58528996 IRAT HO and Re Selection Tuning SA

Aircom International TECHNICAL REPORT Prepared (also subject responsible if other)

1 (73)

No.

Senior Engineer Emerson Eduardo Rodrigues Approved

Checked

Date

Rev

2006-03-286

PA1

Reference

Technical report - GPI IRAT HO

Abstract This is a technical report for the IRAT Handover and Cell reselection project being done in MTN WCDMA network in South Africa


2 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Executive Summary Huawei and MTN SA performed during the last 7 weeks tests of the Inter Radio access Technology (IRAT) mobility between MTN’s 3G and 2G networks in South Africa. For the 3G to 2G handover a choice between CPICH RSCP and CPICH Ec/No can be done to trigger the handover. For the initial network launch when there is no or low load level and no major interference problems in the network CPICH RSCP gives a more stable IRAT handover from 3G to 2G. However in areas where there initially is a relatively high level of interference, such as high-rise buildings, CPICH Ec/No is a more safe choice of measurement measurement quantity for f or triggering the handover. One drawback of using CPICH Ec/No is that a relatively high triggering threshold is needed to have stable handovers in coverage-limited scenarios. This might lead to that t hat some users at some locations move to the 2G network even though they could have reasonable 3G coverage. The choice of measurement quantity for the triggering is a matter of trading pros of one quantity for cons with the t he other. The decision should be taken t aken bearing in mind the strategy of MTN current network deployment and future network expansions.


2 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Executive Summary Huawei and MTN SA performed during the last 7 weeks tests of the Inter Radio access Technology (IRAT) mobility between MTN’s 3G and 2G networks in South Africa. For the 3G to 2G handover a choice between CPICH RSCP and CPICH Ec/No can be done to trigger the handover. For the initial network launch when there is no or low load level and no major interference problems in the network CPICH RSCP gives a more stable IRAT handover from 3G to 2G. However in areas where there initially is a relatively high level of interference, such as high-rise buildings, CPICH Ec/No is a more safe choice of measurement measurement quantity for f or triggering the handover. One drawback of using CPICH Ec/No is that a relatively high triggering threshold is needed to have stable handovers in coverage-limited scenarios. This might lead to that t hat some users at some locations move to the 2G network even though they could have reasonable 3G coverage. The choice of measurement quantity for the triggering is a matter of trading pros of one quantity for cons with the t he other. The decision should be taken t aken bearing in mind the strategy of MTN current network deployment and future network expansions.


3 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Contents 1

INTRODUC INTRODUCTION TION ................................ ................................................. .................................. ................................. ................................. .................... ... 6 1.1 BACKGROUND................................. .................................................. .................................. .................................. ............................ ........... 6 1.2 PURPOSE ............................... ................................................ .................................. ................................. ................................. .................... ... 6 MTN’ S MOBILITY STRATEGY 1.3 IRAT ................................. ................................................. ................................. ................. 7

2

THEORY THEORY ............................... ................................................ .................................. .................................. .................................. ................................. ................ 8 2.1 BACKGROUND TO IRAT HO/CR HO/CR ................................. ................................................. ................................. .................... ... 8 2.2 CPICH RSC RSCP P ................................ ................................................. .................................. .................................. ............................ ........... 8 2.3 CPICH EC /NO ................................ ................................................. .................................. .................................. ............................ ........... 8 2.4 UPLINK AND DOWNLINK .................................. .................................................. ................................. ............................... .............. 9 2.5 IRAT HANDOVER ............................... ................................................ .................................. ................................. ....................... ....... 10 2.5.1 U2G: Triggering of 2G measurements measurements ................... ......... ................... ................... ................... ................. ........ 10 2.5.2 U2G: Compressed mode measurements................................................... 11 2.5.3 U2G: Handover from UTRAN .................................................................... 12 2.5.4 G2U: Triggering of 3G measurements measurements ................... ......... ................... ................... ................... ................. ........ 14 2.5.5 IRAT Handover strategy and possible parameter settings ................... ......... ................ ...... 16 2.6 IRAT CELL RESELECTION.................................. .................................................. ................................. .......................... ......... 17 2.6.1 U2G: Triggering of measurements............................................................. 17 2.6.2 U2G: Cell ranking ...................................................................................... 19 2.6.3 G2U: measurements ................................................................................. 20 2.6.4 G2U: Cell Ranking ..................................................................................... 21 2.6.5 IRAT cell reselection strategy and possible parameter settings ................. ......... ........ 22 2.7 SIZE OF NEIGHBOUR LIST ............................... ................................................ .................................. ............................. ............ 23 2.8 PERFORMANCE INDICATORS ................................ ................................................. .................................. ....................... ...... 24 2.8.1 IRAT HO Success Rate (U2G) .................................................................. 24 2.8.2 IRAT HO Success Rate (G2U) .................................................................. 24 2.8.3 Cell reselection outage time U2G .............................................................. 24 2.8.4 Cell reselection outage time G2U .............................................................. 24

3

METHOD METHOD ................................. .................................................. .................................. ................................. ................................. ............................. ............ 25 3.1 TESTS ................................. .................................................. .................................. .................................. .................................. .................... ... 25 3.1.1 Test phases phases................................... .................................................... .................................. .................................. .......................... ......... 25 3.1.2 Test cases cases ................................ ................................................ .................................. .................................. ............................... ............... 25 3.2 TOOLS................................. .................................................. .................................. .................................. .................................. .................... ... 26 3.2.1 TEMS Investigation scanner ...................................................................... 26 3.2.2 TEMS Investigation WCDMA WCDMA 6.0 + Terminals ................... .......... ................... ................... .............. ..... 26 3.2.3 TEMS Investigation WCDMA WCDMA 3.0.3 + Terminal .................. ......... ................... ................... .............. ..... 26 3.2.4 Equipment Set up (Initial Phase) ............................................................... 27 3.3 NETWORK LOAD................................. .................................................. .................................. ................................. ....................... ....... 27 3.3.1 Uplink Uplink load .................................. .................................................. ................................. .................................. ............................. ............ 27 3.3.2 Downlink load ............................................................................................ 27 3.4 TEST PROCEDURES ............................... ................................................ .................................. .................................. .................... ... 28 3.5 INITIAL PHASE TEST LOCATION ............................... ................................................. ................................... .................... ... 29 3.6 VALIDATION PHASE TEST LOCATION ................................. .................................................. ............................. ............ 30 3.7 TEST 1: U2G AND G2U HANDOVER BASED ON EC /NO ............................... ................................... .... 31 3.7.1 General comments .................................................................................... 31


4 (73)

No.


Checked

3.7.2 3.7.3 3.7.4 3.7.5 3.8 3.8.1 3.8.2 3.8.3 3.8.4 3.8.5 3.9 3.9.1 3.9.2 3.9.3 3.9.4 3.9.5 3.10 3.11

Date

Rev

2006-03-286

PA1

Reference

Purpose..................................................................................................... 31 Parameter ................................................................................................. 31 Theory ....................................................................................................... 31 Performance Indicators ............................................................................. 31 TEST 2: U2G AND G2U HANDOVER BASED ON RSCP .................................... 32 General comments .................................................................................... 32 Purpose..................................................................................................... 32 Parameter ................................................................................................. 32 Theory ....................................................................................................... 33 Performance Indicators ............................................................................. 33 TEST 3: U2G AND G2U CELL RESELECTION .................................................. 33 General comments .................................................................................... 33 Purpose..................................................................................................... 33 Parameters................................................................................................ 33 Test execution ........................................................................................... 34 Performance Indicators ............................................................................. 34 TEST PLAN .................................................................................................. 34 TIME PLAN .................................................................................................. 34

4

RESULTS ............................................................................................................... 35 4.1 INITIAL PHASE ............................................................................................. 35 4.2 VALIDATION PHASE ...................................................................................... 36

5

CONCLUSIONS ..................................................................................................... 37 5.1 U2G USING CPICH RSCP .......................................................................... 37 5.2 U2G USING CPICH EC /NO .......................................................................... 38 5.3 G2U HANDOVER (EC /NO AND RSCP) .......................................................... 39 5.4 IDLE MODE BEHAVIOUR (CR U2G AND G2U)................................................. 39 5.5 SIZE OF NEIGHBOUR LIST ............................................................................. 39 5.6 U2G TRIGGERING : RSCP OR EC /NO ............................................................ 40 5.7 HUAWEI RECOMMENDATION ......................................................................... 40 5.8 FURTHER STUDIES ...................................................................................... 41

6

PROPOSED PARAMETER SETTINGS .................................................................. 42

7

ABBREVIATIONS .................................................................................................. 45

8

REFERENCES ....................................................................................................... 45

9

APPENDIX A - INITIAL PHASE RESULTS ............................................................ 46 9.1 OVERVIEW .................................................................................................. 46 9.2 TEST CASE GROUP (U2G): .......................................................................... 46 9.3 TEST CASE GROUP (G2U): .......................................................................... 53 9.3.1 QSC and MRSL ......................................................................................... 53 9.3.2 FDDMRR................................................................................................... 54

10

APPENDIX B - VALIDATION PHASE RESULTS................................................... 56 10.1 OVERVIEW .................................................................................................. 56 10.2 EC /NO STRATEGY ....................................................................................... 56 10.3 RSCP STRATEGY ....................................................................................... 60 10.4 IDLE MODE RESULTS .................................................................................... 63


5 (73)

No.


Checked

11

Date

Rev

2006-03-286

PA1

Reference

APPENDIX C - GENERAL PARAMETERS ............................................................ 65 11.1 3G TO 2G HO PARAMETERS ........................................................................ 65 11.2 2G TO 3G HO PARAMETERS ........................................................................ 67


6 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Revision History

1

Introduction

1.1

Background Huawei performed an IRAT handover testing service for MTN South Africa. The service was focused on a selection of some WCDMA and GSM parameters. The IRAT handover module was divided into two phases: analysis phase and validation phase. The reason of this was to verify the findings from the initial testing in different environment to get more information to base the parameter settings decision on.

1.2

Purpose The purposes of this technical report are to: 1.

Describe what had been done and technical problems during the project

2.

Present results and findings from the tests

3.

Conclude the findings

4.

Provide final recommendation(s) on parameter settings

This report is divided into four parts: theory, method, results and conclusions


7 (73)

No.


1.3

Checked

Date

Rev

2006-03-286

PA1

Reference

MTN’s IRAT mobility strategy MTN strategy regarding the IRAT mobility can be summarised with the following points: Stay in 3G as long as possible, so as to maximize the enhanced service capacity offered in 3G. Use 2G as a safety net, and allow seamless handovers between the 2G and 3G networks. Move back to 3G as quickly as possible once there is 3G coverage if the UE has to go to 2G due to a lack of 3G coverage. The identified solution for the short term should be consistent with MTN network evolution, be it coverage extension or capacity growth.


8 (73)

No.


Checked

2

Theory

2.1

Background to IRAT HO/CR

Date

Rev

2006-03-286

PA1

Reference

The purpose of IRAT handover functionality is to let a user who cannot access the 3G network or retain its service in the 3G network handover or reselect to a 2G backbone network instead. Whether or not the users can access/retain the connection to the 3G network will be dependent on the required and available power in uplink and downlink. The required power will in among other things be dependent on the pathloss and interference situation in up and downlink. One way of estimating the pathloss is the received signal code power (RSCP) of the CPICH. However the CPICH RSCP does not take any interference into consideration. The Ec/No of the CPICH is a measurement that takes both the pathloss and the interference situation into consideration.

2.2

CPICH RSCP The CPICH Received Signal Code Power (CPICH RSCP) is dependent on the CPICH transmitted code power, the pathloss. Since the transmitted power on the CPICH is constant the CPICH RSCP will primarily be affected by the pathloss. Hence a decrease in the CPICH RSCP will mean that the pathloss in both up and downlink has increased. The absolute accuracy requirement for the CPICH RSCP measurements is 6-8 dB [1].

2.3

CPICH Ec/No The CPICH Ec/No is defined as the energy per chip divided by the total in-band interference. Theoretically it is defined as the CPICH RSCP divided by the RSSI (received signal strength indicator). CPICH Ec/No = CPICH RSCP / RSSI Both the CPICH RSCP and the CPICH Ec/No will increase with a decrease in the pathloss. However the Ec/No is both proportional to the pathloss and inversely proportional to the RSSI. Thus, the relation between CPICH Ec/No and pathloss is not linear as in the RSCP case. Particularly at the cell border in cases where there is mainly one dominant server the behaviour of the CPICH Ec/No is not completely straightforward.


9 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

CPICH Ec/No

RSSI

CPICH RSCP

Pathloss

The figure above describes the principle of the scenario previously mentioned. As can be seen the CPICH Ec/No remains relatively constant since both the pilot signal and the interfering signals are more or less subject to the same pathloss. Once the thermal noise becomes the dominant part of the RSSI the CPICH Ec/No also starts to degrade but not as steep as the RSCP degradation. When the signal finally becomes lower than the thermal noise floor the Ec/No will drop sharply. This description of the scenario is highly simplified. The advantage with the CPICH Ec/No measurement is that it takes into consideration the downlink interference situation in the network through its dependency on the RSSI. The absolute accuracy requirement for the CPICH Ec/No measurements is 1.5-3 dB, which is considerably higher than the requirements for the RSCP [1].

2.4

Uplink and Downlink None of the previously discussed measurements does directly cater for the uplink coverage and interference criteria required for the dedicated channel to retain the connection. However if the uplink and downlink are relatively balanced from a required and available power point of view, one could use the CPICH RSCP to represent the uplink as well as downlink pathloss. When letting the CPICH RSCP represent the maximum tolerable pathloss for the dedicated channel in both up and downlink one has to take into consideration a few issues. The mapping will be dependent on whether or not a ASC (TMA) is used, the UE Tx power class, the downlink maximum code power for the dedicated channel, the CPICH Tx power etc. The mapping of the CPICH Ec/No to the up and downlink interference situation is slightly more complicated. Since the load situation might be different in the up and downlink the CPICH Ec/No has no direct correlation with the uplink interference situation. However in most cases one could suspect that the downlink load level could be higher than the uplink load level due to the asymmetric PS radio bearers and the higher power consumption in the downlink due to soft handover.


10 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

2.5

IRAT Handover

2.5.1

U2G: Triggering of 2G measurements

Reference

To perform the handover from the 3G network the UE has to conduct measurements on the 2G frequencies of the 2G neighbour cells. The set up of those measurements are initiated when the UE is reporting event 2d to UTRAN in a measurement report. The event 2d happens when the CPICH RSCP of all cells in active set drops below the threshold usedFreqThresh2dRscp minus the value hysteresis2d /2 for TTT2d milliseconds. Alternatively will the event 2d happen when the CPICH Ec/No of the cells in active set drops below the threshold usedFreqThresh2dEcno minus the value hysteresis2d /2 for TTT2d milliseconds. After receiving a measurement report with the event 2d UTRAN will sent the message physical channel reconfiguration to the UE for the UE to reconfigure to compressed mode by spreading factor reduction (SF/2). Compressed mode

SF=SF SF=SF

T = 10 ms

0

0

/2 SF=SF

0

time for measurements

f

After this a measurement control message will be sent to the UE containing the monitored neighbour list for the 2G neighbour cells. The UE will then tune into the frequencies of the neighbouring 2G cells in the neighbour list during the empty slots created in the compressed frames. In case there are no 2G neighbours defined for the cells in active set, UTRAN will not command the UE into compressed mode since there are no neighbour cell frequencies to perform measurements on. The compressed mode measurements are stopped either when the UE handover to appropriate 2G cell or if the UE reports the event 2f in a measurement report to UTRAN. UTRAN will then send a measurement control message to the UE to release the compressed mode measurements.


11 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

The event 2f happens when the CPICH RSCP of at least one cell in active set rises above the threshold usedFreqThresh2dRscp plus usedFreqRelThresh2fRscp plus the value hysteresis2f /2 for TTT2f milliseconds. Alternatively the event 2f will happen when the CPICH Ec/No of at least one cell in active set rises above the threshold usedFreqThresh2dEcno plus usedFreqRelThresh2fEcno plus the value hysteresis2f /2 for TTT2f milliseconds.

2.5.2

U2G: Compressed mode measurements The measurement control sent after the physical channel reconfiguration also contains information about the requirements for the triggering of event 3a. Event 3a happens when the CPICH RSCP of all cells in active set drops below the threshold utranThresh3aRscp minus the value hysteresis3a /2 for TTT3a milliseconds or when the CPICH Ec/No of all cells in active set drops below the threshold utranThresh3aEcno minus the value hysteresis3a /2 for TTT3a milliseconds. In addition to either of the two previous conditions the carrier RSSI of the target GSM cell has to be above the threshold gsmThresh3a and the verified that the BSIC of the target cell is the same as the cell in the monitored list.

The 2G measurements required for the UE to be able to evaluate if the event 3a criteria is fulfilled, are performed in the following way.


12 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

The UE measures the RSSI of all carriers in the neighbour list and rank them according to signal strength. When using compressed mode with gap length 7 slots, the UE is required to take 6 samples per gap [1]. Each carrier shall be measured 3 times, which means that 2 carriers can be measured per gap. The time to measure and rank all carriers are hence dependent on the number of 2G neighbours in the neighbour list. In one reporting period of 480ms the UE can measure 12 neighbours, in 960ms 24 neighbours and so on. The RSSI measurements are continuously repeated to maintain the ranking of the cells. 3/8 of the measurement occasions are used for the RSSI measurements. The UE is also required to attempt BSIC decoding of the 8 highest ranked cells (the ranking might be continuously changing). The BSIC decoding can only be done if the BSIC transmission is done fully within one transmission gap. In the worst case the UE will need 2 attempts to decode the BSIC, which will mean that the decoding will take 5.28seconds per carrier [1]. Half of the measurement occasions are dedicated to BSIC decoding and 1/8 of the measurement occasions are dedicated for BSIC reconfirmation, since the BSIC is only considered as verified for a certain period of time. If a cell is BSIC decoded and the criteria for event 3a is fulfilled the UE will send a measurement report indicating the number in the neighbour list for the cell fulfilling the criteria. This will be the target cell to perform the handover to. Since the BSIC verification is quite time consuming it could be so that it is not the best ranked cell that is BSIC decoded since the ranking might have changed during the decoding time. If the event criteria’s are fulfilled the decoded cell will be reported even though it is no longer the strongest in the ranking. It is therefore desirable to keep the neighbour list as short as possible as to minimise the probability of this happening and so that the ranking list can be quickly updated.

2.5.3

U2G: Handover from UTRAN Upon reception of a measurement report indicating the event 3a the RNC attempt to allocate resources in the target GSM cell. If and when the resources are secured a handover from UTRAN command message is sent from the RNC to the UE. When the handover is completed the UE will send the message handover complete to the BSC. This will initiate the release of the resources in the UMTS network. In case the UE fails to connect to the GSM network it will send the message handover from UTRAN failure to UTRAN. If possible the call will be kept and another handover attempt made. The flowchart below depictures a successful handover from UMTS to GSM.


13 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

All parameters defining the different thresholds used during the UMTS to GSM handover are set per RNC and per UMTS cell. Hence, for the RNC parameters one value has to be found that can cater for the whole network. The picture below describes the complete IRAT handover procedure from U2G.


14 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Measurement Quantity (Ec/No and RSSI)

WCDMA cell

hysteresis2f/2

usedFreqRelTresh2fEcno usedFreqTresh2dEcno

hysteresis2d/2

hysteresis2d/2 gsmTresh3a hysteresis3a/2

utranTresh3aEcno

GSM cell Reporting Reporting event 2d event 2f

Reporting event 2d

Reporting event 3a

A user entering compressed mode will be forced to roughly double its Tx power during the part of the compressed frame when the data is transmitted. For downlink Compressed Mode (CM) the available Tx power is increased during the compressed frames but for the uplink the UE max Tx power is setting the limit. If the CM is performed when the UE already is transmitting close to its maximum output power, it might lead to increased BER and BLER due to the inability of the power control to combat the fast fading dips. Since there is an increased interference situation in the network due to the compressed mode, it is desirable to have as few users as possible in CM and each user in CM for as short time as possible. To reduce the number of users in compressed mode, low values on the triggering thresholds are desirable (i.e. moving into CM late). This would mean that users are entering CM at low signal levels/low Ec/No quality levels. The IRAT handover functionality is further described in [2].

2.5.4

G2U: Triggering of 3G measurements The measurements performed on the 3G neighbours while on a voice connection in 2G is initiated at a certain level of signal strength. The parameter QSC defines a signal strength criterion, which has to be fulfilled before measurements on the UMTS neighbours are performed. The threshold can either be set to a level (-98 - -74dBm) for which the GSM signal strength has to be below before the measurements are performed. The other option is to set QSC to a level (-78 - -54dBm) for which the signal strength has to be above before the measurements are performed. Setting the QSC parameter can also disable the evaluation of the GSM signal strength, so that measurements are never performed or permanently enabled by setting the QSC to always.


15 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

The parameter ISHOLEV defines the load level is the GSM network which should be exceeded before starting to evaluate the reported UMTS measurements. ISHOLEV is defined as the percent of idle TCH’s than need to be allocated. Hence the value 99% means that measurements on UMTS are always evaluated. The information on when to measure along with 3G frequency and scrambling codes of the 3G neighbouring cells are sent to the UE in the Measurement Information message. The UE may use the search frames, which are not required for BSIC decoding, for these measurements. If indicated by the parameter SPRIO = YES, the UE may use up to 25 search frames per 13 seconds without considering the need for BSIC decoding in these frames. The UE shall report a new best UTRAN cell, which is part of the neighbour cell list, at the latest 5 seconds after it has been activated under the condition that there is only one UTRAN frequency in the neighbour cell list and that no new GSM cells are activated at the same time and under good radio conditions [3]. If a new GSM frequency is activated during the measurements, the required minimum reporting time is extended by the time to decode the BSIC of the new cell. The reporting of UMTS cell measurements are done in the same measurement reports as the GSM measurements. Since there is only possible to report 6 cells in the measurement report the number of reported GSM cells are reduced for a multi RAT capable UE. The parameter FDDMRR (1-3) defines how many positions in the measurement report that should be dedicated to 3G cells. The GSM measurement report fields RxLev indicates the received signal strength of the reported GSM cells in the neighbour list coded according to [3]. For the reported 3G cells the reported value does not indicate the signal strength but represents the CPICH Ec/No decodes according to the following formula: CPICH Ec/N0 [dB] = Reported value / 2 – 24.5 The reporting field BCCH-INDEX represents the position in the neighbour list for the 2G cells. For the 3G cells it is always reported as 31 (representing only 31 GSM neighbour cells when UMTS neighbour is defined). Finally the reported field BSIC represents the BSIC for the GSM cells and the position in the 3G neighbour list for the 3G cells. The handover decision is made depending on whether or not the reported CPICH Ec/No exceeds the value represented by the parameter MRSL. The reporting and evaluation of 3G neighbour cells does not affect the locating algorithm used for the ranking and evaluation of reported 2G neighbours.


16 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

WCDMA Cell measurement

Filtering Urgency condition B

T

S

T

T

% idle TS: 1/6 Inter System Handover algorithm

T

T

16, 7%

Traffic load

ISHOLEV = 99 %

% idle TS ISHOLEV

Ec/No > MRSL

Basic ranking

Add WCDMA cell to candidate list

Radio Network functions evaluations

Organizing the list

Sending the list

Allocation reply

All parameters used to define the different thresholds used in the G2U handover are all set per GSM cell.

2.5.5

IRAT Handover strategy and possible parameter settings One desirable parameter setting could be to stay with the voice connection in the 3G network as long as the quality of the connection is sufficient. When this criteria is no longer fulfilled it could be desirable to move the connection to the 2G network. Finally it could be desirable to move the connection back to the 3G network again when the quality is sufficient, but this is not so important since the end user perception of the voice call should be fairly similar in both 2G and 3G. When trying to achieve this in practice one has to take many other aspects into consideration. One of the considerations is to be able to perform the actual measurements and the handover before the quality becomes to bad. When in compressed mode the terminal will need to approximately double the power during the compressed slots. This will lead to a reduced coverage since the available power is constant (in uplink). If for example driving in high speed out of the coverage region one has to back off the handover thresholds so that the connection can be maintained during the measurement and relocation time while still in compressed mode. The measurement criteria for moving from U2G and G2U previously described can be descripted as follows:


17 (73)

No.


Checked

3G

Date

Rev

2006-03-286

PA1

Evaluate Ec/No

Reference

2G

or 3G

Evaluate RSCP

2G

When moving from U2G either CPICH RSCP or CPICH Ec/No can be chosen as measurement quantity. When moving from G2U the standard only allows for CPICH Ec/No measurements to be reported.

2G

Evaluate Ec/No

3G 3G

One aspect to consider is the possible ping - pong effects while moving between 3G and 2G. If CPICH RSCP is chosen as measurement for the U2G handover, special care has to be taken when setting the parameter MRSL (Ec/No) for the G2U handover. Since there is no exact relationship between the CPICH Ec/No and the CPICH RSCP it might be so that the CPICH Ec/No is better than the value defined by MRSL at the point when the U2G handover is triggered due to a low CPICH RSCP. This would then result in a ping - pong handover from G2U since the criteria CPICH Ec/No > MRSL is fulfilled. However for the voice service, this scenario can be avoided by not letting the user perform the G2U handover at all (ISHOLEV=0% or QSC = never) or by setting the MRSL threshold to a very high value to delay the G2U handover. However the limitation with being in 2G with a voice service compared to the 3G-voice service is that the availability of the multi-RAB (voice + PS64) in the 3G network.

2.6

IRAT Cell Reselection

2.6.1

U2G: Triggering of measurements In idle mode the UE is required to start doing measurements on 2G cells and evaluate them in a cell ranking procedure when the following criteria is fulfilled: CPICH Ec/No <= qQualMin + sRatSearch


18 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Where the parameter qQualMin is the minimum acceptable CPICH Ec/No level for camping on a 3G cell and the parameter sRatSearch defines the offset from qQualMin at which levels the measurements should start. Both parameters are set per UMTS cell.

This means that the UE is not required to always measure and evaluate the 2G cells in idle mode. The parameters previously defined are transmitted in the system information SIB3. The neighbour list with 2G neighbours is transmitted to the UE in the system information SIB11. The parameter qHyst2 is the hysteresis parameter used for the 3G-cell reselection evaluation. For a 3G-cell to be replaced by another 3G cell it is required to be at least qHyst2 dB better (in terms of CPICH Ec/No) for the cell reselection to take place according to the picture below.


19 (73)

No.


Checked

CPICH_Ec/No

Date

Rev

2006-03-286

PA1

Reference

qHyst2 qHyst2

qHyst2

The blank dotted line is the serving cells.

qQualMin treSelection

treSelection treSelection treSelection

Time

Although this hysteresis parameter only effects the 3G-3G cell re-selection it also have an impact on the 3G-2G cell reselection. If the current 3G cell’s CPICH Ec/No drops below the criteria for 2G measurements, 2G cells will be measured and evaluated and a 3G to 2G cell reselection performed if the predefined criteria’s are fulfilled. At the same time it could be so that there exists a 3G cell with a CPICH Ec/No better than the threshold for 2G measurement evaluation but at the same time less than qHyst2 dB better than the current 3G cell. This would effectively mean that the UE start camping on a 2G cell even though there exist a 3G cell better than the required quality criteria for staying in the 3G network.

2.6.2

U2G: Cell ranking If the parameter qualMeasQuantity = 2 (CPICH Ec/No used for ranking), the cell ranking is done in two steps. First a ranking is done based on the signal strength (CPICH RSCP and GSM RxLev), if a GSM cell turnout to be the highest ranked no more ranking is done and a cell reselection to the GSM cell is done. If a W CDMA cell is strongest, a second ranking is done based on CPICH Ec/No with only WCDMA cells. If the parameter qualMeasQuantity = 1 (CPICH RSCP) only one cell ranking need to be done with all cells (2G and 3G) included. The cell ranking between 3G and 2G cells is done in the following way (also described in the picture below): R(serving 3G cell) = Qmeas(s) + qHyst1 R(neighbour 2G cell) = Qmeas(n) - qOffset1sn If the R criterion for the 2G cell is higher than the R criteria for the 3G cell for the time duration treSelection seconds, a cell reselection to the 2G cell is done.


20 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

The quantity Qmeas is the CPICH RSCP for the 3G cell and RxLev for the 2G cell. The parameter qHyst1 is the hysteresis parameter affecting the signal strength (not the Ec/No), and the parameter qOffset1sn is a parameter used to offset the measured GSM signal strength in the ranking.

2.6.3

G2U: measurements The start of the measurements on the 3G cells is, in the same way as for the dedicated mode, started when the GSM signal strength is either above or below the value indicated by the parameter QSI .

The parameter can also be set so that measurements on the 3G neighbours are either never or always performed. The latter alternative is highly desirable in case one wishes to utilise the 3G network as much as possible.


21 (73)

No.


Checked

Date

Rev

Reference

2006-03-286

PA1

The parameter QSI is defined per GSM cell and is transmitted together with the 3G neighbour cell list in the system information type 2quater. The UE shall be able to identify and select a new best UTRAN cell on a frequency, which is part of the 3G Cell Reselection list, within 30 seconds after it has been activated under the condition that there is only one UTRAN frequency in the list and under good radio conditions [3]. For the measured 3G cell to be considered in the cell ranking procedure the Ec/No of the measured 3G neighbour has to at least fulfil the following criteria: CPICH Ec/No > FDDQMIN Where the parameter FDDQMIN can be set from 0 to 7 representing an Ec/No values from –20 to –6 dB. The table below will

2.6.4

UE Mapping of value

0

1

2

3

4

5

6

7

[dB]

-20

-6

-18

-8

-16

-10

-14

-12

G2U: Cell Ranking The cell ranking for the cell reselection from 2G to 3G is done in the following way: CPICH Ec/No > FDDQMIN (prerequisite for ranking) AND CPICH RSCP > RLA(s+n) + FDDQOFF where the RLA(s+n) is the average (per cell) signal strength of the serving 2G cell and its 2G neighbours. The parameter FDDQOFF defines an offset for the measured 2G signal strength (-28 - +28 dB) used to prioritise the 2G or 3G cell in the ranking. There is also a possibility to set the parameter to - to always prioritise the 3G cell and thereby effectively disable the cell ranking procedure. FDDQOFF is defined per 2G to 3G neighbouring cell relations. Up to 15 seconds after a U2G cell reselection the parameter FDDQOFF is automatically increased by 5 dB to reduce the risk of ping-pong between 3G and 2G if the desire is to try to compare the signal strengths.


22 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Measurement Quantity

CPICH RSCP CPICH Ec/No

FDDQOFF FDDQMIN

5 seconds CPICH Ec/No > FDDQMIN

IRATCC to CPICH RSCP > WCDMA GSM RLA+ FDDQOFF

2.6.5

GSM RLA

t

IRAT cell reselection strategy and possible parameter settings As in the case of connected mode it could be desirable to set the idle mode parameters so that the UE is camped on the 3G network as long as it can access the network with a high probability and with a acceptable quality. However since this probability is dependent on both the signal strength (pathloss) and quality (Ec/No) of the best 3G cell independently (although they are correlated). In the current standard the CPICH Ec/No check is always done first, followed by a CPICH RSCP and GSM RxLev comparison. Hence the standard allows for a cell reselection process, which can be described with the following flowchart:

3G

Evaluate Ec/No

Evaluate RSCP (relative GSM)

2G


23 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Since the signal strength evaluation is done by comparing the signal strengths of the 3G cell with the signal strength of the 2G cells, tuning on cell relation level, of the offset values (qOffset1sn) has to be made. This is only feasible if full co-siting of all 2G and 3G cells are done. In case of complete co-siting of 3G cells with 2G cells, one single offset could probably be used for the whole network. However in the case there are more 2G sites, e.g. indoor sites, hotspot sites etc. the offsets has to be tuned on cell relation level since the 2G signal strength could be considerably higher than the 3G signal strength although sufficient to provide a 3G service. For the G2U cell reselection the current standard can be described with the following flow chart:

2G

Evaluate Ec/No

Evaluate RSCP (relative UMTS)

3G

In this case both the RSCP and the Ec/No criteria should be fulfilled before the cell reselection from G2U is performed. This reduced the probability pf a ping-pong cell reselection between 2G and 3G. Since the signal strength evaluation is here also done by comparing the 3G and 2G signal strength, parameter tuning on cell relation level is required here as well.

2.7

Size of neighbour list It is always recommended to keep the neighbour lists short, both for the 2G neighbours while in 3G and 3G neighbours while in 2G. It is hence crucial that the correct neighbours are defined, and that no unnecessary ones are chosen. As an example, after IRAT cells change to a cell, which turns out to be less preferable than some other, a cell reselection in GPRS is required. That process takes time, during which the overall throughput and user experience Measurements on IRAT neighbours, is a demanding process for the UE. Therefore, the number of inter-system neighbours must be kept low, around 10. Another reason for keeping the neighbour lists short is the scenario described in the CM theory section.


24 (73)

No.


Checked

2.8

Performance Indicators

2.8.1

IRAT HO Success Rate (U2G)

Date

Rev

2006-03-286

PA1

Reference

The IRAT handover Success rate is defined as the number of successful IRAT handovers divided by the number of occasions where an IRAT handover could be expected based on the signal and quality levels. If the handover is not successful the reason for the failure should be evaluated. This performance indicator is highly correlated with the drop call rate. This performance indicator is based on subjective evaluation of the logfiles and not necessary the exact occurrence of a specific message.

2.8.2

IRAT HO Success Rate (G2U) The IRAT handover Success rate is defined as the number of successful IRAT handovers divided by the number of occasions where an IRAT handover could be expected based on the quality levels. If the handover is not successful the reason for the failure should be evaluated. This performance indicator is based on subjective evaluation of the logfiles and not necessary the exact occurrence of a specific message.

2.8.3

Cell reselection outage time U2G The cell reselection outage time is measured as the time from starting to read GSM system information until the UE is has received a location area update accept message from the GSM network.

2.8.4

Cell reselection outage time G2U The cell reselection outage time is measured as the time from starting to read UMTS system information until the UE is has received a location area update accept message from the UMTS network.


25 (73)

No.


Checked

3

Method

3.1

Tests

3.1.1

Test phases

Date

Rev

2006-03-286

PA1

Reference

The testing was divided into an initial test phase followed by an analysis phase and a validation phase. The initial test phase was used to narrow down the possible parameter settings to a few sets. Those parameter settings were later tested in different locations to find the best possible setting. The test cases were defined prior to the start of the testing and can be found in “Test Specifications IRAT HO” and “Validation Specification IRAT HO” documents. After some of the test cases during the initial testing was carried out the priorities of the remaining initial test cases was redone to better utilise the remaining time.

3.1.2

Test cases

3.1.2.1

IRAT Handover Since the time for the testing was limited many parameters were set to default value or to a theoretical value estimated based on other parameter values. The testing was then focused on the following parameters:

usedFreqThresh2dRscp usedFreqThresh2dEcno ttt2d usedFreqThresh2fRscp usedFreqThresh2fEcno ttt2f utranThresh3aRscp utranThresh3aEcno MRSL sRatSearch FDDQMIN FDDMRR QSC The strategy for the testing was basically to find an as aggressive value as possible (i.e. still successful handovers) on each parameter using either CPICH Ec/No or CPICH RSCP as a trigger. This would mean staying as long as possible on the 3G network.


26 (73)

No.


Checked

3.2

Tools

3.2.1

TEMS Investigation scanner

Date

Rev

2006-03-286

PA1

Reference

TEMS Investigation for WCDMA v6.0 Scanner module was used in the drive route preparation phase to measure the following items:

CPICH Ec/No CPICH RSCP CPICH Scrambling Codes GSM BCCH carrier RxLev on target cells C/I on GSM target cells

3.2.2

TEMS Investigation WCDMA 6.0 + Terminals

3.2.2.1

Nokia 6630 The Nokia terminal supports all kinds of handover scenarios in Idle, voice service and PS data service modes. It always uses compressed mode but do not have so good reporting capabilities capabilities to TIW in terms of BLER, SIR and SIR target etc.

3.2.2.2

Sony Huawei V800 The Nokia terminal supports all kinds of handover scenarios in Idle, voice service and PS data service modes.

3.2.3

TEMS Investigation WCDMA 3.0.3 + Terminal

3.2.3.1

Motorola A835 (additional equipment for the Validation phase) The Motorola terminal supports cell reselection in idle mode both from 3G to 2G and back to 3G again as well as voice service handover from 3G to 2G. No other IRAT handover scenarios are supported.


27 (73)

No.


3.2.4

Checked

Date

Rev

2006-03-286

PA1

Reference

Equipment Set up (Initial Phase)

2

Long Call Test Voice

UETR/MTR in OSS-RC

Voice

Uu GPS

1

TEMS Scanner TEMS Investigation WCDMA

Iub

RB S

RN C Call test (B-party)

All the equipments within the dotted box is located in car or portable setup Measurement Point

3.3

Network Load

3.3.1

Uplink load It is practically impossible to generate uplink load in an artificial way. Therefore a load margin has to be considered when analysing analysing the results.

3.3.2

Downlink load Increasing the output power on the BCH channel can generate the downlink load. This will not exactly resemble a real load situation but can be considered an acceptable approximation of how the system and UE will perform under downlink load. Different load situations will have different effects on how many users that are in compressed mode and how long time they are spending in compressed mode. Note that the BchPower parameter defines the power on the BCH relative to the power on the CPICH. The table below shows the load at different BCH power settings assuming the feeder loss parameter set to 0 dB.


28 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

RBS nominal power 17.4W

3.4

BchPower [0.1dB]

PTot at RBS [W])

% Load

-31

1.8

10%

30

3.1

18%

60

4.9

28%

80

7.0

40%

100

10.3

59%

Test procedures At the test location, before the drive test, the t he engineer sets up the equipment equipment in the car c ar or the portable equipment according to the test equipment equipment setting specified in Test specifications. The engineer also verifies that the UETR is logging and the network is set with the proper IRAT HO parameter value before starting the drive tests. During the initial test phase the parameters were changed according to the procedure described in Test Specification. For the Validation phase, three different groups of parameter settings with three test cases each were predefined.


29 (73)

No.


3.5

Checked

Date

Rev

2006-03-286

PA1

Reference

Initial Phase test location Hillbrow, Benrose and the surrounding area have been chosen as the test area for the initial phase. In this area the GSM coverage is always good, while in UMTS (JHB CBD_2 cluster) some RSCP coverage holes and a significant number of low Ec/No-RSCP locations were detected. The radio environment is deeply influenced by the urban structure, where high block buildings, narrow streets and many corners can be seen. All the sites involved in the measurement were set onto Germiston RNC, and for each cell the BchPower value was set to 20 in order to simulate a certain load on the network. The main reason for that was to get comparable results between RSCP and Ec/No analysis strategies. Geography: All along the route there are hills and flat areas where signal can be seen far away (Crown Mines) or blocked (Eastgate).


30 (73)

No.


3.6

Checked

Date

Rev

2006-03-286

PA1

Reference

Validation phase test location Fourways and the surrounding area have been chosen as the test area for the validation phase. A different test area as well another RNC (Randburg) were chosen for the measurements, in order to verify the possibility for the adopted parameter settings in numerous places. In this area the GSM coverage is always suitable, while in UMTS (Fourways cluster) some RSCP coverage holes and a significant number of low Ec/NoRSCP locations were detected. The 3G radio environment is floating due to the presence of several kind of clutter areas: urban and sub urban blocks, residential estates and open fields. Geography: The entire route is changing between hilly and flat areas where signal can be seen far away (e.g. Fairlands).


31 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

3.7

Test 1: U2G and G2U handover based on Ec/No

3.7.1

General comments

Reference

For this test the predefined settings should be used with a predefined load level. 3.7.2

Purpose The purpose of the test is to find out the pros and cons of each setting in each environment.

3.7.3

Parameter Tested parameters utranThresh3aEcno usedFreqThresh2dEcno usedFreqThresh2fEcno timeToTrigger2f MRSL QSI/QSC FDDMRR measQuantity2 = 2 (Ec/No) utranMeasQuantity3 = 2 (Ec/No) Parameters fixed during the tests ueTxPowerThresh6a = 21 ueTxPowerThresh6b = 18 TimeToTrigger2d = 11 sRatSearch = 4 FDDQMIN = 7 FDDQOFF = 0 SPRIO = YES QSCI = 1 ISHOLEV = 99

3.7.4

Theory The procedure should be repeated for each parameter set.

3.7.5



32 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

Drop call rate IRAT HO success rate T3G T2G TCMwithoutHO TCMwithHO THOCompl

3.8

Test 2: U2G and G2U handover based on RSCP

3.8.1

General comments For this test the predefined settings should be used with a predefined load level.

3.8.2

Purpose The purpose of the test is to find out the pros and cons of each setting in each environment.

3.8.3

Parameter Tested parameters utranThresh3aRscp usedFreqThresh2dRscp usedFreqThresh2fRscp timeToTrigger2f MRSL QSI/QSC FDDMRR measQuantity2 = 1 (RSCP) utranMeasQuantity3 = 1 (RSCP) Parameters fixed during the tests ueTxPowerThresh6a = 21 ueTxPowerThresh6b = 18 TimeToTrigger2d = 11 sRatSearch = 4 FDDQMIN = 7 FDDQOFF = 0 SPRIO = YES


33 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

QSCI = 1 ISHOLEV = 99

3.8.4

Theory The procedure should be repeated for each parameter set

3.8.5


Drop call rate IRAT HO success rate T3G T2G TCMwithoutHO TCMwithHO THOCompl

3.9

Test 3: U2G and G2U cell reselection

3.9.1

General comments For this test the predefined setting should be used in each of the predefined environments and which each of the predefined load levels.

3.9.2

Purpose The purpose of the test is to find out the performance with the selected setting

3.9.3

Parameters Tested parameters sRatSearch FDDQMIN Parameters fixed during the tests utranThresh3aEcno usedFreqThresh2dEcno


34 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

usedFreqThresh2fEcno measQuantity2 (2 = Ec/No) utranMeasQuantity3 (2 = Ec/No) FDDQOFF= 0 FDDMRR = 1 SPRIO = YES QSI = 7 QSC = 7 QSCI = 1 ISHOLEV = 99

3.9.4

Test execution The procedure should be repeated for each parameter set in each environment and for each load case.

3.9.5


Ec/No vs. RSCP plot when performing the reselection. Ping pong rate

3.10

Test plan The tests will be performed with three different sets of parameters per test group: Strategy Ec/No e2d e2f ttt2f e3a MRSL QSC/QSI FDDMRR sRatSearch FDDQMIN

3.11

-11 2 13 -13 32 7 2 -

Time Plan Strategy Ec/No

-12 2 13 -13 30 8 3 -

-12 2 12 -13 29 9 3 -

Strategy RSCP -100 2 13 -104 32 7 2 -

-102 3 13 -104 30 8 3 -

-102 3 13 -105 29 9 3 -

Cell Reselection

8 3

4 5

2 7


35 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

st

1 parameter set 2005-09-07 nd

2

parameter set 2005-09-07

rd

3 parameter set 2005-09-07

Strategy RSCP st


2


rd


Cell Reselection st


2


rd


4

Results

4.1

Initial phase During the initial phase several triggering thresholds were tested separately or in pairs of two in the test area. The results of this can be found in Appendix A. Based on those results the test settings were narrowed down to the following table and the load was increased to BCH Power = 40 to test the “brake point” for RSCP measurement.


36 (73)

No.


Checked

Ec/No Strategy

Date

Rev

2006-03-286

PA1

RSCP Strategy

Reference

Idle mode

VTC 1.1

VTC 1.2

VTC 1.3

VTC 2.1

VTC 2.2

VTC 2.3

VTC 3.1

VTC 3.2

VTC 3.3

e2d

-11

-12

-12

-100

-102

-102

-11

-11

-11

e2f

2

2

2

2

3

3

2

2

2

ttt2f

13

13

12

13

13

13

13

13

13

e3a

-13

-13

-13

-104

-104

-105

-12

-12

-12

sRatSearch

4

4

4

4

4

4

8

4

2

MRSL

32

30

29

32

30

29

29

29

29

QSC/QSI

7

8

9

7

8

9

7

7

7

FDDMRR

2

3

3

2

3

3

1

1

1

FDDQMIN

7

7

7

7

7

7

3

5

7

Due to the possibility to use different measurement quantities in the network for IRAT HO events triggering, three different strategies for parameter testing were created. The first strategy is based on Ec/No, the second on RSCP and the last one is focusing on the idle mode behaviour. During the initial phase we experienced missing messages on the NOKIA 6630: due to this reason we decided to add an additional terminal (Motorola A835) with the aim to collect more complete message flows for the analysis.

4.2

Validation phase During the validation phase the settings described in the table above were used together with an increased BchPower value in the specified location. The results for these tests can be found in Appendix B. The analysis of the Nokia 6630 data was more time consuming due to the already mentioned problem of some missing messages, so it was more difficult to follow the message flow during a specific dedicated mode phase (e.g. Call Setup, IRAT HO procedure). Therefore, sometimes it was almost impossible to find the correct trigger time for events. On the other hand, the Sony Huawei V800 and the Motorola A835 showed a more consistent message flow. It was also experienced a different UE sensitivity: the V800 is more sensitive and due to this fact it was exposed to CM ping-pong effects more than the other two mobiles. The A835 instead has less sensitivity in the receiver, and due to the older SW/HW version, some freezing and some abnormal TIW disconnections were experienced.


37 (73)

No.


Checked

5

Conclusions

5.1

U2G using CPICH RSCP

Date

Rev

2006-03-286

PA1

Reference

The CPICH RSCP is a more stable triggering quantity in areas with low interference and low signal strength than CPICH Ec/No. When driving out of coverage in such environment the RSCP triggering quantity can make sure that the handover is made at a relatively low pathloss regardless of interference situation. This reduces the probability of the UE remaining in the 3G network at a high pathloss to high for the UE to access the network. However if the interference situation increases it might not be the pathloss that sets the limit for the accessibility to the network but the interference. The pathloss where the UE loses its coverage will depend on the link budgets for the required service. This will in turn be dependent on e.g. availability of ASC (TMA) in the uplink and UE Tx power capability. The plot below shows the UE Tx power distribution for different CPICH RSCP for all tested UE when driving out of coverage in the Fourways area. UE Tx power

25 20 15 10 r e w o p x T E U

5 0 -5 -10 -15 -130,0

-120,0

-110,0

-100,0

-90,0

-80,0

-70,0

-60,0

-50,0

-40,0

-30,0

RSCP

Total

It can be seen that if selecting the handover threshold for event 3a to –104dBm there are still margins left to the maximum UE Tx power of 21dBm.


38 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

The drawback with using the CPICH RSCP as measurement quantity is its insensitivity to load in both up and downlink. The received power on the pilot channel will always be the same regardless of load situation in either up or downlink. If the load situation increases the UE Tx power and downlink code power will increase necessarily to keep the connection. One type of location where this (at a very sparsely loaded network) is a problem, is at places which are reached by many overshooting cells without having any dominant server. Another drawback with the use of CPICH RSCP as a measurement quantity for the U2G handover and cell change is that CPICH Ec/No has to be used for the G2U handover and cell reselection. If RSCP is used for the U2G triggering there is a potential for ping-pong effects between 3G and 2G. The ping-pong effects might not be such a big problem for the voice service since the end user experience is fairly similar in both networks. Hence the MRSL threshold can be set to a very high value or the QSC parameter could be set so that the UE never attempt to perform the G2U handover. The drawback of doing this is that the user does not have the opportunity to access the multi-bearer service while in 2G.

5.2

U2G using CPICH Ec/No When using CPICH Ec/No as measurement quantity an increase in downlink interference will be reflected on the value. Although an increase in uplink interference will not be reflected on the CPICH Ec/No value, it can be assumed that there is a correlation between the uplink and downlink interference levels and rather more interference in the downlink. The problematic location type for the CPICH Ec/No measurement quantity is the environment where the moving out of coverage (in terms of pathloss) when the interference level is low. In such environments the CPICH Ec/No remains relatively good until the thermal noise becomes the dominant part of the RSSI. At that point the CPICH Ec/No start to degrade quicker to finally drop heavily when the CPICH RSCP is below the sensitivity of the receiver. At those levels the UE might be transmitting near its maximum Tx power and the connection is not very reliable. Whether or not the IRAT handover is successful at such a location is dependent on how fast the quality is degrading, it might very well be so that a slow moving UE manage to handover to GSM whereas a fast moving UE fails due to that the connection drops before the end of the message flow. If choosing CPICH Ec/No as the measurement quantity for the triggering it is easy to achieve a balance and hence avoid ping-pong handovers between 3G and 2G since Ec/No is used both ways. Due to some mobile limitations, ping-pong will occur even when using CPICH Ec/No in case a high value (such as –11 dB) is used for the event 3a triggering.


39 (73)

No.


5.3

Checked

Date

Rev

2006-03-286

PA1

Reference

G2U Handover (Ec/No and RSCP) If CPICH Ec/No is used for the handover triggering the parameter MRSL which controls the G2U handover should be set to a value approximately 1 dB higher than the value used for the event 2f triggering. This is to avoid ping-pong behaviour, which will increase the risk for dropped calls during a later handover attempt. If on the other hand CPICH RSCP is used for the triggering it is recommended to either set the parameter QSC to never perform any measurements (15). In this case the UE will remain in the 2G network until the call is terminated and the idle mode parameters evaluates on which cell the UE should camp. The drawback with staying with a voice call in 2G as opposed to 3G network is the lack of opportunity to use the multi-bearer service (Voice +PS).

5.4

Idle mode behaviour (CR U2G and G2U) If the Reselection from UMTS to GSM is triggered too early, this might force the UE to leave the 3G network for GSM even if the WCDMA coverage is sufficient enough to provide 3G services. This leads the end user to stay unnecessary time in a technology that is limiting his opportunities. On the other hand, leaving UMTS too late can increase the risk to stay in a network that is not able to handle end user expectations anymore (Call set-ups could be blocked/dropped). The GSM to UMTS cell reselection settings should be consistent with the U2G settings in order to support the best usage of both technologies. Since the main strategy is to maintain an end-user as long as possible into the 3G network, it’s advisable to set the CR parameters with a gap of at least 2 dB between themselves. If the value for the 3G network is set at e.g. –14dB to leave towards GSM a suitable value to return should be –12dB Ec/No.

5.5

Size of neighbour list When defining the neighbour lists one has to take into account;. the list should be as short as possible to increase the chance of a successful handover and the cell coverage might change with the increase of load. However one has to keep in mind that CPICH Ec/No is used as measurement quantity for both idle mode and G2U active mode, although it might be of less importance to always reselect to the strongest cell in idle mode.


40 (73)

No.


5.6

Checked

Date

Rev

2006-03-286

PA1

Reference

U2G triggering: RSCP or Ec/No The choice between the CPICH RSCP triggering criteria and the CPICH Ec/No criteria is about deciding what kind of pros and cons are most important for the network. For the tested areas both CPICH RSCP with a threshold of –104dBm and CPICH Ec/No with a threshold of –13dB works well. However at certain locations one of the measurement quantities performs better than the other and vice versa. To be able to safely let users handover to the 2G network when the quality is degraded due to an increased interference situation the CPICH Ec/No has to be used as a measurement quantity. However to have relatively safe handovers when moving out of coverage in a low load situation, a high value of Ec/No has to be used. A CPICH Ec/No of –11dB for the event 3a leads to secure Handovers to the 2G network in most places tested. The biggest drawback with this value is the reduced 3G coverage since the user will leave the 3G network for the 2G network at levels where the connection can be remained with good quality if the signal strength is sufficient. The table below sums the pros and cons with the two different measurement quantities assuming the threshold for event 3a set to –104 dBm for the RSCP case and –11 dB for the Ec/No case.

CPICH Ec/No Triggering

Pros

Ensures that users experiencing Stable at low and medium interference high interference are handed over scenarios to 2G. Balance the U2G and G2U cell change with the FDDQMIN More time in 3G at medium interference parameter (avoid ping-pong). High threshold needed for stable handovers when going out of coverage.

Cons

5.7

CPICH RSCP triggering

Reduced coverage in high signal strength - medium interference areas.

Does not ensure that users experiencing high interference are handed over to 2G. (i. High Rise buildings, future load) Ec/No used for G2U for PS (and CS) potential ping pong problems. Slightly reduced coverage in low signalstrength - low interference areas.

Huawei recommendation In order to ensure a high quality in the network at all locations Huawei initial recommendation for the IRAT parameter setting in South Africa is to u se the CPICH Ec/No measurement quantity for the triggering as a long-term solution.


41 (73)

No.


Checked

Date

Rev

2006-03-286

PA1

Reference

It’s also possible to use as an intermediate solution the RSCP measurement quantity in isolated and low interfered areas. This indication is given because the RSCP measurement quantity in not really affected by interference problems due to UL or DL load, and because actually the network is not fully loaded. The biggest drawback that can be seen using this strategy is that after a certain amount of load in the network, the entire parameter settings initially “tuned” on the RSCP basis, need then to be reviewed on a CPICH EC/No level.

The recommended values are set as to ensure a high probability of successful U2G handovers both when moving out of coverage in coverage limited scenario and when experiencing a high level of interference.

5.8

Further studies During this service the investigation was focused only on the IRAT HO parameters that have a major impact in the network performance. Below is a list of suggested topics to study: Indoor IRAT HO behaviour CM starts triggered with UE Tx power (ueTxPowerThresh6a/ ueTxPowerThresh6b) GSM vs. UMTS traffic load handling made by ISHOLEV parameter It is suggested that more than one type of TEMS investigation terminal are used for further testing.


42 (73)

No.


6

Checked

Date

Rev

2006-03-286

PA1

Reference

Proposed parameter settings This section contains the final proposed parameter settings that were evaluated according to the results obtained from the Validation phase. The lists below show the proposed settings both if the Ec/No and if RSCP quantity are measured. In both cases, the proposed values for Idle Mode behaviour are added.

CPICH Ec/No         

 

CPICH RSCP

utranThresh3aEcno = -13 usedFreqThresh2dEcno = -12 usedFreqRelThresh2fEcno = 2 (-10) timeToTrigger2d = 11 (320 ms) timeToTrigger2f = 13 (1280 ms) QSC = 8 (above -78dBm) QSI = 7 (always) MRSL = 29 (-9,5) FDDMRR = 2



sRatSearch = 4 (-14*) FDDQMIN = 7 (-12)



Assumed that qQualMin = -18

       



utranThresh3aRscp = -104 usedFreqThresh2dRscp = -102 usedFreqRelThresh2fRscp = 3 (-99) timeToTrigger2d = 11 (320 ms) timeToTrigger2f = 13 (1280 ms) QSC = 8 (above -78dBm) QSI = 7 (always) MRSL = 29 (-9,5) FDDMRR = 2

sRatSearch = 4 (-14*) FDDQMIN = 7 (-12)

bold = Huawei default/recommended value

The proposed parameter settings are not the only possible variation. In border cells areas it’s possible to start CM at higher values because e2d is triggered on a cell basis as well as the sRatSearch parameter. This could give the UE the needed time to handover in time. An additional offset could be added for e2d (e.g. 1dB for Ec/No or 2dBm for RSCP). If the users in CM are reaching a certain level per cell (e.g. 30%), the Admission Control Algorithm will deny the access to that cell. In fact Compressed mode has a higher probability to be blocked by Admission Control or by RBS CE limitations -> Requested Soft HO during CM could be rejected, this might lead to a dropped call (release connection offset or RF synch lost). During compressed mode the usage of channel elements (CE) is higher as well the transmitted power from the UE. In case of using the RSCP strategy it is advisable to monitor the load in the system to detect the point of switching back to Ec/No. A drawback is as well that tuning activities

Ericsson Internal TECHNICAL REPORT Prepared (also subject responsible if other)

43 (73)

No.

EDD/IC/GA Michael Weber TEI/I/RFN Pierluigi Reda,

ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

during the RSCP phase may need to be re-done due to the different characteristics of the measurement quantities.


44 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

Some rule to decide which measurement quantity could be used to trigger IRAT HO are: Low load = RSCP Medium to high Load = Ec/No Low to medium interference = RSCP Medium to high interference = Ec/No Isolated area = RSCP High site density = Ec/No Border cells = RSCP Dense urban area = Ec/No A possible separating for measurement quantity of the area for MTM South Africa could be:

Ec/No Strategy

RSCP Strategy

Pretoria

Durban

Randburg

Cape Town

Germiston Bloemfontein (belongs to GERNC)



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

7

8

45 (73)

No.

Checked

Abbreviations GPI

Global Parameter Investigation

U2G

UMTS to GSM

G2U

GSM to UMTS

CM

Compressed Mode

CR

Cell Reselection

IRAT HO

Inter Radio Access Technology Handover

IRAT CR

Inter Radio Access Technology Cell Reselection

References [1] 3GPP TS25.133 V4.15.0 (2005-06) (Release 1999) [2] User Description, Handover 75/1551-HSD10102/1 [3] 3GPP TS 05.08 V8.17.0 (2003-06)

Reference


46 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

9

Appendix A - Initial phase results

9.1

Overview

Reference

The main scope of the Initial phase of the GPI service was to point out the 3G and 2G network parameters to test and to be investigated. Each parameter was chosen accordingly to the major effects that a value change could have on both systems. Consequently, the first step was to define two different strategies concerning Ec/No and RSCP measured quantities for IRAT HO event triggering. In addition to this, a separate strategy was arranged regarding the Idle Mode behaviour (U2G and G2U Cell reselection). As a staring point, a certain number of Test Cases groups were created, in order to test each single parameter (or sometimes a small set) both from RSCP and from Ec/No point of view. The main scope was to figure out what could have been the best behaving parameter values for the next implementation inside the validation phase test cases. Two different mobiles were used all along the drive tests (TEMS Nokia 6630 and TEMS SonyHuawei V800) in order to collect sufficient data for the analysis.

9.2

Test Case Group (U2G): The U2G HO analysis was conducted on several parameters that have a major impact on the IRAT U2G HO procedure: e2d, e2f, e3a, ttt2d and ttt2f. It was also taken in account the number of ping-pong effects between e2d and e2f, in order to have a complete figure on the system performance according to the different parameter settings. The main objective was to gather all the data coming from the initial phase and to decide which were the best behaving values to use for the validation phase Test Cases. The two tables below (divided by Ec/No and RSCP strategy) show the behaving of e3a triggering during an IRAT HO procedure. The tested parameter was utranThresh3aEcno. The data that were used are referring to the threshold level when the e3a request should be reported by the UE, the average time between e3a itself and the DL HO from UTRAN command as well the average RSCP-Ec/No levels measured for the AS best serving cell when the event occurred.



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Test Case

47 (73)

No.

Meas. Quantity e3a

e3aTo HO CommandTime

Reference

As1EcNo As1RSCP

TC 1.1.1 6630

Ec/No

-12

0,94

-13,40

-107,56

TC 1.1.1 V800

Ec/No

-12

1,17

-13,50

-102,50

TC 1.1.2 6630

Ec/No

-13

2,71

-13,00

-102,90

TC 1.1.2 V800

Ec/No

-13

0,71

-12,50

-98,83

TC 1.1.3 6630

Ec/No

-14

1,38

-12,57

-102,44

TC 1.1.3 V800

Ec/No

-14

0,61

-14,25

-101,50

For the Ec/No strategy (table above) it was evident that the best behaving parameter setting was related to TC 1.1.2 and TC 1.1.3: even if the average time between e3a and the DL HO from UTRAN command is floating, the average measured Ec/No and RSCP levels for the AS best serving cell when the event occurred resulted consistent with the adopted event 3a triggering settings. The TC 1.1.1 settings were excluded due to the significant difference between these evaluated thresholds. Following the same criteria, the analysis was conducted as well on the RSCP related test cases. In this scenario, the best behaving test cases for the validation phase were TC 1.1.7 and 1.1.8.

Test Case

Meas. Quantity

e3a

e3aTo HO Command Time

TC 1.1.6 6630

RSCP

-102

0,91

-11,71

-102,79

TC 1.1.6 V800

RSCP

-102

1,07

-11,80

-100,20

TC 1.1.7 6630

RSCP

-104

1,03

-13,20

-106,22

TC 1.1.7 V800

RSCP

-104

1,06

-11,57

-105,57

TC 1.1.8 6630

RSCP

-105

0,72

-11,75

-106,03

TC 1.1.8 V800

RSCP

-105

0,97

-12,50

-106,50

TC 1.1.9 6630

RSCP

-106

0,90

-10,50

-105,10

TC 1.1.9 V800

RSCP

-106

1,53

-14,50

-104,25

As1EcNo As1RSCP

Regarding the parameter usedFreqRelThresh2f(Ec/No or RSCP), the main analysis was conducted onto the ping-pong effects between the event 2f itself triggering and the event 2d. In this way is possible to have a good figure of the correct triggering of both events, in terms of:


48 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

right start/stop CM during a call e2d reporting from the UE to the system that correctly ends with an IRAT U2G HO incorrect e2d/e2f reporting to the network (this means a misuse of the radio resource by the mobile, since both the events are triggered when not needed). The tables below show the results coming from this analysis, divided again by Ec/No and RSCP strategy and sorted by UE. The ping-pong effect is taken into account if the sequence e2d-e2f-e2d is happening in a time interval shorter than 25 seconds per call. During the measurements a certain number of compressed mode ping-pong was detected, and it gave a first figure which test cases can be considered as the best behaving.

TC 1.1.1 e2f=2 V800 Time Interval (s) 20

6630

TC 1.1.2 e2f=1 V800

6630

TC 1.1.3 e2f=3 V800

6630

Ping-Pong Ping-Pong Ping-Pong Ping-Pong Ping-Pong Ping-Pong 1 0 0 0 0 2

15

2

1

4

1

1

1

10

2

0

1

1

0

1

5

4

0

2

0

0

0

0

5

2

5

2

0

0

Ping Pong ratio

54,22

20,01

52,14

20,11

0,01

0,11

TC 1.1.6 e2f=5 V800

6630

Time Interval (s) Ping-Pong Ping-Pong 1 0 20

TC 1.1.7 e2f=3 V800

6630

TC 1.1.8 e2f=2 V800

6630

TC 1.1.9 e2f=4 V800

6630

Ping-Pong Ping-Pong Ping-Pong Ping-Pong Ping-Pong Ping-Pong 0 0 0 0 0 0

15

1

0

0

0

0

1

0

0

10

0

0

0

2

0

1

0

0

5

1

0

0

0

0

0

1

3

0

1

0

2

1

2

0

0

0

Ping Pong ratio

11,01

0,00

20,00

10,20

20,00

0,11

1,00

3,00

An additional indication came from the analysis on the CM usage, both from Ec/No perspective and from the RSCP point of view. In this case the charts below shows the ratio of total time spent from the UE in compressed mode.


49 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

CM usage (Ec/No) 100 e g a t n e c r e P

80 60 40 20 0 6630

V800

6630

TC 1.1.1

V800

6630

TC 1.1.2

V800

TC 1.1.3

CM ratio with NO HO

CM usage (RSCP) 100 e g a t n e c r e P

80 60 40 20 0 6630

V800

TC 1.1.6

6630

V800

TC 1.1.7

6630

V800

TC 1.1.8

6630

V800

TC 1.1.9

CM ratio with NO HO

It’s important to know that in this scenario a significant role is played also by the UE’s behaviour, because the entire “CM start/stop” or “CM start / IRAT HO performing” procedures are depending from the different UE capabilities.

According to the explained methodology, for the validation phase the value for the parameter usedFreqRelThresh2fEc/No was chosen to 2; the corresponding one for RSCP was indicated in 2 and 3. For the e2d triggering (usedFreqThresh2dEcno/RSCP) the analysis was conducted taking in account the total distribution (Ec/No vs. RSCP) of all the samples collected when an e2d request was reported by the UE. The charts below (Test Cases 1.2 group) show the data coming from these measurements.


50 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

e2d (Ec/No, Thr -11 tested) 0

-20

-11 dB -40 P C S R

-60

-80

-100

-120 -25

-20

-15

-10

-5

0

Ec/No 6630

V800

e2d (Ec/No, Thr -12 tested) 0

-20

-40 P C S R

-60

-80

-100

-120 -20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

Ec/No 6630

V800

From the charts above it’s evident that both of the t ested thresholds are consistent with a correct e2d triggering, because the e2d reporting itself is happening in most of the cases at the predefined Ec/No levels, so both the values –11dB and –12dB for Ec/No strategy were adopted for the validation phase. Also in this scenario it’s very clear the different UE behaviour, and the V800 showed a better performance than the 6630.


51 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

The following charts are showing the same data related to the RSCP strategy, and the results are equal to the previous ones, so the two thresholds (-102dBm and –100dBm) for RSCP e2d triggering were chosen for the validation phase. Even if the event 2d reporting is not always happening at the adopted RSCP level, in most of the cases the corresponding Ec/No values are consistent. e2d (RSCP, Thr -100 tested) -99 -100 -101 -102 P C-103 S R

-104 -105 -106 -107 -20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

Ec/No 6630

V800

e2d (RSCP, Thr -102 tested) -100,5 -101,0 -101,5 -102,0 -102,5 P C-103,0 S R

-103,5 -104,0 -104,5 -105,0 -105,5 -25

-20

-15

-10

Ec/No 6630

V800

-5

0


52 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

For the parameters ttt2d the evaluation was conducted matching the data coming from the different parameter setting values and the point (RSCP vs. EC/No) in which the e2d itself was reported by the UE. From the measurements analysis it was evident that even if the ttt2d parameter values are changing, most of the e2d events are reported in the same radio conditions. This was a clear indication that the radio environment is changing very slowly, so the different values of ttt2d don’t have a real impact on the UE behaviour. Due to this fact, in the validation phase the ttt2d was not tested. The chart below shows the figures coming from the analysis.

ttt2d analysis -60 -70 -80 P -90 C S R-100

-110 -120 -130 -25

-20

-15

-10

-5

0

Ec/No ttt2d=11

ttt2d=12

ttt2d=13

For the parameter ttt2f the analysis was conducted by evaluating the total time spent by the UE in CM, the total time in CM without an IRAT HO, the total time in CM with a successful IRAT HO, the average time in CM and the call time rate spent in UMTS. By comparing these four items the output is that if the ttt2f value is too short, the UE is spending much more time in CM before leaving the GSM measurements for the WCDMA network only: in fact the mobile is much more affected by ping-pong effects. Therefore by this point of view, the possibility to spend a longer time in CM without IRAT HO is increasing. Consequently, an improper usage of the power and radio resources is happening.


53 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

The chart below shows the compared data coming from Test Cases Group 1.2. 350

100% 90%

300

80% 250

70% 60%

200

50% 150

40% 30%

100

20% 50

10%

0

0% e2d=-11 ttt2f=12

e2d=-12 ttt2f=14

e2d=-11 ttt2f=12

V800

e2d=-12 ttt2f=14 6630

e2d=-102 e3a=-106 ttt2f=12

e2d=-102 e3a=-104 ttt2f=14 V800

Ec/No TtlTimeInCompMode

TtlTimeInCompModeWithHO

e2d=-102 e3a=-106 ttt2f=12

e2d=-102 e3a=-104 ttt2f=14 6630

RSCP TtlTimeInCompModeWithoutHO

AvTimeInCompMode

CallTimeUMTS rate

As can be seen in the chart above, a value for ttt2f set to 12 (640 ms) leads the UE to a very high usage of CM, and the total time spent in CM without HO is increasing as well. This is happening both in Ec/No and RSCP Test Cases, even if the at lower values in the RSCP scenario. The opposite situation was detected when the ttt2f value was set at 14 (2560 ms), so it was decided to adopt for the validation phase the values 12 and 13 for the Ec/No strategy, and 13 for the RSCP scenario.

9.3

Test Case Group (G2U):

9.3.1

QSC and MRSL For the G2U IRAT HO analysis two Test Cases Groups (2.1.x and 2.2.x) were arranged in order to test the two main parameter that are involved in the G2U HO event itself: QSC (Starting the WCDMA measurements) and MRSL (Command for G2U HO). Both of them can deeply influence the GSM technology usage, in terms of giving the priority to 2G or 3G according to different parameter settings. Due to this fact, for this analysis three items were considered: percentage of call time in GSM, percentage of call time in UMTS and number of successful G2U handovers.


54 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

The chart below shows the output coming from the measurements for all four driven test cases (2.1.1 and 2.1.2 for MRSL, 2.2.1 and 2.2.2 for QSC). 100

8 7

90

7 6

80

6

6 70

5

5

60

4

50

4

40

3

30 2 20

1

1

1

10

0

0

0 V800

6630

V800

6630

V800

6630

6630

TC 2.1.1

TC 2.1.2

MRSL = 30

MRSL = 34

QSC = 9

QSC = 11

QSC = 7

QSC = 7

MRSL = 32

MRSL = 32

% Call Time GSM

TC 2.2.1

V800

% Call Time UMTS

TC 2.2.2

G2U HO successful

In TC 2.1.1 and 2.1.2 the QSC is set to 7 (always measuring WCDMA), and increasing the value of MRSL from 30 to 34 a decreasing number of G2U handovers can be counted, even if the technology usage is varying. This means that increasing the value of MRSL the point of handing over is moved to upper Ec/No values because the threshold for MRSL is becoming stricter. Moreover, in Test Cases group 2.2.1 and 2.2.2 for QSC testing, the MRSL value was kept at 32, while the value for QSC was changed from 9 to 11. This change shows that at a certain strict threshold for MRSL (32), if the QSC is set to even stricter values (9 = starting measuring UMTS neighbour if the GSM signal is above –74dB, 11 = starting measuring UMTS neighbour If the GSM signal is above –66 dB), the number of G2U handovers is decreasing since the handover point itself is moved to a higher threshold. This setting is only usable if the GSM and UMTS sites are co-located, so that a good UMTS signal is expected if the GSM signal level is good. 9.3.2

FDDMRR The parameter tested in this group is responsible for the number of UMTS cell that could be reported in one “Measurement Report”. The total number of reported neighbour cells is 6 per report. The number of cell that can be reported is sent in the Measurement information towards the UE after the QSC value is reached. The analysis is based on the number of “HOtoUTRAN”, “HOCompleteG2U” and the time between starting measuring UMTS until the HOtoUTRAN.


55 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

9

80 72

8

70

67,3 7 58 6 O H 5 U 2 G o 4 N

60 50

44 40

37

30

28

3

s d n o c e S

20

2

10

1 0

0 V800

6630

FDDMRR 1 HOtoUTRAN

V800

6630 FDDMRR 2

HOCompleteG2U

V800

6630 FDDMRR 3

avTimeFromFDDMRRmeasToG2U_HO (s)

The figure above shows an increase of HO from GSM to UMTS and the average HO time is decreasing. This leads to the conclusion that a higher value of FDDMRR increases the speed and the number of HO. The drawback is that with an increase of UMTS cell reported in the “Measurement Report” the chance of connection losses in GSM is increasing. For the validation phase FDDMRR set to 2 and 3 should be tested.


56 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

10

Appendix B - Validation phase results

10.1

Overview

Reference

In the validation phase all the outputs coming from the initial phase were collected and gathered together in order to define three different groups of test cases: RSCP, Ec/No and Idle mode tests. By analysing the results of the initial phase, all the investigated and tested parameters were combined in different ways in order to obtain different scenarios. All the combinations were defined according to the outputs of the initial phase; only the best behaving were chosen, instead the parameters that didn’t have a real effect on the system with different values were excluded as well as all the parameters whose values resulted too strict. After the analysis of the validation test cases, all the best behaving parameters were gathered in two separated strategies, and the data coming from the Idle Mode behaviour measurements were added accordingly: the main scope of this activity was to define the two main strategies (RSCP and Ec/No) for the final parameter settings proposal. A TEMS Motorola A835 was added to gather more data for the analysis and provide a wider spectrum of UE behaviour.

10.2

Ec/No Strategy The charts below showing the results of the analysis from the Ec/No strategy collected during the validation phase. Compressed Mode ping-pong 14

e2d=-11 e2f=2 ttt2f=13 e3a= -13

e2d=-12 e2f=2 ttt2f=13 e3a= -13

e2d=-12 e2f=2 ttt2f=12 e3a= -13

12 10

4

8

6

6

4

3

4 2 0

3 2 V800

6630

1

1 2

A835

V800

VTC 1.1

4 4

1 1

2 1 1

6630

A835

20 sec

15 sec

4

1

2 1

1

V800

6630

A835

VTC 1.2 25 sec

1 1 1

VTC 1.3 10 sec

5 sec

The chart above illustrates the number of CM starts and stops and weights these by time between the CM stop and the next CM start during a call. VTC 1.1 and 1.2 point up the lowest numbers of CM ping-pong within 5 sec. The parameters setting of VTC 1.3 specially the ttt2f seems to lead to a higher ping-pong ratio.


57 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

Compressed Mode usage 35

30

25

20

15

10

5

0 6630

V800

A835

6630

V800

VTC1.1

A835

6630

VTC1.2 TtlStartCM

TtlStopCM

HOfromUTRAN

V800

A835

VTC1.3 HOtoUTRAN

The number of CM start together with the number of HOfromUTRAN indicates the correct usage of CM. As it can be seen above the number of CM starts/stop of VTC1.2 are the lowest and the usage of CM to handover to GSM is in a good ratio.

Time in Compressed Mode 300

250

200

c e 150 s

100

50

0 6630

V800

A835

6630

VTC1.1 TtlTimeInCompMode

V800

A835

6630

VTC1.2 TtlTimeInCompModeWithHO

TtlTimeInCompModeWithoutHO

V800

A835

VTC1.3 AvTimeInCompMode

The time that the UE is spending in CM is crucial due to the fact that the UE Tx power is increased during this time, the soft HO procedure is slower and PS session will be switched down to PS64 to be able to perform a handover. It can be seen in the diagram above that the total time in CM and the related time with HO to GSM for VTC1.2 is the lowest one of all test cases in this group. Together with the impression of the chart “Compressed Mode ping -pong” above the setting from VTC 1.2 performs best.


58 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

Technology distribution 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 6630

V800

A835

6630

VTC1.1

V800

A835

6630

VTC1.2 CallTimeUMTS ratio

V800

A835

VTC1.3

CallTimeGSM ratio

The technology distribution between UMTS and GSM will give an indication about the end user perception and the possibility to use the enhanced services of UMTS. As it can be seen from the diagram above the distribution of technologies in VTC 1.2 and 1.3 give a priority towards the UMTS system.

VTC1.1 totNoDrops

VTC1.2

VTC1.3

6630

V800

A835

6630

V800

A835

6630

2

2

2

1

1

2

1

1

1

1

CM Drop

2 2

2

A835

1

Missing NB GSM

V800

1

1

1

The table above lists the number of drops that occurs during the test case. The drop call reasons are different most of the times. As example the Motorola A835 dropped twice during compressed mode in VTC1.1. The reason for that was it could not perform a soft HO in UMTS and so the call dropped (e1d). This type of drops happened only with the Motorola and this might be related to the older SW/HW version, RLT timer expiring has been seen as well and drops due to missing neighbours. Site U0008 came on air without neighbour cell relations. The UE’s could manage to HO to GSM due to the decrease of Ec/No but as soon the UE was in GSM U0008 reached the criteria for HO back to UMTS. The UE handed over and dropped afterwa rds due to the “releaseConnOffset”. The UMTS parameter setting from VTC 1.2 seems to be the best performing ones and should taking into account for the final proposed settings. VTC1.2 utranThresh3aEcno = -13  usedFreqThresh2dEcno = -12   usedFreqRelThresh2fEcno = 2 (-10) timeToTrigger2d = 11 (320 ms)   timeToTrigger2f = 13 (1280 ms)


59 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

100

14 13

90

12 80 10

70 9 60

8

8

50 40

7 6

6

6

6 5

30

4 3

20

2 10 0

0 A835

6630

V800

A835

VTC 1.1 (QSC= 7, MRSL= 32)

6630

V800

A835

VTC 1.2 (QSC= 8, MRSL= 30)

3G call time %

2G call time %

6630

V800

VTC 1.3 (QSC= 9, MRSL= 29)

HOCompleteG2U

The chart above gives an impression about the technology distribution together with the IRAT HO G2U performance. The G2U IRAT HO is mainly triggered from QSC (starting measuring UMTS neighbours) collectively with MRSL that give the HO command to UTRAN. These two parameters can priories the usage of GSM or UMTS from the GSM point of view. The settings from VTC1.2 and 1.3 give a priority towards UMTS. A combination of both test cases should be considered for the final proposal.

10

51,8 43,0

9

43,2

38,0

8 7

44,6

33,8 34,6

30,4

6

24,9

5

25,9

4 17,3

3 2

08,6

1 0

00,0 V800

6630

VTC 1.1 FDDMRR=2 G2U HO

V800

6630

VTC 1.2 FDDMRR=3

G2U HO COMPL

V800

6630

VTC 1.3 FDDMRR=3

AVERAGE TIME FDDMRR MEAS-G2U HO

FDDMRR is reflecting the max. Number of UMTS cells reported into a measurement report. A high number of cell reported in one Measurement Report decreases the time of selecting the best UMTS cell to HO, but on the other hand a higher number for UMTS cells leaves less for GSM. The max. Number of cell that can be reported into one report is 6 so that this can increase the risk to lose the call in GSM. A suitable solution could be to divide the max. Number of cell per Report by the number of available Frequency Bands (900/1800/2100). This would lean to a possible value of 2 or 3 depending of the area.



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

10.3

60 (73)

No.

Checked

Reference

RSCP Strategy The charts below showing the results of the analysis from the RSCP strategy collected during the validation phase Compressed Mode ping-pong 9 8

e2d=-100 e2f=2 ttt2f=13 e3a =-104

e2d=-102 e2f=3 ttt2f=13 e3a =-104

e2d=-102 e2f=3 ttt2f=13 e3a =-105

7 6

4

5 4

1

3

2

2 1

2 2

2

0 V800

2

1 6630

A835

V800

VTC 2.1

2

1

1

6630

A835

V800

VTC 2.2

25 sec

20 sec

15 sec

3

1

2

1

6630

A835

VTC 2.3 10 sec

5 sec

The best performing test case is VTC2.2 due to the reason the this has the lowest total number of ping-pong effects under 25 seconds and the least number of CM ping-pong within 5 sec. It seems that the lower value of event 2f from VTC2.1 leads to an increase of the ping-pong effect.

Compressed Mode usage 30

25

20

15

10

5

0 6630

V800

A835

6630

V800

VTC2.1

A835

VTC2.2 TtlStartCM

TtlStopCM

HOfromUTRAN

6630

V800

A835

VTC2.3 HOtoUTRAN

The figures from the chart above strengthen the impression from the “Compressed Mode ping-pong” chart. VTC 2.2 has a low number of CM starts/stops with a suitable number of HO to GSM. Furthermore is can be seen that the behaviour of the UE’s doesn’t follow a global trend anymore.


61 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

Time in Compressed Mode 160

140

120

100 c e s

80

60

40

20

0 6630

V800

A835

6630

VTC2.1

V800

A835

6630

VTC2.2

TtlTimeInCompMode

TtlTimeInCompModeWithHO

V800

A835

VTC2.3

TtlTimeInCompModeWit houtHO

AvTimeInCompMode

The intuition from the chart before compared with the one above shows that it is not possible to predict a global trend. The performance of the SE V800 seems to be consistent but the Nokia 6630 and Motorola A835 showing the opposite behaviour. This might be correlated with the fast floating of RSCP value and the UE Rack receiver performing.

Technology distribution 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 6630

V800 VTC2.1

A835

6630

V800

A835

VTC2.2 CallTimeUMTS rate

6630

V800

A835

VTC2.3

CallTimeGSM rate

The technology distribution chart shows that the Nokia 6630 is staying more time into GSM and the change of system settings in the different test case are not able to turn this around. A similar effect can be seen for the Motorola A835 the different is that this UE is staying more time in UMTS. The V800 seems to be the only UE that is changing the behaviour according the system settings. On reason for that is that the used measurement quantity to leave UMTS is RSCP and the quantity to return is based on Ec/No.


62 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

VTC2.1 6630

V800

totNoDrops

Reference

VTC2.2 A835

6630

2

VTC2.3

V800

A835

6630

V800

A835

1

1

2

1

3

1

1

2

1

3

CM Drop Missing NB GSM

2

The table above is listing the dropped calls that occur during the drive test. A higher number of drops can be seen for missing neighbour. The reason for this is slightly different as to the one in the VTC1.X group. In this case the UE is loosing the connection due to “releaseConnOffset” straight. The quality decreases but the coverage level is on a stable level until the new cell becomes strongest. 100

14

90 12 80

60

10

10

70

8

8

50 6

40

6

5 30

4

4

4

4

4

3

20

2 10 0

0 A835

6630

V800

VTC 2.1 (QSC= 7, MRSL= 32)

A835

6630

V800

V TC 2.2 (QSC= 8, MRSL= 30)

3G call time %

2G call time %

A835

6630

V800

V TC 2.3 (QSC= 9, MRSL= 29)

HOCompleteG2U

The settings for G2U handover need to be very deeply correlated with the settings from U2G. The different quantities we are measuring with, increases the risk of a ping-pong effect and this makes it difficult to define a good handover strategy in both directions. The test cases separate doesn’t give a good solution a combination of them is recommended.



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

9

63 (73)

No.

43,2

39,1

38,0

37,0

8

Reference

38,9

33,3 34,6

7

30,2

6 5

25,9

20,6

21,6 4 12,7

3

17,3 13,0

2

08,6

1

04,3

0

00,0 V800

6630

VTC 2.1 FDDMRR=2 G2U HO

V800

6630

VTC 2.2 FDDMRR=3

G2U HO COMPL

V800

6630

VTC 2.3 FDDMRR=3

AVERAGE TIME FDDMRR MEAS-G2U HO

FDDMRR is reflecting the max. Number of UMTS cells reported into a measurement report. A high number of cell reported in one Measurement Report decreases the time of selecting the best UMTS cell to HO, but on the other hand a higher number for UMTS cells leaves less for GSM. The max. Number of cell that can be reported into one report is 6 so that this can increase the risk to lose the call in GSM. A suitable solution could be to divide the max. Number of cell per Report by the number of available Frequency Bands (900/1800/2100). This would lean to a possible value of 2 or 3 depending of the area.

10.4

Idle mode results The following parameters have been used to discover the best setting for MTN South Africa idle mode behaviour. VTC 3.1 VTC 3.2 VTC 3.3

sRATsearch 8 -10 dB 4 -14 dB 2 -16 dB

FDDQMIN 3 -8 dB 5 -10 dB 7 -12 dB

The basic principle of the idle mode cell reselection is to keep the end user as long as possible in the enhanced system to give him the opportunity to used the additional services, but on the other hand the camp too long into a system that is already degraded increases the chance of bad end user experience.


64 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Reference

22 20 18 16 14 12 10 8 6 4 2 0 6630

V800

A835

CR U2G TOT

6630

V800

A835

CR U2G on the Best GSM

VTC 3.1

VTC 3.2

6630

V800

A835

CR U2G not on the Best GSM

VTC 3.3

22 20 18 16 14 12 10 8 6 4 2 0 6630

V800 CR G2U TOT

A835

6630

V800

A835

CR G2U on the Best WCDMA VTC 3.1

VTC 3.2

6630

V800

A835

CR G2U not on the Best WCDMA

VTC 3.3

The idea behind this settings should be the leave 3G at the same level as the event 3a and return to 3G before it starts CM. The compromise that has to be faced is that values for sRatSearch and FDDQMIN are only even values. This leads into the results that a possible CR U2G is done slightly after event e3a and the return G2U at the CM start level. A wider ranch between sRatSearch and FDDQIN decreases the UMTS footprint.



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

11

65 (73)

No.

Checked

Reference

Appendix C - General Parameters The following tables include the list of all parameters involved in the IRAT HO (UMTS to GSM and vice versa).

11.1

3G to 2G HO parameters Parameter utranThresh3aEcno

utranThresh3aRscp

gsmThresh3a

fddGsmHoSupp hysteresis2d

hysteresis2f hysteresis3a measQuant2

Description Threshold for event 3a (the estimated quality of the currently used WCDMA RAN frequency is below a certain threshold and the estimated quality of the GSM system is above a certain threshold in the same time interval) for WCDMA RAN Threshold for event 3a (the estimated quality of the currently used WCDMA RAN frequency is below a certain threshold and the estimated quality of the GSM system is above a certain threshold in the same time interval) for WCDMA RAN Threshold for event 3a (the estimated quality of the currently used UTRAN RAN frequency is below a certain threshold and the estimated quality of the GSM system is above a certain threshold) for GSM. Indicates if the RNC supports Inter-RAT Handover Hysteresis used for event 2d

Range CPICH Ec/No. –24 – 0 dB

Default/Resolution -13 dB/1dB

CPICH Rscp. –115 to -25 dB

-105 dBm/1 dB

GSM carrier RSSI. -115 – 0 dBm

-104 dBm / 1 dBm

0 = off, 1 = on

0/1

0 – 14.5 dB

0 / 0.5 dB

Hysteresis used for event 2f Hysteresis used for event 3a Measurement quantity for connection quality monitoring and reporting evaluation

0 – 14.5 dB 0 – 7.5 dB Enum {CPICH Ec/No, CPICH RSCP}

0 / 0.5 dB 0 / 0.5 dB CPICH Ec/No / N/A



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter usedFreqThresh2dE cno

usedFreqThresh2dR scp

usedFreqRelThresh2 fEcno

usedFreqRelThresh2 fRscp

utranMeasQuant3

66 (73)

No.

Description Threshold for event 2d (the estimated quality of the currently used UTRAN frequency is below a certain threshold). Used if measurement quantity is configured to be Ec/No Threshold for event 2d (the estimated power level of the currently used UTRAN frequency is below a certain threshold). Used if measurement quantity is configured to be Rscp Relative threshold for event 2f versus event 2d. The estimated quality of the currently used WCDMA RAN frequency is above a certain threshold +usedFreqThresh2dEcno. Relative threshold for event 2f versus event 2d. The estimated quality of the currently used WCDMA RAN frequency is above a certain threshold +usedFreqThresh2dRscp. Defines the measurement quantity of the UTRAN quality for inter-RAT Handover evaluation

Reference

Range CPICH Ec/No. -24 – 0 dB

Default/Resolution -11 dB / 1 dB

CPICH Rscp. -115 to -25 dBm

-103 dBm / 1dBm

0 - 20 dB

2dB / 1dB

0 - 20 dB

2dB / 1dB

Enum {CPICH Ec/No, CPICH RSCP}

CPICH Ec/No / N/A



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

11.2

67 (73)

No.

Checked

Reference

2G to 3G HO parameters Parameter SPRIO

Description Search Priority

Range Yes/no

Default / Resolution Yes

On/Off

1(on)

Indicates if 3G cells may be searched when BSIC decoding is required. NO = Multi-RAT MS may not use the search frames required for BSIC decoding, for UTRAN FDD measurements. YES = Multi-RAT MS may use up to 25 search frames per 13 seconds without considering the need for BSIC decoding in these frames. A setting of “No” would not affect measurements of GSM cells. “Yes” needed for Nokia 6650

COEXUMTS

Defined per cell. BSC exchange property. Determine if cell reselection and handover from GSM  UMTS is allowed or not.



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter MFDDARFCN (idle mode)

Description Indicates the absolute RF channel number of the neighbouring UTRAN cell measured on by the MultiRAT MSs MFDDARFCN, the absolute RF channel number of the neighbouring UTRAN cell to be measured by a multi-RAT mobile in GSM idle mode. According to the 3GPP Recommendations, the channels are numbered as follows:

MSCRCODE (idle mode) UMFI (idle mode)

68 (73)

No.

f (n) = 5/n in MHz, where n (MFDDARFCN) goes from 0 to 16383 and f is a frequency of the carrier, downlink. There are 12 frequencies (bandwidth 5MHz) in the UMTS spectrum (21102170MHz), with values of MFDDARFCN from 10550 to 10850. Indicates the scrambling code for the UTRAN neighbouring cell MFDDARFCN-MSCRCODEDIVERSITY Up to 64 UMFIs can be defined in a cell.

Range

Reference

Default / Resolution



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter FDDQMIN

69 (73)

No.

Description Defines the minimum threshold for the “quality” measure Ec/No for cell reselection to UTRAN.

Range

Reference

Default / Resolution 7(-13dB)

The available settings are from 0 – 7 which represents –20 to –13 dB.

FDDQOFF

Per cell parameter and applies to all its UTRAN neighbours. Defines cell reselection offset to UTRAN cells. Available settings are from 0 – 15 which represented – infinity (0) , -28dB (1) to +28dB (15) in steps of 4 dB. A setting of 0 is meant to select UTRAN whenever possible. Per cell parameter and applies to all its UTRAN neighbours

0 (-inf)



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter QSI

70 (73)

No.

Description GSM-UMTS cell reselection quality search indicator.

Range

Reference

Default / Resolution 7(always)

Indicates the threshold to start UTRAN FDD measurements in (a) Idle 1 mode and (b) Stand-by and Ready states Defines if the monitoring of UMTS cells will be performed if the signal is below (0-6) or above the threshold (8-14), always (7) or never (15). 0 -98 dBm 1 -94 dBm ··· 6 -74 dBm 7 always 8 -78 dBm 9 -74 dBm ··· 14 -54 dBm 15 never

COEXUMTSINT

FDDARFCN (active mode)

1

During packet data usage

Per cell parameter Used to control the time interval between traffic load checking. BSC exchange property Indicates the absolute RF channel number of the neighbouring UTRAN cell in the GSM active mode.

1000ms



ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter SCRCODE

UTRANID

FDDMRR

QSC

71 (73)

No.

Description Defines the scrambling code for the neighbouring UTRAN cell in active mode. Consists of RNC identity (RNCID), Cell Identity within RNC (CI), MCC, MNC and LAC RNCID-CI-MCC-MNC-LAC Defines how many measured neighbouring UTRAN cells should be included in measurement reports. The remaining positions in measurement reports will be used for reporting GSM cells according to the parameter MBCR

GSM-UMTS handover quality search indicator. Indicates the threshold to start UTRAN FDD measurements in Active mode. Defines if monitoring of UMTS cells will be performed if the signal level is below (0-6) or above the threshold (8-14), always (7) or never (15). 0 -98 dBm 1 -94 dBm ··· 6 -74 dBm 7 always 8 -78 dBm 9 -74 dBm ··· 14 -54 dBm 15 never

Defined per GSM cell

Reference

Range

Default / Resolution

1-3

1

7 (always)


72 (73)

No.


ESA/SK 05:0144

Approved

Date

Rev

2006-03-28

PA1

Checked

Parameter ISHOLEV

Description Defines the traffic load threshold of the serving GSM cell that needs to be exceeded in order to evaluate UMTS measurements for handover.

MRSL

Minimum Reported Signal Level for CPICH Ec/No UTRAN cell quality criteria. Within the locating algorithm, the measured energy per chip on the Common Pilot Channel (CPICH Ec/No) a UTRAN cell is compared with the minimum quality threshold defined by MRSL. Only if these criteria are fulfilled, handovers to the UTRAN cell are possible. The values are: MRSL < -24 dB 1 -24 dB <= MRSL < -23.5 dB ... 48 -0.5 dB <= MRSL < 0 dB 49 0 <= MRSL Defies a minimum threshold for the “quality”measureEc/No for handovers to UTRAN. To be set the same value as the UTRAN parameter usedFreqRelThresh2fEcno. The value of MRSL shall be set so that a handover to UTRAN is not immediately followed by a ping-pong handover back to GSM. This is achieved if the MRSL parameter has the relation as follows: MRSL > usedFreqThreshold2dEcno + a hysteresis A suitable hysteresis is achieved if MRSL = usedFreqThreshold2fEcno Defined per UTRAN cell

Range

Reference

Default / Resolution 99 (always)

-9 dB

58528996 IRAT HO and Re Selection Tuning SA

Recommend Documents