc -24usedFreqRelThresh2fEcNo 1utranRelThresh3aEcNo -1TimeToTrigger2dEcNo 200TimeToTrigger2fEcNo 1280utranRelThreshRscp 5MRSL

Enter Compress Mode For RSCP -101Exit Compress Mode for RSCP -97Handover for RSCP -80

Enter Compress Mode for EcNo -25Exit Compress Mode for EcNo -22Handover for EcNo -25

Enter Compress mode for UeTx 21Exit Compress Mode for UeTx 18Handover for UeTx (RSCP ) -75

P 3/P 4 P olicy


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

9

P3 IRAT Handover Strategy Analysis We split 2d events in 2 classes:

ClassA: 2d events which are followed by either IRAT Handover or event 2f or dropped call

ClassB: 2d events which are not followed by any event or dropped call (most likely 2d events on short RRC connection for registration)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

P3 Policy

2d Event count

ClassA 2d Events (All RABs) ClassB 2d Events (All RABs)


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

10

P3 IRAT Handover Strategy Analysis

0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics

Averall Performance Statistics

TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm

Only almost half of the phones that

trigger 2d actually perform handover

The majority returns in good coverage either before CM

activation or before handover

execution

This graph is only reporting ClassA 2d events


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

11

Averall Time Statistics(all RABs, all triggers)

0

5

10

15

20

25

30

35

40

Time (s)

PD

F (

%)

0

10

20

30

40

50

60

70

80

CD

F (

%)

TimeToCmAct PDF TimeTo3a PDF TimeToHo PDF TimeInCm PDF TimeTo2fCm PDF TimeTo2fNoCm PDF

TimeToCmAct CDF TimeTo3a CDF TimeToHo CDF TimeInCm CDF TimeTo2fCm CDF TimeTo2fNoCm CDF

P3 IRAT Handover Strategy Analysis

In 10% of the cases the 2f event happens before the UE has time to enter CM

In other cases the Ue might be in a connection state where it is not possible to enter CM (Signalling, or high packet rates).

It takes around 2.4 seconds for the UE to enter CM and start measuring GSM

50% of the handover need more than 4 seconds to be executed, but in general it is soon after event 3a


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

12

P3 IRAT Handover Strategy Analysis IRAT handover is occurring too early and/or in areas where the

radio environment is changing fast (2d ending in 2f events)

– Lowering handover threshold and increasing time to trigger for 2d and 2f might help making IRAT handover more efficient


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

13

Test Case description

Different test cases have been defined:– Activate Ec/No based IRAT Handover

– Move RSCP based IRAT Handover to lower coverage areas

– Move Ec/No based IART Handover to more highly interfered areas

– Combine the parameter settings from previous tests cases

– Increase time to trigger (delay decision on when entering compress mode based on both RSCP and Ec/No)

– Activate UeTx power based handover

– Increase time to trigger (delay decision on when entering compress mode based on UeTx Power)

– Combine parameter settings from previous test cases (RSCP, Ec/No, UeTx Power)

– Move handover decision based on RSCP to more extreme coverage areas

– In the attached file all test cases are explainedMicrosoft Excel Worksheet


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

14

The Ec/No trigger is introduced and threshold for CM activation is set to -14 dB.

The Immediate handover strategy is also applied for Ec/No and the 3a threshold is set thigher than the 2d threshold

Ue Tx power triggering is still disabled by mean of high filter coefficient values

Test Case Description – Introducing Ec/No trigger gsmThresh3a -95

hysteresis2d 4hysteresis2f 4hysteresis3a 0usedFreqThresh2dRscp -100usedFreqThresh2dRscpDrnc -100usedFreqRelThresh2fRscp 1utranRelThresh3aRscp 20TimeToTrigger2dRscp 320TimeToTrigger2fRscp 1280TimeToTrigger3a 0timeTrigg6a 320timeTrigg6b 1280utranFilterCoefficient3 2gsmFilterCoefficient3 1filterCoefficient1 3filterCoefficient2 2filterCoeff6 19ueTxP owerThresh6a 21ueTxP owerThresh6b 18usedFreqThresh2dEcNo -13usedFreqThresh2dEcNoDrnc -12usedFreqRelThresh2fEcNo 1utranRelThresh3aEcNo 8TimeToTrigger2dEcNo 200TimeToTrigger2fEcNo 1280utranRelThreshRscp 5MRSL -7




Enabling Ec/No


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

15

Test Case Analysis – Introducing Ec/No trigger

We split 2d events in 2 classes:


ClassB: 2d events which are not followed by any other event (most likely 2d events on RRC connection only)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Enabling Ec/No Trigger

2d Event count



dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

16

0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics




Only almost half of the phones that

trigger 2d actually perform handover

The majority returns in good coverage either before CM

activation or before handover

execution

This graph is only reporting ClassA 2d events


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

17


0

5

10

15

20

25

30

35

40

Time (s)

PD

F (

%)

0

10

20

30

40

50

60

70

80

CD

F (

%)



In other cases the Ue might be in a connection state where it is not possible to enter CM (Signalling, or high packet rates).


It takes around 2.4 seconds for the UE to enter CM and start measuring GSM

In 10% of the cases the 2f event happens before the UE has time to enter CM

50% of the handover need more than 4 seconds to be executed, but in general it is soon after event 3a


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

18

Test Case Analysis – Introducing Ec/No trigger Introducing Ec/No trigger at -14 without changing time to trigger values does

not give any significant change. The reason for this is that most of the 2d events are still being triggered based on RSCP rather than on Ec/No.

0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality

Averall Perf. Statistic(All RBs)


Most of the 2d events are followed by 2f event in the Ec/No case

Most of the 2d events are followed by CM activation in the RSCP case


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

19

RSCP thresholds have been lowered in two steps of 1 dB each.

Same policy is kept for Ec/No

Same Policy is kept for UeTx power

Test Case Description – Lowering RSCP thresholds gsmThresh3a -95 -95

hysteresis2d 4 4hysteresis2f 4 4hysteresis3a 0 0usedFreqThresh2dRscp -101 -102usedFreqThresh2dRscpDrnc -100 -100usedFreqRelThresh2fRscp 1 1utranRelThresh3aRscp 20 20TimeToTrigger2dRscp 320 320TimeToTrigger2fRscp 1280 1280TimeToTrigger3a 0 0timeTrigg6a 320 320timeTrigg6b 1280 1280utranFilterCoefficient3 2 2gsmFilterCoefficient3 1 1filterCoefficient1 3 3filterCoefficient2 2 2filterCoeff6 19 19ueTxP owerThresh6a 21 21ueTxP owerThresh6b 18 18usedFreqThresh2dEcNo -13 -13usedFreqThresh2dEcNoDrnc -12 -12usedFreqRelThresh2fEcNo 1 1utranRelThresh3aEcNo 8 8TimeToTrigger2dEcNo 200 200TimeToTrigger2fEcNo 1280 1280utranRelThreshRscp 5 5MRSL -7 -7

Enter Compress Mode For RSCP -102 -103Exit Compress Mode for RSCP -99 -100Handover for RSCP -81 -82

Enter Compress Mode for EcNo -14 -14Exit Compress Mode for EcNo -11 -11Handover for EcNo -5 -5

Enter Compress mode for UeTx 21 21Exit Compress Mode for UeTx 18 18Handover for UeTx (RSCP ) -76 -77

Reducing RSCP threshold

Reducing RSCP threshold


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

20

Test Case Description – Lowering RSCP thresholds

We split 2d events in 2 classes:


ClassB: 2d events which are not followed by any other event (most likely 2d events on RRC connection only)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Lowering RSCP Thresholds 1

2d Event count


0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Lowering RSCP Thresholds 2

2d Event count


2d event at -102 dBm



dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

21

Test Case Description – Lowering RSCP thresholdsThese graphs are only

reporting ClassA 2d events

0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics



0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics



0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics






No majour difference between the three test cases

No additional dropped calls

Still the handover is happening either in fast changing environment


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

22

Test Case Description – Lowering RSCP thresholds

0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality






Reducing the RSCP triggering does not significantly affect the unbalance between the amount of handover started due to RSCP and Ec/No reasons.

Ec/No triggered handover attempts are mostly generated in fast changing environment

RSCP triggered attempts are mostly followed by handover -> Indoor users? Coverage border users?


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

23

RSCP threshold for event 2d has been moved to -105 dBm.

Ec/No Threshold is -13 dB

UeTx Power trigger is enabled

Time to trigger for the 2d and 6a events have been increased (640ms) in order to remove false bad radio environment detection

Time to trigger for 2f and 6b events have been increased (2560) in order to give time to CM activation

Test Case Description – Delaying TTT and reducing RSCP threshold

gsmThresh3a -95hysteresis2d 4hysteresis2f 4hysteresis3a 0usedFreqThresh2dRscp -104usedFreqThresh2dRscpDrnc -100usedFreqRelThresh2fRscp 1utranRelThresh3aRscp 20TimeToTrigger2dRscp 640TimeToTrigger2fRscp 2560TimeToTrigger3a 0timeTrigg6a 640timeTrigg6b 2560utranFilterCoefficient3 2gsmFilterCoefficient3 1filterCoefficient1 3filterCoefficient2 3filterCoeff6 3ueTxP owerThresh6a 21ueTxP owerThresh6b 18usedFreqThresh2dEcNo -12usedFreqThresh2dEcNoDrnc -12usedFreqRelThresh2fEcNo 1utranRelThresh3aEcNo 8TimeToTrigger2dEcNo 640TimeToTrigger2fEcNo 2560utranRelThreshRscp 20MRSL -6




Extreeme RSCP (-> keep good Ec/No) and UeTx P ower


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

24


0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics



This graph is only reporting ClassA 2d events 2d event at -101 dBm

TTT2d=320ms; TTT2f=1280

0

10

20

30

40

50

60

70

80

90

100

%

Averall Statistics



2d event at -105 dBmTTT2d=640ms;TTT2f=2

560

The longer TTT is reducing the amount of false 2d event detections: less useless signalling and higher IRAT efficiency


0

5

10

15

20

25

30

35

40

Time (s)

PD

F (

%)

0

10

20

30

40

50

60

70

80

CD

F (

%)




0

5

10

15

20

25

30

35

40

Time (s)

PD

F (

%)

0

10

20

30

40

50

60

70

80

CD

F (

%)



In all these cases the Ue is on a signalling connection or packet with higher bitrate


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

25


0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



Even with a 2d threshold for RSCP at -105 most of the IRAT Handover attempts are based on poor RSCP rather then poor Ec/No or high UeTx Power.


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

26


Most of the 2d events are triggered while on Speech connection.

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

%

TrigGsmMeasStart

RB distribution when Start Measuring 2G

Other Speech PS64/64 PS64/128 PS64/384


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

27


0

10

20

30

40

50

60

70

80

90

100

%

EcNo RSCP UeTxPower

Triggering Quality

Averall Perf. Statistic(Speech)

TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCm TrigGsmMeasStopNoCm

0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality

Averall Perf. Statistic(PS64)


0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



Almost all Speech connection perform handover, some other exit CM due to 2f event triggering or other reasons

When 2d is triggered on packet 64 most of the time the handover is performed, in the remaining cases CM might be disactivated due to upswitch to higher bitrates (with 2d still being fulfilled)

Packet is triggering handover to GSM based on Ec/No more than speech (where packet is present there is more interference maybe also due to HSDPA)

When the UE is on higher bitrates the event 2d is often followed by a 2f event.

In other cases the UE has still time to switch down to 64 and perform handover or switch up again

The connection almost never drops.

Even at -105dBm There is still room for reduction!!!


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

28


Distribution of the time between Rab Establishment and IRAT Handover

0

5

10

15

20

25

30

35

40

45

50

55

60

0 0.4

0.8

1.2

1.6

2 2.4

2.8

3.2

3.6

4 4.4

4.8

5.2

5.6

6 6.4

6.8

7.2

7.6

8 8.4

8.8

9.2

9.6

10 10.4

0

1

2

3

4

5

6

7

8

9

10

11

12

TimeRabEstToHo PDF TimeRabEstToHo CDF

50% of the handover happen within 9 seconds from the rab establishment.

Considering that it takes around 4 seconds to perform handover from the event 2d, it means that the event is usually triggered 5 seconds after the call setup


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

29

Full Test Case comparison

Handover Rate and Drop rate

0

0.25

0.5

0.75

1

1.25

1.5

1.75

2

2.25

2.5

SummaryOption01

SummaryOption02

SummaryOption03

SummaryOption04

SummaryOption06

SummaryOption07

SummaryOption08

SummaryOption09

SummaryOption10

Test Case

Dro

p R

ate

(%)

0

5

10

15

20

25

30

35

40

45

50

Han

do

ver

rate

(%

)

Sum of Drop Rate due to Speech IRAT Sum of Total Drop Rate due IRATSum of Speech Drop Rate Sum of Total Drop RateSum of Ho Perc Speech1 Sum of Ho Perc Tot1


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

30

RSCP threshold for event 2d has been moved to -107 dBm.

Ec/No Threshold is -13 dB

UeTx Power trigger is enabled

Time to trigger for the 2d and 6a events have been kept longer (640ms) in order to remove false bad radio environment detection

Time to trigger for 2f and 6b events have been kept longer (2560) in order to give time to CM activation

Final Test Case Description –reducing RSCP threshold further

gsmThresh3a -95hysteresis2d 4hysteresis2f 4hysteresis3a 0usedFreqThresh2dRscp -107usedFreqThresh2dRscpDrnc -100usedFreqRelThresh2fRscp 0utranRelThresh3aRscp 20TimeToTrigger2dRscp 640TimeToTrigger2fRscp 2560TimeToTrigger3a 0timeTrigg6a 640timeTrigg6b 2560utranFilterCoefficient3 2gsmFilterCoefficient3 1gsmThresh3a -95filterCoefficient1 3filterCoefficient2 3filterCoeff6 3ueTxP owerThresh6a 21ueTxP owerThresh6b 18usedFreqThresh2dEcNo -13usedFreqThresh2dEcNoDrnc -13usedFreqRelThresh2fEcNo 0utranRelThresh3aEcNo 8TimeToTrigger2dEcNo 640TimeToTrigger2fEcNo 2560utranRelThreshRscp 20





dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

31

Final Test Case Results –reducing RSCP threshold further

Test Case results

-1.00%

-0.80%

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

Date

Sp

eech

Dro

p R

ate

(%)

0.00

0.50

1.00

1.50

2.00

2.50

3.00

Erl

ang

(%

)

RNC01 Erlang RNC10 Erlang RNC11 Erlang RNC12 Erlang RNC16 Erlang RNC19 Erlang

RNC01 Drop RNC10 Drop RNC11 Drop RNC12 Drop RNC16 Drop RNC19 Drop

Small increase in drop rate (0.15%)

Significative increase in traffic (45%)


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

32

Test Case results

-1.00%

-0.80%

-0.60%

-0.40%

-0.20%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

Date

Sp

eech

Dro

p R

ate

(%)

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

2.6

2.8

MH

T (

Min

.)

RNC01 Drop RNC10 Drop RNC11 Drop RNC12 Drop RNC16 Drop RNC19 Drop

RNC01 MHT RNC10 MHT RNC11 MHT RNC12 MHT RNC16 MHT RNC19 MHT


Users stay connected longer in 3G than before (MHT = 15% longer). Notice this is the only RNC where a significative change in MHT can be observed


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

33

Final Test Case Results –reducing RSCP threshold further Percentage of Speech IRAT

0%

5%

10%

15%

20%

25%

30%

35%

40%

Date

IRA

T (

%)

RNC01 RNC10 RNC11 RNC12 RNC16 RNC19

30% less users move to 2G, again this is the only RNC where a visible change in IRAT rate can be observed


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

34


0

5

10

15

20

25

30

35

40

45

50

55

60

0 0.4

0.8

1.2

1.6

2 2.4

2.8

3.2

3.6

4 4.4

4.8

5.2

5.6

6 6.4

6.8

7.2

7.6

8 8.4

8.8

9.2

9.6

10 10.40

1

2

3

4

5

6

7

8

9

10

11

12


50% of the handover happened within 9 seconds from the rab establishment.

0

5

10

15

20

25

30

35

40

45

50

55

60

0

1

2

3

4

5

6

7

8

9

10

11

12


50% of the handover now happens within 14 seconds from the rab establishment.


Move to 2G at -105 RSCP

-13 Ec/No

Move to 2G at -108 dBm

-14 Ec/No


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

35

0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



Even with a 2d threshold for RSCP at -107 most of the IRAT Handover attempts are based on poor RSCP rather then poor Ec/No or high UeTx Power

Tipical outdoor/indoor and idle/connected coverage


0

10

20

30

40

50

60

70

80

90

%

EcNo RSCP UeTxPower

Triggering Quality



Move to 2G at -105 RSCP

-13 Ec/No

Move to 2G at -108 dBm

-14 Ec/No


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

36

Idle-Mode/Connected Mode IRAT Strategy (3G->2G)

Start measuring GSM in Idle

RSCP>qRxLevMin+P (-115/-112)

Ec/No>qQualMin (-18)

Ec/No>qQualMin+sRatSearch (-14)

UM

TS

C

ell

GSM Coverage

Stab

le U

MTS

Idle

Cove

rage

RSCP>usedFreqThresh2dRscp – hysteresis2d/2 (-101)

RSCP>usedFreqThresh2dEcNo – hysteresis2d/2 (-13)

Small gap in Ec/No strategy between Idle and Connected mode

High gap in RSCP strategy between Idle and Connected mode

Start measuring GSM in Connected mode

Stable UMTS

Coverage in

Connected

mode


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

37


Idle-Mode/Connected Mode IRAT Strategy Proposal (3G->2G)

Start measuring GSM in Idle




UM

TS

C

ell

GSM Coverage

Stab

le U

MTS

Idle

Cove

rage

RSCP>qRxLevMin+sHcsRat (-111/-108)

RSCP>usedFreqThresh2dRscp – hysteresis2d/2 (-108)

RSCP>usedFreqThresh2dEcNo – hysteresis2d/2 (-13)

Small gap in both Ec/No and RSCP strategies between Idle and Connected mode

Start measuring GSM in Connected mode

Stable UMTS

Coverage in

Connected

mode


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

38

qHyst1=qHyst2=4dB and treSelection=2s

– This settings increases the probability that late reselections trigger un-necessary measurements on 2G

qHyst1=4dB and qHyst2=2dB will reduce the risk keeping reselection thresholds between 2G and 3G unchanged

Idle mode Parameter Investigation

qHyst2=4dB

2 s

Measure GSM

During this time the Ue might reselect GSM even with Good RSCP coverage

Ec/No

UMTS CellA

UMTS CellB


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

39

2G/3G Coverage values

Area TypeGSM 900

[dBm]GSM 1800

[dBm]EDGE[dBm]

UMTS[dBm]

3G - 2G [dB]

Macro sites for In-train coverage -66.3 -64.3 -65 -72.4 7.4Dense Urban Macro for Indoor coverage -69.3 -67.3 -68 -81.4 13.4Urban Macro for Indoor coverage -73.3 -71.3 -72 -85.4 13.4Suburban Macro for Indoor coverage -73.3 -71.3 -72 -85.4 13.4Rural Macro for Indoor coverage -77.3 -75.3 -76 -89.4 13.4Macro for In-car coverage -80.3 -78.3 -79 -93.4 14.4Macro for Outdoor Coverage only -87.3 -89.3 -88 -102.4 14.4Indoor site -85 -85 -85 -90.0 5.0

To avoid useless re-selection to 2G we should move to 2G only when the 2G signal is 14 dB better than the 3G signal


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

40

A first attempt has been made to reduce the probability for 3G connections to make false reselections to 2G:

Reducing hysteresis for 3G->3G cell reselection (to avoid false GSM reselections)

Increasing the offset towards 2G neighbour cells to take into account the different planning values for coverage

Delaying measurements on 2G (sRatSearch reduction)

Idle-Mode camping Strategy Step1 (3G->2G)

Idle Mode U->G

qHyst1 4

qHyst2 2

qOffset1sn 15

qOffset2sn 0

qRxLevMin -115

qQualMin -18

sRatSearch 2

sHcsRat 4

measQuantity1 2

maxTxP owerUl 24

Main KPI to monitor performances:

GtoUReselPerTraffic[%]=100*

GtoUReselPerRrc[%]=100*

pmTotNoRrcConnectAttIratCellResel+pmTotNoRrcConnectAttIratCcOrderpmTotNoRrcConnectReqCs+pmTotNoRrcConnectReqPs+pmTotNoRrcCo

nnectReqSmspmTotNoRrcConnectAttIratCellResel+pmTotNoRrcConnectAt

tIratCcOrderpmTotNoRrcConnectReq


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

41

Idle-Mode camping Strategy Step1 (3G->2G) - Results

Test Case results

-40%

0%

40%

80%

120%

160%

200%

Date

Res

el/T

raff

ic (

%)

25%

35%

45%

55%

65%

75%

85%

Res

el/R

RC

(%

)

RNC01 Resel/Traff ic RNC10 Resel/Traff ic RNC11 Resel/Traff ic RNC12 Resel/Traff ic RNC16 Resel/Traff ic RNC19 Resel/Traff ic

RNC01 Resel/RRC RNC10 Resel/RRC RNC11 Resel/RRC RNC12 Resel/RRC RNC16 Resel/RRC RNC19 Resel/RRC

First reduction in IRAT Reselection

Second possible reduction: further monitoring needed

Connected-Mode/Idle-Mode alignment (Connected-Mode parameter change)

New Idle Mode Settings


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

42

Possible problems to be solved

Indoor/Outdoor coverage mismatch

3G 3G

3G3G

2G 2G2G2G


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

43

Possible problems to be solved

Coverage continuity in high traffic environment

3G

3G

2G


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

44

Idle-Mode camping Strategy (3G->2G/2G->3G)

RSCP+qHyst>GSM_RSSI-qOffset1sn

Start measuring GSM




UM

TS

C

ell

GSM Coverage

Stab

le U

MTS

Idle

Cove

rage

Transition to GSM

Transition to UMTS

Ec/No>FDDMIN (-10)

Ping-Pong Areas

Users in Ping-Pong areas will move from GSM to UMTS and back in GSM. They should not cause signalling because they are not allowed to camp on cells with that poor RSCP

The real Ping-Pong is due to the absence of hysteresis, users at the border of the Ping-Pong areas might continuously reselect the two networks where signalling is generated at every reselection


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

45

Idle-Mode camping Strategy (3G->2G/2G->3G)

Ping-Pong areas might very well be located at the edge of UMTS coverage

RSCP = qRxLevMin+P

Ec/No = FDDMIN

Ping-Pong area

GSM Coverage

Extended UMTS Coverage (Ec/No)

Release 5 phones should support RSCPMIN parameter, they will reselect UMTS only if Ec/No>FDDMIN and RSCP>RSCPMIN. Putting RSCPMIN > qRxLevMin+P will remove ping pong areas, but phones seem not to support this parameter yet.


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

46

Changing Neighbouring Strategy will create hysteresis

RSCP = qRxLevMin+P

Ec/No = FDDMIN

Ping-Pong area

GSM Coverage


Idle-Mode camping Strategy Proposal (3G->2G/2G->3G)

Hysteresis

To prevent un-necessary 3G-2G reselection and handover the neighbour relation between the 3G cell and the cosited GSM 1800 might be removed


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

47

Still a cosited GSM 900 Could be a reason for Ping-Pong Areas

RSCP = qRxLevMin+P

Ec/No = FDDMIN

Ping-Pong areaGSM 900



GSM 1800

Start Measuring

UMTS

Starting UMTS measurements only when GSM is good will create hysteresis and prevent Ping-Pong

Hysteresis

QSI could also be used for UMTS only sites where it is not possible to remove the 2G->3G neighbour relation


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

48

Summary:

– Border and Core UMTS Sites:

– Keep all 3G->2G relations but remove the co-sited 3G->2G relation to avoid useless reselections or handover

– Do not remove 3G->2G co-sited relations where 3G is outdoor only and 2G is indoor only or vice versa or where the 2G is 900!

– Remove all un-necessary 2G->3G relations: keep only the 2G->3G relation with the co-sited if any or all 2G->3G relations with close neighbouring sites (face-to-face) if no co-sited is available

– For the co-sited GSM activate the QSI parameter to enable Hysteresis

– Use qOffset1sn=15 dB for all 2G neighbours

– Use qHyst1=2 dB instead of 4 dB (reduces the risk that late reselections trigger measurements on 2G)

– QSI Activation Strategy:

– If 2G Indoor and 3G Outdoor or vice-versa: QSI=3 (GSM<-86 to measure UMTS)

– If both Indoor or both Outdoor: QSI=10 (GSM>-70 to measure UMTS)

Idle-Mode camping Strategy Proposal (3G->2G/2G->3G) Option1


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

49

Summary:

– Border UMTS Sites (as in Option1):

– Keep all 3G->2G relations but remove the co-sited 3G->2G relation to avoid useless reselections or handover



– For the remaining co-sited GSM activate the QSI parameter to enable Hysteresis

– Core UMTS Sites (different from Option1):

– Remove all 3G->2G relations



– If the co-sited GSM is a 900 activate the QSI parameter to enable Hysteresis

– Use qOffset1sn=15 dB for all 2G neighbours (as in Option1)

– Use qHyst1=2 dB instead of 4 dB (as in Option1)

– QSI Activation Strategy (as in Option1):

– If 2G Indoor and 3G Outdoor or vice-versa: QSI=3 (GSM<-86 to measure UMTS)

– If both Indoor or both Outdoor: QSI=10 (GSM>-70 to measure UMTS)

Idle-Mode camping Strategy Proposal (3G->2G/2G->3G) Option2


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

50

In addition to each of the two previous option the parameter SHcsRat should be used to start measuring GSM also when RSCP get closer to the limit (sHcsRat=4 as a starting point)


RSCP+qHyst>GSM_RSSI-qOffset1

Start measuring GSM




UM

TS

C

ell

GSM Coverage

Stab

le U

MTS

Idle

Cove

rage

Transition to GSM

RSCP>qRxLevMin+sHcsRat (-111/-108)


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

51

The proposed parameter settings is an attempt to create the required hysteresis in order to reduce the ping-pong between 2G and 3G cells:

trigger start of UMTS measurements based on GSM coverage (QSI)

Activate the RSCP criteria to come back on UMTS (R12 only phones)

Review all neighbour strategies according to the new recommendations

Idle-Mode camping Strategy Step2 (2G->3G)

Idle Mode G->U

QSI 10

FDDQMIN 3

FDDQMINOFF 0

FDDRSCPMIN 7

FDDQOFF 0

COEXUMTS 1

Measure UMTS when GSM is > -70

move to UMTS if Ec/No is > -8

move to umts if RSCP is > -100

Move to UMTS when RSCP is greater than GSM + -inf

Allow reselection and HO to UMTS on

Maxi

mum

exp

ected

impro

vement o

ver

signalli

ng

reduct

ion


dd

MM

MM

yyyy, S

peaker,

Path

/File

nam

e, C

lass

ifica

tion (

Menu V

iew

, H

ead

er

and

Foote

r)

52

The proposed parameter settings is an attempt to activate 2G->3G handover with minimum risk of ping-pong handover:

trigger start of UMTS measurements based on GSM coverage (QSI)

Activate the RSCP criteria to come back on UMTS (R12 only phones)

It might be recommendable to start with higher values of ISHOLEV to avoid abrupt changes.

2G->3G Handover Activation Strategy

Connected Mode G->UQSC 13ISHOLEV 0MRSL 31COEXUMTS 1

Measure UMTS when GSM is > -58Move to UMTS if GSM load is higher than 0Move to UMTS if EcNo is better than -9Allow reselection and HO to UMTS on

Param

ter s

ettings m

ight

change d

epending o

n

resu

lts fr

om th

e

previo

us test

173312178 IRAT Handover Optimization Project

Documents

classification menu

slide masterdd mmmm

readye2de2fe3ae3add

cpich settingsdd mmmm

event triggeringdd mmmm

way forwarddd mmmm yyyy

irat handover failure

rscp onlyperform irat