Está en la página 1de 52

IRAT Handover Optimization Project

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

Agenda
Background P4 IRAT Handover P3 IRAT Handover Strategy analysis Test cases Test Results Additional Analysis and the way forward
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

P3 IRAT Handover Strategy

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

Background P3 IRAT Handover Strategy


IRAT based on RSCP only Perform IRAT as soon as 3G gets below -101 dBm
-80 -95 -97
-98
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

GSM Coverage
e3a

UMTS Coverage

UMTS and GSM Satisfied UMTS Satisfied GSM not ready

e3a e2f

-100

-101 RSCP

e2d

CM

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

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 MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Background P4 IRAT Handover


IRAT based on Ec/No or RSCP or UeTx Power (Uplink and Downlink Monitoring)

uplink

Ue Tx Power
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

downlink

CPICH RSCP CPICH Ec/No


quality = good if all measurement items are good quality = bad if at least one measurement is bad

CONNECTION QUALITY MONITORING

Configurable thresholds for event triggering

Ec/No

RSCP

UeTxPwr

bad

good

bad

good

good

bad

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

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
RSSI (Interference )
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

RSSI (Interference ) RSCP (Signal ) Thermal


Noise

Better Ec/No but poor RSCP

Good RSCP everywhere Thermal but bad Ec/No

RSCP (Signal)

Noise

P4: HSDPA -> check maxTxPowerDl and cpich settings

UeTx Power: Typically high Ul Interference

P4: 384/HSDPA P5: E-UL

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

Check to be performed on all P4 cells


HSDPA Margin Admission Margin (R99)

Ec/No=CPICH - L - (IN + OUT + TH) Ec/No = CPICH - L - IN = CPICH - L - (TotTxPower - L) = CPICH TotTxPower Ec/No = CPICH - TotTxPower + HsdpaMargin Ec/No = CPICH - TotTxPower + AdmissionMargin (primaryCpichPower-maximumTransmissionPower+hsPowerMargin)/10 >
Total TX Power

Power available for traffic

usedFreqThresh2dEcNo-hysteresis2d/4

Common Channel Power

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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 hysteresis2d hysteresis2f hysteresis3a usedFreqThresh2dRscp usedFreqThresh2dRscpDrnc usedFreqRelThresh2fRscp utranRelThresh3aRscp TimeTo Trigger2dRscp TimeTo Trigger2fRscp TimeTo Trigger3a timeTrigg6a timeTrigg6b utranFilterCo efficient3 gsmFilterCo efficient3 filterCo efficient1 filterCo efficient2 filterCo eff6 ueTxP o werThresh6a ueTxP o werThresh6b usedFreqThresh2dEcNo usedFreqThresh2dEcNo Drnc usedFreqRelThresh2fEcNo utranRelThresh3aEcNo TimeTo Trigger2dEcNo TimeTo Trigger2fEcNo utranRelThreshRscp M RSL Enter Co mpress M o de Fo r RSCP Exit Co mpress M o de fo r RSCP Hando ver fo r RSCP Enter Co mpress M o de fo r EcNo Exit Co mpress M o de fo r EcNo Hando ver fo r EcNo Enter Co mpress mo de fo r UeTx Exit Co mpress M o de fo r UeTx Hando ver fo r UeTx (RSCP ) -95 4 4 0 -1 00 -1 00 2 20 200 1 280 4 (60) 320 1 280 1 1 3 3 1 9 21 1 8 -24 -24 1 -1 200 1 280 5

-1 01 -97 -80 -25 -22 -25 21 1 8 -75

P 3/P 4 P o licy

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

P3 IRAT Handover Strategy Analysis


We split 2d events in 2 classes:
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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)
2d Event count

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% P3 Policy ClassA 2d Events (All RABs) ClassB 2d Events (All RABs)

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

P3 IRAT Handover Strategy Analysis


This graph is only reporting ClassA 2d events
Averall Performance Statistics

100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM

The majority returns in good coverage either before CM activation or before handover execution

TrigGsmMeasStopNoCm

10

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Only almost half of the phones that trigger 2d actually perform handover

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

P3 IRAT Handover Strategy Analysis

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

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

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

11

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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)
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

12

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

Test Case description


Different test cases have been defined:
Activate Ec/No based IRAT Handover 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)
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Move RSCP based IRAT Handover to lower coverage areas

Move handover decision based on RSCP to more extreme coverage areas


In the attached file all test cases are explained

13

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

Test Case Description Introducing Ec/No trigger


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

Enter Co mpress M o de Fo r RSCP Exit Co mpress M o de fo r RSCP Hando ver fo r RSCP Enter Co mpress M o de fo r EcNo Exit Co mpress M o de fo r EcNo Hando ver fo r EcNo Enter Co mpress mo de fo r UeTx Exit Co mpress M o de fo r UeTx Hando ver fo r UeTx (RSCP )

-1 01 -98 -80 -1 4 -1 1 -5 21 1 8 -75

Enabling Ec/No

14

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

gsmThresh3a hysteresis2d hysteresis2f hysteresis3a usedFreqThresh2dRscp usedFreqThresh2dRscpDrnc usedFreqRelThresh2fRscp utranRelThresh3aRscp TimeTo Trigger2dRscp TimeTo Trigger2fRscp TimeTo Trigger3a timeTrigg6a timeTrigg6b utranFilterCo efficient3 gsmFilterCo efficient3 filterCo efficient1 filterCo efficient2 filterCo eff6 ueTxP o werThresh6a ueTxP o werThresh6b usedFreqThresh2dEcNo usedFreqThresh2dEcNo Drnc usedFreqRelThresh2fEcNo utranRelThresh3aEcNo TimeTo Trigger2dEcNo TimeTo Trigger2fEcNo utranRelThreshRscp M RSL

-95 4 4 0 -1 00 -1 00 1 20 320 1 280 0 320 1 280 2 1 3 2 1 9 21 1 8 -1 3 -1 2 1 8 200 1 280 5 -7

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

Test Case Analysis Introducing Ec/No trigger


We split 2d events in 2 classes:
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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 other event (most likely 2d events on RRC connection only)
2d Event count

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Enabling Ec/No Trigger ClassA 2d Events (All RABs) ClassB 2d Events (All RABs)

15

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

Test Case Analysis Introducing Ec/No trigger This graph is only reporting
ClassA 2d events
Averall Performance Statistics

100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM

The majority returns in good coverage either before CM activation or before handover execution

TrigGsmMeasStopNoCm

16

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Only almost half of the phones that trigger 2d actually perform handover

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

Test Case Analysis Introducing Ec/No trigger

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

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

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

17

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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.
Averall Perf. Statistic (All RBs)
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP

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

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

UeTxPower

Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm

18

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

Test Case Description Lowering RSCP thresholds


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

Enter Co mpress M o de Fo r RSCP Exit Co mpress M o de fo r RSCP Hando ver fo r RSCP Enter Co mpress M o de fo r EcNo Exit Co mpress M o de fo r EcNo Hando ver fo r EcNo Enter Co mpress mo de fo r UeTx Exit Co mpress M o de fo r UeTx Hando ver fo r UeTx (RSCP )

-1 02 -99 -81 -1 4 -1 1 -5 21 1 8 -76

-1 03 -1 00 -82 -1 4 -1 1 -5 21 1 8 -77

Reducing RSCP thresho ld

Reducing RSCP thresho ld

19

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

gsmThresh3a hysteresis2d hysteresis2f hysteresis3a usedFreqThresh2dRscp usedFreqThresh2dRscpDrnc usedFreqRelThresh2fRscp utranRelThresh3aRscp TimeTo Trigger2dRscp TimeTo Trigger2fRscp TimeTo Trigger3a timeTrigg6a timeTrigg6b utranFilterCo efficient3 gsmFilterCo efficient3 filterCo efficient1 filterCo efficient2 filterCo eff6 ueTxP o werThresh6a ueTxP o werThresh6b usedFreqThresh2dEcNo usedFreqThresh2dEcNo Drnc usedFreqRelThresh2fEcNo utranRelThresh3aEcNo TimeTo Trigger2dEcNo TimeTo Trigger2fEcNo utranRelThreshRscp M RSL

-95 4 4 0 -1 01 -1 00 1 20 320 1 280 0 320 1 280 2 1 3 2 1 9 21 1 8 -1 3 -1 2 1 8 200 1 280 5 -7

-95 4 4 0 -1 02 -1 00 1 20 320 1 280 0 320 1 280 2 1 3 2 1 9 21 1 8 -1 3 -1 2 1 8 200 1 280 5 -7

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

Test Case Description Lowering RSCP thresholds


We split 2d events in 2 classes:
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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 other event (most likely 2d events on RRC connection only)
2d Event count 2d Event count

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Lowering RSCP Thresholds 1 ClassA 2d Events (All RABs) ClassB 2d Events (All RABs)

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Lowering RSCP Thresholds 2 ClassA 2d Events (All RABs) ClassB 2d Events (All RABs)

2d event at -102 dBm

2d event at -103 dBm

20

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

Test Case Description Lowering RSCP These graphs are only thresholds
reporting ClassA 2d events
Averall Performance Statistics

No majour difference between the three test cases

2d event at -101 dBm


100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm

No additional dropped calls Still the handover is happening either in fast changing environment
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Averall Performance Statistics

Averall Performance Statistics

2d event at -102 dBm


100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm TrigGsmMeasStart CmAct Trig3a PerfHo 100 90 80 70 60 % 50 40 30 20 10 0

2d event at -103 dBm

Averall Statistics Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm

21

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

Test Case Description Lowering RSCP thresholds Reducing the RSCP triggering
Averall Perf. Statistic (All RBs)
90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm UeTxPower

2d event at -101 dBm

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?
Averall Perf. Statistic (All RBs)
90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm TrigGsmMeasStart CmAct Trig3a PerfHo UeTxPower 40 30 20 10 0 EcNo RSCP Triggering Quality Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm UeTxPower

Averall Perf. Statistic (All RBs)

90 80 70 60 50 %

2d event at -102 dBm

2d event at -103 dBm

22

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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

Test Case Description Delaying TTT and reducing RSCP threshold

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

Enter Co mpress M o de Fo r RSCP Exit Co mpress M o de fo r RSCP Hando ver fo r RSCP Enter Co mpress M o de fo r EcNo Exit Co mpress M o de fo r EcNo Hando ver fo r EcNo Enter Co mpress mo de fo r UeTx Exit Co mpress M o de fo r UeTx Hando ver fo r UeTx (RSCP )

-1 05 -1 02 -84 -1 3 -1 0 -4 21 1 8 -64

Extreeme RSCP (-> keep go o d Ec/No ) and UeTx P o wer

23

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

gsmThresh3a hysteresis2d hysteresis2f hysteresis3a usedFreqThresh2dRscp usedFreqThresh2dRscpDrnc usedFreqRelThresh2fRscp utranRelThresh3aRscp TimeTo Trigger2dRscp TimeTo Trigger2fRscp TimeTo Trigger3a timeTrigg6a timeTrigg6b utranFilterCo efficient3 gsmFilterCo efficient3 filterCo efficient1 filterCo efficient2 filterCo eff6 ueTxP o werThresh6a ueTxP o werThresh6b usedFreqThresh2dEcNo usedFreqThresh2dEcNo Drnc usedFreqRelThresh2fEcNo utranRelThresh3aEcNo TimeTo Trigger2dEcNo TimeTo Trigger2fEcNo utranRelThreshRscp M RSL

-95 4 4 0 -1 04 -1 00 1 20 640 2560 0 640 2560 2 1 3 3 3 21 1 8 -1 2 -1 2 1 8 640 2560 20 -6

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

Test Case Description Delaying TTT and reducing RSCP threshold The longer TTT is reducing the
This graph is only reporting ClassA 2d events
Averall Performance Statistics
100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm TrigGsmMeasStart CmAct Trig3a PerfHo

amount of false 2d event detections: less useless signalling Averall Performance Statistics and higher IRAT efficiency 2d event at -101 dBm 2d event at -105 dBm TTT2d=320ms; TTT2f=1280 TTT2d=640ms;TTT2f=2560
100 90 80 70 60 % 50 40 30 20 10 0 Averall Statistics Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm

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

24

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Test Case Description Delaying TTT and reducing RSCP threshold


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.
Averall Perf. Statistic (All RBs)
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm UeTxPower

25

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

Test Case Description Delaying TTT and reducing RSCP threshold


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

RB distribution when Start Measuring 2G

70.0

60.0

50.0

40.0 % 30.0

20.0

10.0

0.0 TrigGsmMeasStart Other Speech PS64/64 PS64/128 PS64/384

26

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Test Case Description Delaying TTT and reducing RSCP threshold


Almost all Speech connection perform handover, some other exit CM due to 2f event triggering or other reasons
Averall Perf. Statistic (Speech)

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)
90 80 70 60

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)
Averall Perf. Statistic (PS64)

100 90 80 70 60

50 % 50 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCm TrigGsmMeasStopNoCm TrigGsmMeasStart CmAct Trig3a PerfHo UeTxPower 30 20 10 0 EcNo RSCP Triggering Quality Dropped after 2d CmDeAct TrigGsmMeasStopCm TrigGsmMeasStopNoCm UeTxPower % 40

Averall Perf. Statistic (PS128)

90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct

Averall Perf. Statistic When the UE is on higher (PS384) bitrates the event 2d is The connection almost never drops. often followed by a 2f 90 event. 80

In other cases the UE has 60 still time to switch 50 down to 64 and perform % handover or switch up 40 again 30
20 10 0 UeTxPower EcNo RSCP Triggering Quality TrigGsmMeasStopCm TrigGsmMeasStopNoCm TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCm TrigGsmMeasStopNoCm UeTxPower

70

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

27

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Test Case Description Delaying TTT and reducing RSCP threshold


Distribution of the time between Rab Establishment and IRAT Handover

60 55 50 45 40 35 30 25 20 15 10 5 0

12 11 10

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

9 8 7

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

6 5 4 3 2 1 0

TimeRabEstToHo PDF

TimeRabEstToHo CDF

28

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

0.4

0.8

1.2

1.6

2.4

2.8

3.2

3.6

4.4

4.8

5.2

5.6

6.4

6.8

7.2

7.6

8.4

8.8

9.2

9.6

10

10.4

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

Full Test Case comparison


Handover Rate and Drop rate

2.5 2.25 2 1.75

50 45 40 35 30 25 20 15 10 5 0 Summary Option01 Summary Option02 Summary Option03 Summary Option04 Summary Option06 Test Case Sum of Drop Rate due to Speech IRAT Sum of Speech Drop Rate Sum of Ho Perc Speech1 Sum of Total Drop Rate due IRAT Sum of Total Drop Rate Sum of Ho Perc Tot1 Summary Option07 Summary Option08 Summary Option09 Summary Option10
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Drop Rate (%)

1.5 1.25 1 0.75 0.5 0.25 0

Handover rate (%)

29

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

Final Test Case Description reducing RSCP threshold further

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

Enter Co mpress M o de Fo r RSCP Exit Co mpress M o de fo r RSCP Hando ver fo r RSCP Enter Co mpress M o de fo r EcNo Exit Co mpress M o de fo r EcNo Hando ver fo r EcNo Enter Co mpress mo de fo r UeTx Exit Co mpress M o de fo r UeTx Hando ver fo r UeTx (RSCP )

-1 08 -1 06 -87 -1 4 -1 2 -5 21 1 8 -67

30

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

gsmThresh3a hysteresis2d hysteresis2f hysteresis3a usedFreqThresh2dRscp usedFreqThresh2dRscpDrnc usedFreqRelThresh2fRscp utranRelThresh3aRscp TimeTo Trigger2dRscp TimeTo Trigger2fRscp TimeTo Trigger3a timeTrigg6a timeTrigg6b utranFilterCo efficient3 gsmFilterCo efficient3 gsmThresh3a filterCo efficient1 filterCo efficient2 filterCo eff6 ueTxP o werThresh6a ueTxP o werThresh6b usedFreqThresh2dEcNo usedFreqThresh2dEcNo Drnc usedFreqRelThresh2fEcNo utranRelThresh3aEcNo TimeTo Trigger2dEcNo TimeTo Trigger2fEcNo utranRelThreshRscp

-95 4 4 0 -1 07 -1 00 0 20 640 2560 0 640 2560 2 1 -95 3 3 3 21 1 8 -1 3 -1 3 0 8 640 2560 20

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

Final Test Case Results reducing RSCP threshold further

Small increase in drop rate (0.15%)

Significative increase in traffic (45%)

31

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Final Test Case Results reducing RSCP threshold further

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

32

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Final Test Case Results reducing RSCP threshold further

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

33

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Final Test Case Results reducing RSCP threshold further


Distribution of the time between Rab Establishment and IRAT Handover

Move to 2G at -105 RSCP -13 Ec/No


60 55 50 45 40 35 30 25 20 15 10 5 0 12 11 10

Move to 2G at -108 dBm -14 Ec/No

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

9 8 7 6 5 4 3 2 1 0

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

TimeRabEstToHo PDF

TimeRabEstToHo CDF

34

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

0.4

0.8

1.2

1.6

2.4

2.8

3.2

3.6

4.4

4.8

5.2

5.6

6.4

6.8

7.2

7.6

8.4

8.8

9.2

9.6

10

10.4

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

Final Test Case Results reducing RSCP threshold further


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


dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Averall Perf. Statistic (All RBs)

Averall Perf. Statistic (All RBs)

90 80 70 60 50 % 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d CmDeAct TrigGsmMeasStopCM TrigGsmMeasStopNoCm UeTxPower %

90

Move to 2G at -105 RSCP 80 -13 Ec/No 70


60 50 40 30 20 10 0 EcNo RSCP Triggering Quality TrigGsmMeasStart CmAct Trig3a PerfHo Dropped after 2d

Move to 2G at -108 dBm -14 Ec/No

UeTxPower

CmDeAct

TrigGsmMeasStopCM

TrigGsmMeasStopNoCm

35

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

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

Ec/No>qQualMin+sRatSearch (-14)

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

UMTS Cell

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


High gap in RSCP strategy between Idle and Connected mode

RSCP>usedFreqThresh2dRscp hysteresis2d/2 (101)

RSCP>usedFreqThresh2dEcNo hysteresis2d/2 (13)

GSM Coverage

36

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Start measuring GSM in Connected mode Start measuring GSM in Idle

Ec/No>qQualMin (-18)

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

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


Distribution of the time between Rab Establishment and IRAT Handover

Ec/No>qQualMin+sRatSearch (-16)

RSCP>qRxLevMin+P (-115/-112)
RSCP>qRxLevMin+sHcsRat (-111/-108)

RSCP>usedFreqThresh2dRscp hysteresis2d/2 (108)

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

RSCP>usedFreqThresh2dEcNo hysteresis2d/2 (13)

GSM Coverage

37

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Start measuring GSM in Connected mode Start measuring GSM in Idle

Ec/No>qQualMin (-18)

UMTS Cell

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

Idle mode Parameter Investigation

qHyst1=qHyst2=4dB and treSelection=2s This settings increases the probability that late reselections trigger un-necessary measurements on 2G
qHyst2=4dB
UMTS CellA Ec/No dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

2s Measure GSM

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

38

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

2G/3G Coverage values


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

Area Type Macro sites for In-train coverage Dense Urban Macro for Indoor coverage Urban Macro for Indoor coverage Suburban Macro for Indoor coverage Rural Macro for Indoor coverage Macro for In-car coverage Macro for Outdoor Coverage only Indoor site

GSM 900 GSM 1800 [dBm] [dBm] -66.3 -64.3 -69.3 -67.3 -73.3 -71.3 -73.3 -71.3 -77.3 -75.3 -80.3 -78.3 -87.3 -89.3 -85 -85

EDGE [dBm] -65 -68 -72 -72 -76 -79 -88 -85

UMTS [dBm] -72.4 -81.4 -85.4 -85.4 -89.4 -93.4 -102.4 -90.0

3G - 2G [dB] 7.4 13.4 13.4 13.4 13.4 14.4 14.4 5.0

39

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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


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 M o de U- >G qHyst1 qHyst2 qOffset1 sn qOffset2sn qRxLevM in qQualM in sRatSearch sHcsRat measQuantity1 maxTxP o werUl 4 2 1 5 0 -1 1 5 -1 8 2 4 2 24

Main KPI to monitor performances:

pmTotNoRrcConnectAttIratCellResel+pmTotNoRrcConnectAttIratCcOrd er pmTotNoRrcConnectReqCs+pmTotNoRrcConnectReqPs+pmTotNoRrcConnectReqS ms pmTotNoRrcConnectAttIratCellResel+pmTotNoRrcConnectAttIratCcOrd GtoUReselPerRrc[%]=100* er pmTotNoRrcConnectReq


GtoUReselPerTraffic[%]=100*

40

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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

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

New Idle Mode Settings

First reduction in IRAT Reselection

Second possible reduction: further monitoring needed

41

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Possible problems to be solved


Indoor/Outdoor coverage mismatch
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

2G
2G
3G

2G

2G

3G

3G

3G

42

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

Possible problems to be solved


Coverage continuity in high traffic environment
3G
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

2G

3G

43

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

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

RSCP+qHyst>GSM_RSSI-qOffset1sn dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer) Ec/No>qQualMin (-18) Ec/No>qQualMin+sRatSearch (-14) Ec/No>FDDMIN (-10) RSCP>qRxLevMin+P (-115/-112)

Start measuring GSM UMTS Cell

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

Transition to UMTS Transition to GSM

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 GSM Coverage
Ping-Pong Areas

44

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

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

Ping-Pong area

Ec/No = FDDMIN

RSCP = qRxLevMin+P

Extended UMTS Coverage (Ec/No) GSM Coverage

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.

45

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

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

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

Changing Neighbouring Strategy will create hysteresis


dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Ping-Pong area Ec/No = FDDMIN

RSCP = qRxLevMin+P

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

Extended UMTS Coverage (Ec/No) GSM Coverage

46

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

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

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


dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

Ec/No = FDDMIN RSCP = qRxLevMin+P


Start Measuring UMTS

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

Extended UMTS Coverage (Ec/No)

Ping-Pong area

GSM 1800

GSM 900

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

47

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

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


Summary:
Border and Core UMTS Sites:
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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)

48

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

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


Summary:
Border UMTS Sites (as in Option1):
dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

Core UMTS Sites (different from Option1):

Remove all 3G->2G relations


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 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)

49

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

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


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 dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer) Ec/No>qQualMin (-18) Ec/No>qQualMin+sRatSearch (-14)

RSCP>qRxLevMin+P (-115/-112)
RSCP>qRxLevMin+sHcsRat (-111/-108)

Start measuring GSM UMTS Cell

Transition to GSM

GSM Coverage

50

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

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


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 M o de G - >U QSI FDDQM IN F D D Q M IN O F F F D D R S C P M IN FDDQOFF COEXUM TS 1 0 3 0 7

M easure UM TS when GSM is mo ve to UM TS if Ec/No is > mo ve to umts if RSCP is > M o ve to UM TS when RSCP is greater than GSM + A llo w reselectio n and HO to UM TS

> -70 -8 -1 00 -inf on

51

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

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

2G->3G Handover Activation Strategy


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.

C o nne c t e d M o de G - >U QSC ISHOLEV M RSL COEXUM TS M easure UM TS when GSM is M o ve to UM TS if GSM lo ad is higher than M o ve to UM TS if EcNo is better than A llo w reselectio n and HO to UM TS

1 3 0 31 1 > -58 0 -9 on

52

dd MMMM yyyy, Speaker, Path/Filename, Classification (Menu View, Header and Footer)

También podría gustarte