ETSI TR 101 329-7 V2.1.1 (2002-02)

SLOVENSKI STANDARD
01-april-2004
+DUPRQL]DFLMDWHOHNRPXQLNDFLMLQLQWHUQHWQHJDSURWRNRODSUHNRPUHåLM7,3+21
L]GDMD.DNRYRVWVWRULWYHRGNRQFDGRNRQFDYVLVWHPLK7,3+21GHO9RGLOR]D
QDþUWRYDQMHHOHPHQWRYSRYH]DY7,3+21]YLGLND]PRJOMLYRVWLSUHQRVDJRYRUDRG
NRQFDGRNRQFD
Telecommunications and Internet Protocol Harmonization Over Networks (TIPHON)
Release 3; End-to-end Quality of Service in TIPHON systems; Part 7: Design guide for
elements of a TIPHON connection from an end-to-end speech transmission performance
point of view
Ta slovenski standard je istoveten z: TR 101 329-7 Version 2.1.1
ICS:
33.020 Telekomunikacije na splošno Telecommunications in
general
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

Technical Report
Telecommunications and Internet Protocol
Harmonization Over Networks (TIPHON) Release 3;
End-to-end Quality of Service in TIPHON systems;
Part 7: Design guide for elements of a TIPHON
connection from an end-to-end speech
transmission performance point of view

2 ETSI TR 101 329-7 V2.1.1 (2002-02)

Reference
RTR/TIPHON-05014
Keywords
coding, E-model, internet, IP, network,
performance, planning, protocol, QoS, quality,
speech, transmission, voice
ETSI
650 Route des Lucioles
F-06921 Sophia Antipolis Cedex - FRANCE

Tel.: +33 4 92 94 42 00  Fax: +33 4 93 65 47 16

Siret N° 348 623 562 00017 - NAF 742 C
Association à but non lucratif enregistrée à la
Sous-Préfecture de Grasse (06) N° 7803/88

Important notice
Individual copies of the present document can be downloaded from:
http://www.etsi.org
The present document may be made available in more than one electronic version or in print. In any case of existing or
perceived difference in contents between such versions, the reference version is the Portable Document Format (PDF).
In case of dispute, the reference shall be the printing on ETSI printers of the PDF version kept on a specific network drive
within ETSI Secretariat.
Users of the present document should be aware that the document may be subject to revision or change of status.
Information on the current status of this and other ETSI documents is available at
http://portal.etsi.org/tb/status/status.asp
If you find errors in the present document, send your comment to:
editor@etsi.fr
Copyright Notification
No part may be reproduced except as authorized by written permission.
The copyright and the foregoing restriction extend to reproduction in all media.

© European Telecommunications Standards Institute 2002.
All rights reserved.
ETSI
3 ETSI TR 101 329-7 V2.1.1 (2002-02)
Contents
Intellectual Property Rights.5
Foreword.5
Introduction .6
1 Scope.7
2 References.7
3 Definitions and abbreviations.8
3.1 Definitions.8
3.2 Abbreviations.9
4 General considerations.10
4.1 Transmission planning.11
4.2 User interaction.11
4.3 Maintenance.11
4.4 Monitoring & verification.11
4.5 Interconnection of TIPHON systems with other IP networks .11
5 Guidance on main transmission parameters.14
5.1 Loudness ratings.14
5.1.1 General considerations.14
5.1.2 IP terminals.14
5.1.3 IP gateways.14
5.2 Mean one-way delay .14
5.2.1 Absolute delay.14
5.3 Delay jitter.15
5.3.1 Jitter buffer implementations .15
5.3.1.1 Static jitter buffers.15
5.3.1.2 Dynamic Jitter Buffers .16
5.3.2 Jitter buffer monitoring capabilities.16
5.3.3 Impact.17
5.4 Echo loss, echo cancellation.18
5.4.1 General considerations.18
5.4.2 IP terminals.18
5.4.3 IP gateways.18
5.5 Coding distortion.18
5.5.1 General considerations.18
5.6 Speech processing other than coding.19
5.6.1 General considerations.19
5.6.2 IP terminals.20
5.6.3 IP gateways.20
5.7 Transcoding in network elements.20
5.8 Packet loss.22
5.9 Example of TIPHON QoS parameter allocation .22
6 Calculation and planning examples.23
6.1 Examples with respect to loudness ratings .25
6.2 Examples with respect to mean one-way delay.25
6.2.1 Delay due to speech processing and packetization .25
6.2.2 Planning examples regarding the occurrence of long delay.28
6.2.2.1 Introduction.28
6.2.2.1.1 Application of the advantage factor A with respect to the following examples .28
6.2.2.1.2 Distinction between different communication situations for the following examples with
regard to the grade of interactivity between the two parties.29
6.2.2.1.3 Introduction of an additional equipment impairment factor with respect to double-talk
situations for the following examples.29
6.2.2.1.4 Purpose and general structure of the following examples .30
ETSI
4 ETSI TR 101 329-7 V2.1.1 (2002-02)
6.2.2.2 Connections to regions to which significantly shorter delay is available ("Competition").31
6.2.2.2.1 Speech transmission performance as perceived in listening-only communication situations.31
6.2.2.2.2 Speech transmission performance as perceived in typical communication situations.32
6.2.2.2.3 Speech transmission performance as perceived in highly interactive communication
situations.33
6.2.2.3 Connections to regions to which no shorter delay is available ("Hard-to-reach") .34
6.2.2.3.1 Speech transmission performance as perceived in listening-only communication situations.34
6.2.2.3.2 Speech transmission performance as perceived in typical communication situations.35
6.2.2.3.3 Speech transmission performance as perceived in highly interactive communication
situations.36
6.2.2.4 Summary on planning results for long delay.36
6.2.3 VoIP end-to-end delay budget planning for private networks .37
6.2.3.1 VoIP end-to-end delay sources overview.37
6.2.3.2 VoIP end-to-end delay sources definitions .38
6.2.3.3 VoIP End-to-end Delay Budget Case 1.41
6.2.3.4 VoIP End-to-end Delay Budget Case 2.43
6.2.4 E-Model analysis of the VoIP over MPLS reference model.44
6.2.4.1 Introduction.44
6.2.4.2 MPLS in core networks only.45
6.2.4.2.1 Scenario 1 - Multiple MPLS Networks .45
6.2.4.2.2 Scenario 2 - PSTN and typical DCME practice .46
6.2.4.2.3 Scenario 3 - GSM and Typical DCME practice .47
6.2.4.2.4 VoMPLS core network summary .48
6.2.4.3 Extending VoMPLS into the access network.49
6.2.4.3.1 Scenario 4 - VoMPLS access on USA to Japan .49
6.2.4.3.2 Scenario 5 deployment of GSM and VoMPLS access .50
6.2.4.4 Effects of Voice Codecs in the access network.51
6.2.4.4.1 Scenario 6 - Deployment of Codecs in one Access Leg (USA – Japan) .52
6.2.4.4.2 Scenario 7 - Codec Deployment in both Access Legs (USA - Japan).53
6.2.4.4.3 Scenario 8 codec deployment and mobile access (USA - Australia).54
6.2.4.4.4 Voice codec summary .54
6.2.4.5 Overall conclusions.55
6.3 Examples with respect to the provision of proper echo control .55
6.3.1 TIPHON QoS class "narrowband high (2H)" .55
6.3.2 TIPHON QoS class "narrowband medium (2M)".57
6.3.3 TIPHON QoS class "narrowband acceptable (2A)" .59
6.4 Examples with respect to coding distortion.61
6.5 Examples with respect to speech processing other than coding .61
6.6 Examples with respect to packet loss .61
6.7 Interpretation of the results.61
6.8 Guidance on the relation and Interdependency between Auditory MOS, Objective MOS and Predicted
MOS .62
Annex A: Bibliography.64
History .72

ETSI
5 ETSI TR 101 329-7 V2.1.1 (2002-02)
Intellectual Property Rights
IPRs essential or potentially essential to the present document may have been declared to ETSI. The information
pertaining to these essential IPRs, if any, is publicly available for ETSI members and non-members, and can be found
in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in
respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the ETSI Web
server (http://webapp.etsi.org/IPR/home.asp).
Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guarantee
can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web
server) which are, or may be, or may become, essential to the present document.
Foreword
This Technical Report (TR) has been produced by ETSI Project Telecommunications and Internet Protocol
Harmonization Over Networks (TIPHON).
The present document is part 7 of a multi-part deliverable covering End-to-end Quality of Service in TIPHON systems,
as identified below:
TR 101 329-1: "General aspects of Quality of Service (QoS)";
TS 101 329-2: "Definition of speech Quality of Service (QoS) classes";
TS 101 329-3: "Signalling and control of end-to-end Quality of Service (QoS)";
TS 101 329-5: "Quality of Service (QoS) measurement methodologies";
TR 101 329-6: "Actual measurements of network and terminal characteristics and performance parameters in
TIPHON networks and their influence on voice quality";
TR 101 329-7: "Design guide for elements of a TIPHON connection from an end-to-end speech
transmission performance point of view".
Quality of Service aspects of TIPHON Release 4 and 5 Systems will be covered in TS 102 024 and TS 102 025
respectively, and more comprehensive versions of the Release 3 documents listed above will be published as part of
Releases 4 and 5 as work progresses.
ETSI
6 ETSI TR 101 329-7 V2.1.1 (2002-02)
Introduction
The present document forms one of a series of technical specifications and technical reports produced by TIPHON
Working Group 5 addressing Quality of Service (QoS) in TIPHON Systems. The structure of this work is illustrated in
figure 1.
Introduction Definition of 5
Speech Classes
TR 101 329-1 TS 101 329-2
General Speech
QoS
Aspects
of QoS Classes
SPEC
REPORT
Generic QoS
Specific Aspects of QoS
TS 101 329-5 TR 101 329-6 TR 101 329-7
TS 101 329-3
QoS Measure- Actual Design
Control ment Test Guidelines
Methods Results
SPEC REPORT REPORT
SPEC
QoS signalling Measurement Repository of Useful info for
requirements methodologies real test results designers

Figure 1: Structure of TIPHON QoS Documentation for Release 3
For a concise understanding of the guidance provided in the present document it is strongly recommended that the
reader be aware of the content of the most recent version of TS 101 329-6 [3] which is a repository of real results.
Figure 2: Void
ETSI
7 ETSI TR 101 329-7 V2.1.1 (2002-02)
1 Scope
The present document provides a collection of informative background information and guidance to supplement
parts 1 to part 6 of TS 101 329. The issues covered concern the practical design phases for both equipment and
networks with respect to speech performance, and therefore is relevant to TIPHON equipment manufacturers, service
providers and network designers.
2 References
For the purposes of this Technical Report (TR) the following references apply:
[1] ETSI TS 101 329-2: "Telecommunications and Internet Protocol Harmonization Over Networks
(TIPHON) Release 3; End-to-end Quality of Service in TIPHON Systems; Part 2: Definition of
Speech Quality of Service (QoS) Classes".
[2] ETSI TS 101 329-5: "Telecommunications and Internet protocol Harmonization Over Networks
(TIPHON) Release 3; End-to-end Quality of Service in TIPHON Systems; Part 5: Quality of
Service (QoS) measurement methodologies".
[3] ETSI TS 101 329-6: "Telecommunications and Internet Protocol Harmonization Over Networks
(TIPHON) Release 3; End-to-end Quality of Service in TIPHON systems; Part 6: Actual
measurements of network and terminal characteristics and performance parameters in TIPHON
networks and their influence on voice quality".
[4] ITU-T Recommendation G.100: "Definitions used in Recommendations on general characteristics
of international telephone connections and circuits".
[5] ITU-T Recommendation G.122: "Influence of national systems on stability and talker echo in
international connections".
[6] ITU-T Recommendation G.131: "Control of talker echo".
[7] ITU-T Recommendation G.111: "Loudness ratings (LRs) in an international connection".
[8] ITU-T Recommendations P.64: "Determination of sensitivity/frequency characteristics of local
telephone systems".
[9] ITU-T Recommendation G.109: "Definition of categories of speech transmission quality".
[10] ITU-T Recommendation G.726: "40, 32, 24, 16 kbit/s adaptive differential pulse code modulation
(ADPCM)".
[11] ITU-T Recommendation G.114: "One-way transmission time".
[12] ETSI I-ETS 300 245-2: "Integrated Services Digital Network (ISDN); Technical characteristics of
telephony terminals; Part 2: PCM A-law handset telephony".
[13] ITU-T Recommendation P.800: "Methods for subjective determination of transmission quality".
[14] ITU-T Recommendation P.310: "Transmission characteristics for telephone-band (300-3400 Hz)
digital telephones".
[15] ANSI/TIA/EIA-810-A-2000: "Telecommunications-Telephone Terminal Equipment-Transmission
Requirements for Narrowband".
[16] ITU-T Recommendation G.168: "Digital network echo cancellers".
[17] ITU-T Recommendation G.107: "The E-Model, a computational model for use in transmission
planning".
[18] ITU-T Recommendation G.113: "Transmission impairments due to speech processing".
ETSI
8 ETSI TR 101 329-7 V2.1.1 (2002-02)
[19] ETSI EG 201 050: "Speech Processing, Transmission and Quality Aspects (STQ); Overall
Transmission Plan Aspects for Telephony in a Private Network".
[20] ITU-T Recommendation G.108: "Application of the E-model: A planning guide".
[21] ETSI ETR 250: "Transmission and Multiplexing (TM); Speech communication quality from
mouth to ear for 3,1 kHz handset telephony across networks".
[22] ITU-T Recommendation P.833: "Methodology for derivation of equipment impairment factors
from subjective listening?only tests".
3 Definitions and abbreviations
3.1 Definitions
For the purposes of the present document, the following terms and definitions apply:
dBm: power level with reference to 1 mW
dBm0: at the reference frequency (1 020 Hz), L dBm0 represents an absolute power level of L dBm measured at the
transmission reference point (0 dBr point), and a level of L + x dBm measured at a point having a relative level of x dBr
NOTE: See ITU-T Recommendation G.100 [4], annex A.4.
echo: unwanted signal delayed to such a degree that it is perceived as distinct from the wanted signal
talker echo: echo produced by reflection near the listener's end of a connection, and disturbing the talker
listener echo: echo produced by double reflected signals and disturbing the listener
Loudness Rating (LR): as used in the G-Series Recommendations for planning; loudness rating is an objective
measure of the loudness loss, i.e. a weighted, electro-acoustic loss between certain interfaces in the telephone network
NOTE: If the circuit between the interfaces is subdivided into sections, the sum of the individual section LRs is
equal to the total LR. In loudness rating contexts, the subscribers are represented from a measuring point
of view by an artificial mouth and an artificial ear respectively, both being accurately specified.
Overall Loudness Rating (OLR): loudness loss between the speaking subscriber's mouth and the listening subscriber's
ear via a connection
Talker Echo Loudness Rating (TELR): loudness loss of the speaker's voice sound reaching his ear as a delayed echo
NOTE: See ITU-T Recommendation G.122 [5], clause 4.2 and ITU-T Recommendation G.131 [6], figure I.1.
Terminal Coupling Loss weighted (TCLw): weighted coupling loss between the receiving port and the sending port
of a terminal due to acoustical coupling at the user interface, electrical coupling due to crosstalk in the handset cord or
within the electrical circuits, seismic coupling through the mechanical parts of the terminal
NOTE: For a digital handset it is commonly in the order of 40 db to 46 dB.
weighted Terminal Coupling Loss-single talk (TCLwst): weighted loss between Rin and Sout network interfaces
when AEC is in normal operation, and when there is no signal coming from the user
weighted Terminal Coupling Loss-double talk (TCLwdt): weighted loss between Rin and Sout network interfaces
when AEC is in normal operation, and when the local user and the far-end user talk simultaneously
Send Loudness Rating (SLR) (from ITU-T Recommendation G.111): loudness loss between the speaking
subscriber's mouth and an electric interface in the network
NOTE: The loudness loss is defined here as the weighted (dB) average of driving sound pressure to measured
voltage. The weighted mean value for ITU-T Recommendations G.111 [7] and G.121 (see Bibliography)
is 7 to 15 in the short term, 7 to 9 in the long term. The rating methodology is described in ITU-T
Recommendations P.64 [8], P.76 (see Bibliography) and P.79 (see Bibliography).
ETSI
9 ETSI TR 101 329-7 V2.1.1 (2002-02)
Receive Loudness Rating (RLR) (from ITU-T Recommendation G.111): loudness loss between an electric interface
in the network and the listening subscriber's ear
NOTE: The loudness loss is here defined as the weighted (dB) average of driving e.m.f. to measured sound
pressure. The weighted mean value for ITU-T Recommendations G.111 [7] and G.121 (see Bibliography)
is 1 to 6 in the short term, 1 to 3 in the long term. The rating methodology is described in ITU-T
Recommendations P.64 [8], P.76 (see Bibliography), P.79 (see Bibliography).
Circuit Loudness Rating (CLR): loudness loss between two electrical interfaces in a connection or circuit, each
interface terminated by its nominal impedance which may be complex
toll quality: In general "toll quality" is a term which is not well defined. Currently, there are two different views:
• ITU-T Recommendation G.109 [9] provides the following guidance:
"Finally, to relate the definitions provided by this Recommendation to concepts and terminology used in the past,
a comment about "toll quality" is in order. "Toll quality" has been used by many different people to mean
different things, but to network planners it really meant that technology being introduced into the network was
robust to the effects of transmission impairments from other sources, and could thus be used in many
configurations where inter-working with other systems would be necessary. In this context, the term "toll
quality" does not have any absolute relation to speech transmission quality today, because, for example, the
impairments of systems such as wireless access or packet-based transport will have the same impact regardless
of whether on a local or a long-distance connection. Instead, the terminology provided here is recommended (i.e.
"best" for R in the range from 90 to 100, "high" in the range from 80 to 90 and "medium" in the range from 70 to
80)."
• Experts on low bit-rate coding (members of ITU-T Study Group 16 and SQEG) use the following explanation:
"In summary, we define toll quality as equivalent to wire-line telephone quality. Basically the 32 kb/s ADPCM
(ITU-T Recommendation G.726 [10]) is considered to be a toll quality coder, and when some low rate coders get
standardized in ITU-T, 32 kb/s ADPCM is used as reference, and if a low rate coder produce equivalent
performance to the 32 kb/s ADPCM, then this is considered to be toll quality."
Consequently, at this time the term "toll quality" should be considered as an internal term of speech coder experts only
which is obsolete and which should be avoided in conjunction with the TIPHON QoS documentation. TIPHON
equipment manufacturers and network designers should rather use the Quality Categories defined in ITU-T
Recommendation G.109 [9] or the QoS Classes specified in TIPHON (TS 101 329-2 [1]).
NOTE: Harmonization of the views regarding the term "toll quality" are envisaged to be discussed during the
Study Period 2001-2004.
3.2 Abbreviations
For the purposes of the present document, the following abbreviations apply:
ACR Absolute Category Rating
ADSL Asymmetric Digital Subscriber Line
AEC Acoustic Echo Control
ALC Automatic Level Control
ALC Automatic Level Control
ASL Active Speech input Level
ATM Asynchronous Transfer Mode
DCME Digital Circuit Multiplication Equipment
DTMF Dual Tone Multi Frequency
DTX Discontinuous Transmission
ECD Echo Control Devices
GSM EFR GSM Enhanced Full Rate Speech Coder
GSM FR GSM Full Rate Speech Coder
GSM HR GSM Half Rate Speech Coder
GSM Global System for Mobile communications
ETSI
10 ETSI TR 101 329-7 V2.1.1 (2002-02)
IP Internet Protocol
ISDN Integrated Services Digital Network
ISP Internet Service Provider
IWF Inter Working Function
LAN Local Area Network
MOS Mean Opinion Score
MPLS Multi Protocol Layer Switching
NIC Network Interface Card
NS Noise suppressors
PPP Point to Point Protocol
PSTN Public Switched Telephone Network
QoS Quality of Service
RSVP Resource Reservation Set-Up Protocol
RTP Real-Time Transport Protocol
SBM Subnet Bandwidth Manager
SCN Switched Communications Network
TCP Transmission Control Protocol
TRM Transmission Rating Model
UDP User Datagram Protocol
VAD Voice Activity Detection
VAD Voice Activity Detectors
VDSL Very High Speed Digital Subscriber Line
VED Voice Enhancement Devices
VoIP Voice over IP
VTOA Voice and Telephony Over ATM
WAN Wide Area Network
xDSL ADSL, VDSL and other Digital Subscriber Line Techniques
4 General considerations
The realization of end to end speech quality in a TIPHON system is determined by a combination of user equipment
design, service provider equipment performance and network transmission planning. To guarantee end to end speech
quality:
• user equipment must meet specified performance requirements; and
• service provider equipment must meet specified performance requirements and be correctly configured by
service providers;
• underlying transport networks, involved in the call end-to-end (IP as well as SCN), must be designed to deliver
specific performance criteria at all times. It is implicit that guarantees can only be achieved over managed IP
networks, engineered to deliver a given level of performance, and where traffic levels are controlled.
The issues of end-to-end speech transmission quality need to be considered from various perspectives therefore:
• user equipment and service provider equipment design by manufactures;
• system configuration by service providers;
• transmission planning by network operators.
The purpose of the present document is to provide design guidelines in each of these areas.
The following steps are likely to be involved in implementing a TIPHON system:
• Planning and configuration;
• Pre-qualification;
• User interaction;
ETSI
11 ETSI TR 101 329-7 V2.1.1 (2002-02)
• Maintenance;
• Monitoring and Verification,
which are summarized in the following clauses.
4.1 Transmission planning
In order to deliver the intended end-to-end speech transmission quality in TIPHON systems, transmission planning
should be performed during the design phase of TIPHON related equipment. It is not sufficient to design equipment or
networks just along the requirement limits of the respective TIPHON class.
An advantage factor A (see ITU-T Recommendation G.107 and Appendix II to ITU-T Recommendation G.113) which
is sometimes discussed for Internet-Telephony does not generally apply for business applications and consequently does
also not apply for TIPHON systems. However, this is not a general planning rule, but a business and customer related
decision for this single example case or for a specific service.
Any variation of transmission parameters should only be judged on the basis of E-model calculations for critical end-to-
end connections. Any assumption whether or whether not a specific parameter variation will be perceived by the user
should always be based on E-model calculations.
Special care should be taken with devices which dynamically vary one or more transmission parameters, e.g. Automatic
Level Control (ALC) devices; experiences with such devices have shown that they have the potential to impact
end-to-end speech transmission quality, severely.
4.2 User interaction
User interaction with regard to the change of certain transmission parameters may be provided by equipment which
forms part of a TIPHON connection, e.g. a TIPHON terminal may include a PC client software which provides
adjustment of Loudness Rating to the user (see clause 5.1.2 for further guidance).
4.3 Maintenance
After TIPHON equipment and networks have been designed, planned and rendered operative in compliance with one
the TIPHON QoS classes it might - nevertheless - occur that users complain about too low speech quality.
In such cases, it is very important to be able to carry through a diagnosis of end-to-end speech transmission
performance. For that it will be needed to keep track of all parameter changes (e.g. of Send and Receive Loudness
Rating) carried out either automatically or by user interaction.
This should be considered already during the design phase of TIPHON equipment and networks, e.g. by providing tools
to set parameters back to default values or by providing a log file function.
4.4 Monitoring & verification
Even if a specific TIPHON system has been operated for some time at the desired level of customer satisfaction it will
be required to continuously monitor and check the end-to-end speech transmission quality.
Verification will require access to the actual settings of all major transmission parameters - including those which were
accessible to the user.
4.5 Interconnection of TIPHON systems with other IP networks
Implementers of TIPHON networks are advised that IP networks other than those following the TIPHON regime,
eventually may employ different QoS classification schemes than the one defined in [1].
As an example, in the following the TIPHON approach is compared to the TIA approach taken in [2].
ETSI
12 ETSI TR 101 329-7 V2.1.1 (2002-02)
Explanation of the arrangement of the figures
The figures are drawn with LSQ as the Y-axis and delay as the X-axis as these are the main design parameters over
which the user has some control.
TIPHON sets three criteria for its QoS classes as shown in the following table which is an excerpt from [1].
2H 2M 2A
OVR 80 70 50
Delay 100 ms 150 ms 400 ms
LSQ 86 73 50
Whilst the criteria for LSQ and delay are independent of each other, the criterion on OVR depends on LSQ, delay and
other parameters. Some assumptions therefore have to be made for the other parameters as follows:
• Perfect echo cancellation is assumed for all TIPHON systems therefore, TELR = 65 dB and WEPL = 110 dB
(default as per G.107). Whereas the need for proper echo control is recognized in general, other IP networks
may consider the quality of echo cancellation as actually achieved in their QoS classification.
• The E-model default values are assumed for all terminal related parameters.
• The Overall E-Model Rating R is calculated according to E-model, with TELR, WEPL and terminal related
parameters as above (G.107 default) and the values of delay and LSQ as on the graph.
Figure 3 illustrates the TIPHON QoS classes as described in TS 101 329-2 [1].
TIPHON QoS classes are defined by a combination of all three metrics, LSQ, delay and OVR. Therefore, the colored
areas represent the TIPHON QoS classes as follows:
Green  Narrowband High (2H)
Yellow Narrowband Medium (2M)
Red Narrowband Acceptable (2A)
Mauve Not Recommended (1)
Figure 3: TIPHON QoS classes
ETSI
13 ETSI TR 101 329-7 V2.1.1 (2002-02)
Figure 3 shows how the criteria for LSQ and delay are more stringent for the high and medium quality classes than the
criterion on OVR under the other conditions chosen. But for acceptable and not recommended quality, OVR is the most
stringent criterion.
Figure 4 illustrates the TIA approach as contained in [2].
TIA QoS classes are defined by a pure E-Model approach, employing OVR, only. Therefore, the colored areas represent
the TIA QoS classes as follows:
Green  High
Yellow Medium
Red Low
Mauve 3Not Recommended
Please, note that, the vertical and horizontal lines which indicate the TIPHON requirements on LSQ and delay have
been added to the TIA diagram in figure 4 in order to accomplish a more convenient comparability with figure 3.

Figure 4: TIA QoS classes
Furthermore, it should be noted that in IP networks other than TIPHON one or all of the following may differ:
• name or description of QoS classes;
• guarantees provided or given for each individual class;
• availability and description of wideband classes.
In cases where TIPHON systems are interconnected to other IP networks, implementers of TIPHON systems, therefore,
should thoroughly inspect the basic transmission parameters of the end-to-end connection.
Any conclusion of the resulting end-to-end speech transmission performance of such an interconnection scenario should
be based on end-to-end E-model calculations. A mapping or transformation of TIPHON QoS classes with classes of
other IP networks is, in general, not feasible.
ETSI
14 ETSI TR 101 329-7 V2.1.1 (2002-02)
5 Guidance on main transmission parameters
Overall Packet
Network
Packet
Loss
Packet Loss
Loss
Concealment
Codec
Perceived
Performance
Jitter
Quality
Network Jitter
Buffers
Overall
Network Delay
Delay
Figure 5: Interaction of transmission aspects, e.g. the influence of jitter buffers
on packet loss and delay
5.1 Loudness ratings
5.1.1 General considerations
5.1.2 IP terminals
This is for further study.
5.1.3 IP gateways
This is for further study.
5.2 Mean one-way delay
5.2.1 Absolute delay
Absolute delay is the end-to-end delay from mouth to ear.
See ITU-T Recommendation G.114 [11] and clause 5.3 of TS 101 329-2 [1] for further information.
ETSI
15 ETSI TR 101 329-7 V2.1.1 (2002-02)
5.3 Delay jitter
5.3.1 Jitter buffer implementations
5.3.1.1 Static jitter buffers
A static dejittering mechanism delays the first arriving packet for a time equal to the dejittering delay. After this
dejittering delay the dejittering buffer is read at constant rate. Packets that arrive before their play-out instant are
temporary stored in the dejittering buffer. Packets that arrive too late (after they were supposed to be read out) are
effectively lost. Hence, this dejittering delay should be chosen close to the maximum (e.g. a 99 % quantile) of the
variable part of the end-to-end delay. Dejittering introduces additional delay. The choice of the dejittering delay
involves a trade-off between delay and packet loss. When it is chosen too small a lot of packets will arrive too late for
play out and will be considered to be lost (although they arrive at the receiver). In this case the dejittering buffer
underflows (runs empty at the moment a packet is supposed to be read). The physical size of the dejittering buffer
should be chosen large enough such that no (or rarely) buffer overflows will occur. It can be proven that a buffer size of
twice the amount of packets that are generated during a time equal to the dejittering delay very rarely leads to buffer
overflow.
Several approaches for filling the jitter buffer at the start of a connection have been found:
1) At the start of a connection, the first packet is delayed by the size of the jitter buffer. During this artificial initial
delay, other packets are likely to arrive, thus creating a buffer of packets that can be used if, for some time, no
new packets arrive. This is really the same as approach 3, except that the maximum size of the jitter buffer is not
specified (but in reality it is usually limited and fixed).
2) The entire jitter buffer is filled with data, including silence packets, and kept full if possible. When a voice
packet arrives too soon, silence packets may be removed from the buffer to make space for the new packet.
3) The buffer is filled to an initial value (the low water mark) but has a larger size (the high water mark). Packets
are not played out until the low water mark is reached.
In general, packets that arrive out of order can be put back into the right order in the jitter buffer if they are still on time.
Packets that arrive too late are discarded due to the limited size of the jitter buffer
Various ideas about the optimal size of a jitter buffer exist:
• An integral multiple of the expected packet inter-arrival time.
• Depends on network.
• Sufficient number.
• At least twice the anticipated jitter.
• Roughly twice the length of size of the expected packet arrival time variance.
Typical example jitter buffer sizes:
• 80 ms.
• 50 ms to 100 ms.
• 20 ms to 100 ms.
• Low water mark 30 ms, high water mark 150 ms.
• 150 ms.
• 50 ms.
• Up to 200 ms.
It is recommended that static jitter buffer sizes be configurable in a quantitative way, i.e., in terms of ms or bytes.
ETSI
16 ETSI TR 101 329-7 V2.1.1 (2002-02)
5.3.1.2 Dynamic Jitter Buffers
It is possible to use an algorithm that will dynamically adjust the dejittering delay, it will estimate the variable part of
the delay of the first packet and adjust the dejittering delay accordingly. Such an algorithm has several advantages. The
delay is minimized and buffer underflows are avoided. Also, when a static dejittering delay is used in a gateway or
terminal it should be configured correctly at the establishment of each call to avoid buffer underflows and unnecessary
delay. This configuration is not needed when a dynamic dejittering delay is used.
Finally, the codecs in the sender and receiver might not be (perfectly) synchronized. A dynamic dejittering algorithm
will also compensate for this.
Several approaches to adapting the jitter buffer size have been f
...