GRQ Roaming QoS ParametersWorking Document version 0.6

15 November 2007

Security Classification Category (see next page)

Restricted Group Members

1

Security Classification: RESTRICTED Access to and distribution of this document is restricted to the persons listed under the heading Security Classification Category. This document is confidential to the Association and is subject to copyright protection. This document is to be used only for the purposes for which it has been supplied and information contained in it must not be disclosed or in any other way made available, in whole or in part, to persons other than those listed under Security Classification Category without the prior written approval of the Association. The GSM Association (“Association”) makes no representation, warranty or undertaking (express or implied) with respect to and does not accept any responsibility for, and hereby disclaims liability for the accuracy or completeness or timeliness of the information contained in this document. The information contained in this document may be subject to change without prior notice.

Copyright NoticeCopyright © 2007 GSM Association

Antitrust NoticeThe information contained in this document is in full compliance with the GSM Association’s antitrust compliance policy.

Document History

Version Date Brief DescriptionEditor /Company

0.3 7 June 2007Re-written draft document provided by GRQ SPT adhoc team lead by Eduard Jebbink

0.4 15 June 2007 Re-written after SPT meeting #6 in LondonEduard Jebbink

0.5 20 September 2007

Following further work by the GRQ SPT Ad-Hoc team on the grid QoS parameters / Monitoring Methods, specification of the MO or MT contexts when measuring the voice Parameters (# 4, 6 and 8)

Additional proposal for Packet Loss definition (#34)

Changes to SMS parameters (#21 – 24)

Changes to Packed Switched LU success rate and delay (#31 and 32)

Thierry Aubert, Hayato Yoshimi, David Guilot,Luc Vidal, Pradeep Bhardwaj and Eduard Jebbink (editor)

0.6 15 November 2007

Alignment of data KPI names with ETSI

Definition of the customer experience phases

General revision of parameters for SMS and Data to cover both the needs of getting an end-user perspective of QoS and the appropriate parameters for setting up SLAs

Luc Vidal, Thierry Aubert, David Gillot

2

Purpose/Introduction/background

The purpose of this document is to define a set of parameters required to monitor and test data roaming quality of services end-to-end in accordance with the GRQ principles.The document is for internal discussions, and for discussions with IREG. The aim is to have the content of this document incorporated into the IREG PRDs.

For a subset of these parameters, it is important to have a 1 on 1 relationship with the Roaming SLA (PRD AA13, Annex 12), as this is where roaming partners agree on the values of these QoS Parameters. Once operators have agreed on the values, they can start monitoring and measuring the performance. How this can be done will be described in other documents.

However, it is important to keep in mind that the exact definitions of these QoS Parameters, can depend on how they are measured / monitored. So the work on how these QoS Parameters need to be monitored can result in slight modifications of these definitions.

The aim of GRQ is to be able to monitor end-to-end quality of service of roaming. This was not deemed possible. Instead we are aiming for controlling the end-to-end quality by monitoring parts of the service. Together this should give enough control over the service as the customers are experiencing it.

For that reason we have divided the service in 5 logical and chronological stages, and identified QoS Parameters in each of these stages. So the list is created from a customer perspective, but translated into technical items.The 5 phases are listed below along with their definition :

1. Network Accessibility : the customer is registering to the network (either the circuit switched network for voice or the packet switched network for data). It is assumed that the network is available where the customer is located

2. Service Accessibility : the end-to-end bearer connection is provided to the customer. For voice services, the customer hears the ring tone ; for data services, the end-to-end PDP context is activated ; for SMS, the connection is established between the end-user terminal and the SMSC.

3. Connection Establishment : the customer actually accesses the requested service. For voice services, the call is actually set up end-to-end (even in case of call forward to voicemail) ; for data services, the customer sends the MMS or accesses the WAP portal or accesses a Web page, etc….

4. Connection Retain Ability : the customer terminates his voice call or data connection without cut-off

5. Connection quality : the connection is not impaired by quality problem, such as speech quality for voice or data rate for data services, etc…

For Packet Switched services we are only looking on a bearer level, and not at specific services. It is partly a pragmatic approach, but also because we feel there is not enough information at this stage on specific services. There is still not QoS negotiation and management mechanism in place between operators. Everything is based on ‘best-effort’. KPI for Connection Establishment for Packet Switched services are therefore not proposed in the document.

3

In order to do tests, we always use the 3 standard services: MMS, WAP and internet and their specific APN’s.

QoS Parameters1

Voice QoS Parameters

Service stage QoS Parameter

1. Network Accessibility (customer being able to register on the network)

1. Circuit Switched LU success rate2. Circuit Switched LU delay

2. Service accessibility (from customer hitting the send bottom until hearing a ring tone)

3. NER-MO or SAT-MO (success ratio)4. NER-MT or SAT-MT (success ratio)5. PDD-MO or STT-MO (duration)6. PDD-MT or STT-MT (duration)

3. Connection establishment (from customer hitting the send bottom until a successful establishment of the call)

7. CSSR-MO (success ratio)8. CSSR-MT (success ratio)9. REL (ISUPv2)10. OCN & RDN (ISUPv2)

4. Connection retain ability (from a successful establishment of the call until service is terminated)

11. CCR (success ratio)12. ALOC (duration)2

5. Connection quality13.CLI delivery14.SpQ (Speech Quality)

SMS QoS Parameters

Service stage QoS Parameter

1. Network Accessibility (customer being able to register on the network)

No QoS Parameter (as not SMS specific)

2. Service accessibility (MO) 21. Service Accessibility for SMS-MO22. Service Accessibility for SMS-MT

1 Unless explicitly specified, the following Parameters are related to Mobile Originated (MO) calls from the VPMN.

2 Not relevant for a roaming QoS SLA.

4

23. Access Delay for SMS-MO24. Access Delay for SMS-MT25. End-to-End Delivery Time for SMS-MOError: Reference source not found26. End-to-End Delivery Time for SMS-MTError: Reference source not found

3. Connection establishment (from customer hitting the send bottom until a successful delivery of the SMS)

No QoS parameter (Store & Forward Mechanism)

4. Connection retain ability (from a successful establishment of the service until service is terminated)

No QoS Parameter (Store & Forward Mechanism)

5. Connection qualityNo QoS Parameter (Store & Forward Mechanism)

Data QoS Parameters

Service stage QoS Parameters

1. Network Accessibility (customer being able to register on the network)

31. Packet Switched Location Update success rate32. Packet Switched Location Update Delay

2. Service accessibility (from customer hitting the send / connection bottom until accessing the data bearer)

33. PDP Context Activation success rate34. PDP Context Activation time

3. Connection establishment (from customer hitting the send bottom until a successful establishment of the service)

No QoS Parameter for the moment. Once QoS negotiation procedures between home and serving networks will come in place, this can be addressed in this section. For the moment nothing like connection establishment exists in data roaming.

4. Connection retain ability (from a successful establishment of the service until service is terminated)

35. PDP Context Cut-Off Ratio36. PDP Context Average Session TimeError: Reference source not found.

5. Connection quality37. Throughput (Kbit/sec) 3

38. Goodput (Kbit/sec)39. Roundtrip time (expressed in miliseconds) Error:

Reference source not found 40. Packet loss Error: Reference source not found

3 Because a ‘bearer level’ approach adopted for the data services QoS monitoring, this indicator has to be measured at the ‘bit pipe’ level and not at the User Application level (e.g. HTTP).

5

QoS Parameters definitions

Circuit switched service1. Circuit Switched LU Success Rate (CS LU – SR)

Abstract definition

The CSLU success ratio describes the probability that a subscriber can successfully

attach to the CS network.

Computation

Formula:

Trigger points:

Start:

Mobile sends the CS attach request message.

Stop:

Mobile receives the CS attach accept message.

2. Circuit Switched Location Update Delay (CS LU - D)

Abstract Definition

This CSLU delay describes the time period needed to attach to the CS network.

6

Computation

Formula:

Trigger points (for the computation of the unit CSLU delay):

Start:

Point of time when the mobile sends the attach request message.

Stop:

Point of time when the mobile receives the attach accept message.

Remarks:

The difference between an attach of a known subscriber and an unknown

subscriber will be reflected in the time period indicating the attach setup time. In

case of an unknown subscriber (meaning that the MSC has changed since the

detach, or if it is the very first attach of the mobile to the network), the MSC

contacts the HLR in order to receive the subscriber data. The attach setup time

of an unknown subscriber will be slightly longer than the one of a known

subscriber.

While determining the average attach setup time only successful attach attempts

are included in the calculations.

7

3. Service Accessibility Telephony - MO (SA-T-MO)

Abstract definition

Probability that the end-customer can access the Mobile Telephony Service when

requested if it is offered by display of the network indicator on the Mobile Equipment.

Term NER (Network Effectiveness Ratio defined in ITU-E 425) can be understood as

Service Accessibility Telephony.

Computation

There are two possibilities for a successful call attempt:

The customer hears the alerting

B-party is busy

Abstract formula:

Trigger points:

Beginning of the call attempt:

successful pressing send button (it is important to check, if coverage has been given

when send button is pressed, otherwise this Call Attempt counts to Network Non

Accessibility (NNA)).

Successful call attempt:

connect measurement (e.g. alerting or busy heard by A- party).

Remarks:

SAT/NER is not catching the voice carriers that are sending fake ring tones like sending

fake ACM/CPG messages.

8

4. Service Accessibility Telephony - MT (SA-T-MT)

Abstract definition

Probability that the end-customer can access the Mobile Telephony Service when

requested if it is offered by display of the network indicator on the Mobile Equipment.

Term NER (Network Effectiveness Ratio defined in ITU-E 425) can be understood as

Service Accessibility Telephony.

Computation

There are two possibilities for a successful call attempt:

The customer hears the alerting

B-party is busy

It is assumed that the A Party is in the home network in order to avoid an international

tromboning of the call between the VPMN and the HPMN.

Abstract formula:

Trigger points:

Beginning of the call attempt:

successful pressing send button (it is important to check, if coverage has been given

when send button is pressed, otherwise this Call Attempt counts to Network Non

Accessibility (NNA)).

9

Successful call attempt:

connect measurement (e.g. alerting or busy heard by A- party).

5. Setup Time Telephony – MO (ST-T-MO)

Abstract definition

Time between sending of complete address information and receipt of call set-up

notification.

ITU-E 431 PDD (Post Dialling Delay) can be understood as ST-T.

Computation

Abstract formula:

t2: point of time where connect is established (e.g. alerting4 or subscriber busy is

detected by test equipment))

t1: point of time where the customer presses the send button on mobile equipment

Trigger points:

Beginning of the Setup Time measurement:

successful pressing send button (it is important to check, if coverage has been given,

otherwise this Call Attempt counts to Network Non Accessibility (NNA))

Successful connection:

connect measurement (e.g. alerting or busy heard by A- party)

4 [If you don’t establish an end to end connection afterwards you must ignore this measurement.] It is assumed that early traffic channel assignment is used.

10

6. Setup Time Telephony – MT (ST-T-MT)

Abstract definition

Time between sending of complete address information and receipt of call set-up

notification.

ITU-E 431 PDD (Post Dialling Delay) can be understood as ST-T.

Computation

It is assumed that the A Party is in the home network in order to avoid an international tromboning of the call between the VPMN and the HPMN

Abstract formula:

t2: point of time where connect is established (e.g. alerting5 or subscriber busy is

detected by test equipment))

t1: point of time where the customer presses the send button on mobile equipment

Trigger points:

Beginning of the Setup Time measurement:

successful pressing send button (it is important to check, if coverage has been given,

otherwise this Call Attempt counts to Network Non Accessibility (NNA))

Successful connection:

connect measurement (e.g. alerting or busy heard by A- party)

5 [If you don’t establish an end to end connection afterwards you must ignore this measurement.] It is assumed that early traffic channel assignment is used.

11

7. CSSR - MO (Call Setup Success Ratio)

Abstract definition

CSSR expresses the relationship between the number of seizures and the sum of the

number of seizures resulting in a successful established call.

Computation

Abstract formula:

Trigger points:

Beginning of the call attempt: successful pressing send button (it is important to

check, if coverage has been given when send button is pressed, otherwise this

Call Attempt counts to Network Non Accessibility (NNA)).

Successful call establishment: open connection between A-party and B-party.

8. CSSR – MT (Call Setup Success Ratio)

Abstract definition

CSSR expresses the relationship between the number of seizures and the sum of the

number of seizures resulting in a successful established call.

Computation

It is assumed that the A Party is in the home network in order to avoid an international

tromboning of the call between the VPMN and the HPMN.

Abstract formula:

12

Trigger points:

Beginning of the call attempt:

successful pressing send button (it is important to check, if coverage has been

given when send button is pressed, otherwise this Call Attempt counts to Network

Non Accessibility (NNA)).

Successful call establishment:

open connection between A-party and B-party.

9. REL (ISUPv2 signalling transparency)

Abstract definition

Effective uncorrupted transmission by the VPMN of the Cause Value in the Release

(REL) ISUP messages, as defined in ITU-T Q850

Computation

Testing protocol :

the HPMN sends to the VPMN a REL ISUP message with a valid populated ‘Cause Value’ field

the VPMN must send back to the HPMN the REL ISUP message with the same ‘Cause Value’

the HPMN uses SS7 monitoring tool to measure the KPI

Abstract formula:

13

10. OCN & RDN (ISUPv2 signalling transparency)

Abstract definition

Effective uncorrupted transmission by the VPMN of Original Called Number and

Redirecting Number, as defined in ITU-T Q.732.2.

Computation

Using a testing tool:

A, B and C all belong to the HPMN

A is roaming on the VPMN network while B is located in the HPMN

C is a HPMN service number or subscriber located in the HPMN

A late call forward takes places and is not optimally routed (i.e the calls is effectively routed towards the A destination and returned to the HPMN).

Abstract formula:

 

%100 testsForward Call ofNumber

OCN/RDN including Forwards Call ofNumber [%] RDN&OCN

Trigger points:

OCN & RDN are correctly transmitted .

14

11. Call Completion Rate Circuit Switched Telephony (CCR-CS-T)

Abstract definition

Probability that a successful call attempt is maintained for a predetermined time until it is

released intentionally by A- or B-party.

Computation

Abstract formula:

Trigger points:

Successful call attempt:

connect measurement (e.g. ‘alerting’ or ‘busy’ detected by A- party)

Terminated call:

release of connection directly by A- or B-party

Remarks:

This KPI is not indicating the carriers that are sending fake Busy or Normal Unspecified

voice carriers.

12. ALOC (Average Length of a Call)

Abstract Definition

As defined in ITU E-437 and in IN01: average duration of calls. The advice is to

measure this for MOC and MTC separately, as there could be a significant natural

difference between these 2 call types.

15

Computation

Using traffic report:

13. CLI Transparency

Abstract Definition

CLI needs to be delivered correctly and complete, in a way it can be used to dial back

the original calling party.

Computation

14. Speech Quality (SpQ)

Abstract definition

Indicator representing the quantification of the end-to-end speech transmission quality of

the Mobile Telephony Service.

16

Computation

The validation of the end-to-end quality is made using the MOS scale. This scale

describes the opinion of customers with voice transmission and it’s troubles (noise, robot

voice, echo, dropouts etc). The speech quality measurement is taken per call.

Reference: ITU-T P.862 (PESQ Algorithm)

Abstract formula:

Optionally it might be useful to aggregate both speech quality values into one. In this

case the worst of both shall be used. This aggregated speech quality value shall be

called SpQ (min).

Trigger points:

Beginning of connection:

interchange speech samples between a-party and b-party

End of connection:

release of connection

Remarks:

The acoustic behaviour of terminals is not part of this speech quality measurement.

17

Short Message Service

21. Service Accessibility SMS MO (SA SMS MO)

Abstract definition

Probability that the end-customer can access the Short Message Service when

requested while it is offered by display of the network indicator on the Mobile Equipment.

In this case the customer wants to send a Short Message.

Computation

Abstract formula:

Trigger points [e.g. Layer-3 messages]:

Start SMS service attempt:

Initiate sending a SMS

Successful SMS service attempt:

receiving acknowledgement of the SMSC

22. Service Accessibility SMS MT (SA SMS MT)

Abstract definition

Probability that the end-customer can receive a Short Message from its Home Network

SMS-C while it is offered by display of the network indicator on the Mobile Equipment. In

this case the customer wants to receive a Short Message.

18

Computation

Only the valid attempts have to be measured. Errors due to user mistake (e.g. memory

full) should be excluded.

Abstract formula:

Trigger points [e.g. Layer-3 messages]:

Start SMS service attempt:

Initiate sending a SMS from Home SMS-C

Successful SMS service attempt:

receiving Short Message at Roaming side

23. Access Delay SMS MO (AD SMS-MO)

Abstract definition

Time between sending a Short Message to a Short Message Centre and receiving the

notification from the Short Message Centre.

Computation

Abstract formula:

treceive: point of time the mobile equipment receives the confirmation from the SMS Centre

tsend SMS: point of time the customer sends his SMS to the SMS Centre

Trigger points [e.g. Layer-3 messages]:

19

Start SMS service attempt:

Initiate sending a SMS

Successful SMS service attempt:

receiving acknowledgement of the SMSC

24. Access Delay SMS MT (AD SMS-MT)

Abstract definition

Time between sending a Short Message from the Home Short Message Centre and

receiving the notification at the Short Message Centre.

Computation

Abstract formula:

treceive: point of time the SMS Centre receives confirmation that the Short Message was

correctly delivered.

tsend SMS: point of time the Short Message leaves the e SMS Centre

Trigger points [e.g. Layer-3 messages]:

Start SMS service attempt:

Initiate sending a SMS from Home SMS-C

Successful SMS service attempt:

Receiving Short Message Confirmation of Delivery.

20

25. End-to-End Delivery Time for SMS MO (E2E DT SMS-MO)

Abstract definition

Time between sending a Short Message to a Short Message Centre in the VPMN and

receiving the SMS reception notification from the Short Message Centre in the HPMN.

Computation

Abstract formula:

Trigger points:

treceive SMS:

point of time the mobile equipment in the HPMN receives the new message

notification from the SMS Centre

tsend SMS:

point of time the customer sends his SMS to the SMS Centre from the VPMN

Remarks:

Not relevant for a QoS Roaming SLA since time measured is dependent on the

performance of the HPMN SMS-C

26. End-to-End Delivery Time for SMS MT (E2E DT SMS-MT)

Abstract definition

Time between sending a Short Message to a Short Message Centre in the HPMN and

receiving the SMS reception notification from the Short Message Centre in the VPMN.

21

Computation

Abstract formula:

Trigger points:

treceive SMS:

point of time the mobile equipment in the VPMN receives the new message

notification from the SMS Centre

tsend SMS:

point of time the customer sends his SMS to the SMS Centre from the HPMN

Remarks: Not relevant for a QoS Roaming SLA since time measured is dependent on

the performance of the HPMN SMS-C

22

Packet Switched Service

31. Packet Switched LU Success Rate (PS LU - SR)6

Abstract definition

The PSLU success ratio describes the probability that a subscriber can successfully

attach to the PS network.

Computation

Formula:

Trigger points:

Start:

Mobile sends the PS attach request message.

Stop:

Mobile receives the PS attach accept message.

Remark(s):

32. Packet Switched Location Update Delay (PS LU - D)

Abstract Definition

6 For sake of clarity, the Packet-Switched Location Update indicator mentioned is based on the MAP_Update_GPRS_Location procedure, as described in ETSI 129.002.The Packet-Switched term - rather than GPRS - is used to avoid confusion between the Service offered (Data access via GPRS, EDGE, UMTS,..) and the GPRS bearer technology.

23

This PSLU delay describes the time period needed to attach to the PS network.

Computation

Formula:

Trigger points (for the computation of the unit PSLU delay):

Start:

Point of time when the mobile sends the attach request message.

Stop:

Point of time when the mobile receives the attach accept message.

Remarks:

The difference between an attach of a known subscriber and an unknown subscriber will

be reflected in the time period indicating the attach setup time. In case of an unknown

subscriber (meaning that the SGSN has changed since the detach, or if it is the very first

attach of the mobile to the network), the SGSN contacts the HLR in order to receive the

subscriber data. The attach setup time of an unknown subscriber will be slightly longer

than the one of a known subscriber.

33. PDP Context Activation Success Rate (SA PSD)

A packet-switch data session will be considered set-up successfully if a PDP Context

can be successfully activated.

24

Abstract formula:

emptsivationAttContextActNrOfAllPDP

ssfultionsSuccetextActivaNrOfPDPConeSuccessRatActivationPDPContext %100

Trigger points:

Any PDP Context Activation request is considered as an attempt.

PDP Context Activations are considered successful upon the reception of notifications of

successful PDP context activation (Activate PDP Context Accept)

Remarks:

The suggestion is to use a limited list of 6 APN’s to measure. The HPMN should publish

these HPMN’s either in the IR21 or RSLA. This could include APN’s for MMS, WAP

and/or internet, both for pre-paid and post-paid. This needs to be defined further in the

test conditions.

25

34. PDP Context Activation time

Time between sending the PDP Context Activation request and receiving the notification

of successful completion of that activation.

Abstract formula:

12 t- t[s] Data SwitchedPacket Time up-Set

Trigger points:

Start:

sending of the PDP Context Activation request

End:

reception of the notification of successful PDP context activation (Activate PDP Context Accept)

Remarks:

When averaging the PDP Context Activation Delays, only the successful activations

should be considered.

The suggestion is to use a limited list of 6 APN’s to measure. The HPMN should

publish these HPMN’s either in the IR21 or RSLA. This could include APN’s for

MMS, WAP and/or internet, both for pre-paid and post-paid. This needs to be defined

further in the test conditions.

We also recognize there are 2 different activation types:customer activation (GPRS PDP context creation procedure)

26

network activation (GPRS PDP notification procedure) For the moment we suggest to only use the first one. Reasons are network activations tend to be for local services (Home usage) and it is not clear if the VPMN can replicate network activations.

35. PDP Context Cut-Off Ratio

Abstract definition

The PDP context cut-off ratio denotes the probability that a PDP context is deactivated

without being deactivated intentionally by the user.

Computation

Abstract equation:

Trigger points:

Different trigger points for a PDP context deactivation not initiated intentionally by the

user are possible: SGSN failure or GGSN failure on which the PDP context will be

deactivated by the SGSN or GGSN.

Remarks:

o When analyzing how to practically measure this parameter, a key point will be to

identify how to assess whether a PDP context loss has been initiated by the user or

not. Active and passive monitoring methods might adopt different approaches for

that. In the active monitoring case, the test case specifies when the PDP context

should be deactivated and any loss prior to that can thus (generally) be considered

as ‘not initiated by the user’. In the passive monitoring case, the use of PDP context

failure codes can be used.

27

o Precondition for measuring this parameter is that a PDP context was successfully

established first.

36. Average PDP Context Average Session Time

Abstract definition

The average PDP context session time is the average duration of the PDP context

sessions successfully completed.

Computation

Abstract equation:

)( ActivationPDPContextonDeactivatiPDPContext ttAveragesionTimeAverageSesPDPContext

The trigger points defining the duration are:

Start:

notification of successful PDP context activation (Activate PDP Context Accept)

End:

PDP context deactivation request initiated by the user (Deactivate PDP Context Request)

Remarks:

The PS bearer has to be active in the cell used by a subscriber (cf. Unavailability) and

the mobile station has to be attached as well as the respective PDP context has to be

activated.

37. Throughput (Kbit/sec)

28

This parameter describes the average data transfer rate at the ‘bit pipe’ level (and not at

the User Application level).

The prerequisite for this parameter is network and service access.

Computation

Passive monitoring:

The average throughput is measured between the border gateways of the HPMN and

VPMN from opening the data connection to its end It is derived from the total exchanged

data between the parties over a period of time.

Active monitoring:

The average throughput can be estimated using an FTP transfer from/to a server that

would be very close to the HPMN’s border gateway (to minimize the impact of network

29

links that are “after the bit pipe” on the HPMN side). Using a reference file of a known

volume, the throughput can be derived from the measure of the actual data volume

transferred and the corresponding needed time.

Remarks:

The “bit pipe” starts from the MS to the HPMN Border Gateway.

38 Goodput (Kbit/sec)

This parameter describes the average data transfer rate at the ‘bit pipe’ level (and not at

the User Application level).

The prerequisite for this parameter is network and service access.

Computation

30

Passive monitoring:

The average goodput is measured between the border gateways of the HPMN and

VPMN from opening the data connection to its end. It is derived from the total effective

useful exchanged data (excl. overhead and retransmission) between the parties over a

period of time.

Active monitoring:

The average goodput can be estimated using an FTP transfer from/to a server that

would be very close to the HPMN’s border gateway (to minimize the impact of network

links that are “after the bit pipe” on the HPMN side). Using a reference file of a known

volume, the goodput can be derived from the measure of the known data volume to be

transferred and the corresponding required time.

31

Remarks:

The “bit pipe” starts from the MS to the HPMN Border Gateway.

39. Roundtrip time

This parameter describes the average time required for a packet to travel from the

VPMN side of the ‘bit pipe’ to the HPMN side and come back. This parameter has to be

considered at the ‘bit pipe’ level (and not at the User Application level).

The prerequisite for this parameter is network and service access.

It is used to measure the delay on a network at a given time.

Computation:

Round Trip Time [ms] = (timestamp Packet received) − (timestamp Packet sent)

Passive monitoring:

The measurement of the round trip time is done by evaluating, for example the TCP handshake: Points of time between the [SYN], [SYN,ACK] and [ACK].

32

Active monitoring:

The average round trip time can be estimated using a PING to a server that would be

very close to the HPMN’s border gateway (to minimize the impact of network links that

are “after the bit pipe” on the HPMN side).

Remarks:

The “bit pipe” starts from the MS to the HPMN Border Gateway.

40. Packet loss

Abstract definition

This parameter describes the average percentage of packets lost in the ‘bit pipe’ (not at

the User Application level).

The prerequisite for this parameter is network and service access.

Computation

Abstract equation:

ceivedPacket

PacketSentlossPacket

Re1%100_

Passive monitoring:

The ratio of packet send over packet received of the unreliable UDP-based transport

33

layer. The estimation mechanism, is based on the measurement of loss packet thanks to

the tracking of the reliable application-layer TCP procotol.

Active monitoring:

The ratio of ICMP packages responses sent to a HPMN server and the ICMP packages

sent to a HPMN server can be used to compute the packet loss, provided that the

PINGed server is close to the HPMN’s border gateway (to minimize the impact of

network links that are “after the bit pipe” on the HPMN side).

Remarks: Preconditions for measuring this parameter are:

o The IP address should be one of the HPMN network (GGSN, WAP GW, tests server, …)

o The firewalls are not allowed to block these ICMP echo requests for the tested IP address and Port

The “bit pipe” starts from the MS to the HPMN Border Gateway.

General note: the jitter parameter has not been considered in the current edition of the

document in order to limit the number of parameters, considering that such parameter

mainly affects real-time data services, which are not significantly used in roaming today.

34