1、 ETSI TR 102 720 V1.1.1 (2009-10)Technical Report Speech and multimedia Transmission Quality (STQ);Delay variation on unshared access linesETSI ETSI TR 102 720 V1.1.1 (2009-10) 2Reference DTR/STQ-00150 Keywords quality, voice ETSI 650 Route des Lucioles F-06921 Sophia Antipolis Cedex - FRANCE Tel.:
2、+33 4 92 94 42 00 Fax: +33 4 93 65 47 16 Siret N 348 623 562 00017 - NAF 742 C Association but non lucratif enregistre la Sous-Prfecture 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 ava
3、ilable 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
4、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
5、errors in the present document, please send your comment to one of the following services: http:/portal.etsi.org/chaircor/ETSI_support.asp Copyright Notification No part may be reproduced except as authorized by written permission. The copyright and the foregoing restriction extend to reproduction i
6、n all media. European Telecommunications Standards Institute 2009. All rights reserved. DECTTM, PLUGTESTSTM, UMTSTM, TIPHONTM, the TIPHON logo and the ETSI logo are Trade Marks of ETSI registered for the benefit of its Members. 3GPPTM is a Trade Mark of ETSI registered for the benefit of its Members
7、 and of the 3GPP Organizational Partners. LTE is a Trade Mark of ETSI currently being registered for the benefit of its Members and of the 3GPP Organizational Partners. GSM and the GSM logo are Trade Marks registered and owned by the GSM Association. ETSI ETSI TR 102 720 V1.1.1 (2009-10) 3Contents I
8、ntellectual Property Rights 4g3Foreword . 4g3Introduction 4g31 Scope 5g32 References 5g32.1 Normative references . 5g32.2 Informative references 5g33 Abbreviations . 6g34 Introduction 6g35 Serialisation delay 6g36 Prioritisation . 7g37 Measurement examples 9g38 Theoretical considerations 10g39 Concl
9、usions 11g3History 12g3ETSI ETSI TR 102 720 V1.1.1 (2009-10) 4Intellectual 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-member
10、s, 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.as
11、p). 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 pre
12、sent document. Foreword This Technical Report (TR) has been produced by ETSI Technical Committee Speech and multimedia Transmission Quality (STQ). Introduction Limited bandwidth is a major source for delay variation in packet networks. This is especially the case for residential customers with a low
13、 physical bandwidth connection to their ISP, and it is of concern for VoIP services. Why this is the case, even with a perfect implemented prioritisation, is shown in the present document. There are also some numbers shown to get a feeling of the impact a low bandwidth connection has in respect of d
14、elay variation. The content of the present document is valid for unshared access lines. ETSI ETSI TR 102 720 V1.1.1 (2009-10) 51 Scope The intention of the present document is to provide guidance for VoIP transmission planners and SDOs in the area of delay and delay variation, especially on access l
15、ines. The present document provides an introduction on the effect of different IP services on lines with limited bandwidth (e.g. DSL). It explains the mechanism of serialisation delay, gives a (very general) overview over prioritisation and shows how the maximum delay variation due to concurrent tra
16、ffic can be calculated. The calculations shown in the present document are valid for unshared lines only, shared lines are excluded. 2 References References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For a specific reference, subs
17、equent revisions do not apply. Non-specific reference may be made only to a complete document or a part thereof and only in the following cases: - if it is accepted that it will be possible to use all future changes of the referenced document for the purposes of the referring document; - for informa
18、tive references. Referenced documents which are not found to be publicly available in the expected location might be found at http:/docbox.etsi.org/Reference. NOTE: While any hyperlinks included in this clause were valid at the time of publication ETSI cannot guarantee their long term validity. 2.1
19、Normative references The following referenced documents are indispensable for the application of the present document. For dated references, only the edition cited applies. For non-specific references, the latest edition of the referenced document (including any amendments) applies. Not applicable.
20、2.2 Informative references The following referenced documents are not essential to the use of the present document but they assist the user with regard to a particular subject area. For non-specific references, the latest version of the referenced document (including any amendments) applies. i.1 ITU
21、-T Recommendation Y.1540: “Internet protocol data communication service - IP packet transfer and availability performance parameters“. ETSI ETSI TR 102 720 V1.1.1 (2009-10) 63 Abbreviations For the purposes of the present document, the following abbreviations apply: ATM Asynchronous Transfer Mode DS
22、L Digital Subscriber Line IP Internet ProtcolIPDV IP Packet Delay Variation ISP Internet Service Provider PVC Permanent Virtual Connection SDO Standards Development Organization VoIP Voice over Internet Protocol 4 Introduction For some time now, the coverage of broadband accesses for customers is ge
23、tting higher and higher. These broadband accesses are used for different services, starting with Internet, nowadays more and more also TV and voice services. Packet based networks offer a high flexibility to deliver all of these services over the same network. If more than one service is used at the
24、 same time or if one service uses more than one session at a time, there is the possibility of interference. One very real effect will be the influence of other services/sessions on VoIP-media traffic. 5 Serialisation delay Serialisation delay of a packet is the time it takes to clock every bit of a
25、 packet onto the line. A packet ready to be sent will be normally put in a play out buffer, from where it will be clocked onto the line with the physical line speed (see figure 1). 11101110100010000100110111001100111110011 0 1 0 1 1 0 1 0 0 0 1 1 0Play out bufferLineLinespeedFigure 1: Serialisation
26、delay The formula to calculate the serialisation delay tserialisationis as follows: /sBitLinespeedBitPacketsizestionSerialisat= ETSI ETSI TR 102 720 V1.1.1 (2009-10) 7With: Packetsize = size of a packet on the physical layer. Linespeed = linespeed on the physical layer. Table 1 shows some example ca
27、lculations. NOTE: It is recommended to do this calculation on the physical layer even if it could be done on any other layer, as long as the packet size and the line speed are calculated for the same layer. It has to be taken into account that the packet size need to represent the size of a packet,
28、including all headers and trailers (for further calculations it may also be necessary to include the minimal distance between two packets). Table 1: Example of serialisation delays with different linespeed and packetsizes Linespeed Packetsize Serialisation delay 1 Gbit/s 1 500 Bytes 0,012 ms 1 GBit/
29、s 200 Bytes 0,0016 ms 100 MBit/s 1 500 Bytes 0,12 ms 100 MBit/s 200 Bytes 0,016 ms 10 MBit/s 1 500 Bytes 1,2 ms 10 MBit/s 200 Bytes 0,16 ms 1 MBit/s 1 500 Bytes 12 ms 1 MBit/s 200 Bytes 1,6 ms 100 kbit/s 1 500 Bytes 120 ms 100 kbit/s 200 Bytes 16 ms 6 Prioritisation The general concept of prioritisa
30、tion is that traffic with higher priority is favoured against traffic with lower priority on the same line. Prioritisation is important in the case where a bandwidth limitation exists (more input capacity than output capacity) or in case of congestion (these two effects can be related). The prioriti
31、sation is a strong way of queue management (strict priority), which means that this traffic is prioritised in any case, or it can be used a weaker bandwidth allocation (like a weighted fair queuing) to allow also lower “prioritised“ traffic to pass even in the case of congestion. Normally VoIP media
32、 traffic is in the highest priority class which will use strict priority queues (as a matter of fact, internal network control traffic will be even higher prioritised). For the reminder of the traffic there are several priority classes possible, which normally will use a fair queuing. ETSI ETSI TR 1
33、02 720 V1.1.1 (2009-10) 8Figure 2 shows the functional diagram of a typical prioritisation algorithm in network equipment. W1W2W3Strict priorityClass Xe.g. AFxClass Ye.g. AFyClass Ze.g. BERealtime TrafficLinePlayout BufferFigure 2: Prioritisation principle Classes X, Y and Z are three differently pr
34、ioritised classes (in the picture, classes X and Y are two different Assured Forwarding classes, while class Z is a Best Effort class). In this case the prioritisation is according to different weights (W1, W2 and W3) of the classes, but every class will get its time slot to send (based on the weigh
35、ting). This means, that it is possible that a lowest prioritised packet from the queue of class Z can be sent to the playout buffer, even if there are packets in the higher prioritised queues of class X and Y. Since the realtime class is allocated strict priority, no packet out of the queues X, Y an
36、d Z can be sent to the playout buffer, if there is a packet in the realtime queue (this implicitly means that there has to be some sort of bandwidth control for the realtime class, to avoid a blockage of all other traffic). In the playout buffer, there is no prioritisation. Packets in the playout bu
37、ffer will be sent to the line, in the order they arrived (FIFO, first in, first out). If the playout buffer is full, not even a packet from the realtime queue can be sent to it. The prioritised packet from the realtime queue has to wait until the packet in the playout buffer is sent to the line. Wit
38、h the formula for the serialisation delay, it can be calculated, how much time this takes. This effect leads to delay variation for realtime traffic. In a usual priority implementation the playout buffer has the capability to hold two packets; this means that in the worst case, two low prioritised p
39、ackets have to be sent, before a high priority packet can be sent. The formula for the maximum delay variation tdelayvariationdue to this effect will be the one for the serialisation delay multiplied with the number of packets, which the playout buffer can hold. NOTE 1: In difference to the formula
40、for the serialisation delay, where the actual packet size of the packet under interest has to be taken, in the formula for the maximum delay variation, the maximum packet size possible on the link has to be taken. /varsBitLinespeedBitizeMaxPacketstsNbrofPackestiationdelay= ETSI ETSI TR 102 720 V1.1.
41、1 (2009-10) 9With: NbrofPackets = max. number of packets in the playout buffer. MaxPacketsize = maximum size of a packet on the link (physical layer). Linespeed = linespeed on the physical layer. NOTE 2: A typical maximum packet size for IP networks is around 1 500 Bytes at the IP-layer. 7 Measureme
42、nt examples Figures 3 to 5 show real IP packet delay variation measurements for a VoIP call between two DSL-Accesses, customer A with 6 400 / 640 kbit/s access speed, customer B with 4 608 / 576 kbit/s access speed. 024681012140 2 4 6 8 10121416182022242628303234363840424446Time sDelay variation(IPD
43、V)msInternet-Downloads during this timesmax. Delayvariation due to concurrent trafficFigure 3: Delayvariation (IPDV) of a call A to B, with parallel download at B Figure 3 shows the delay variation (IPDV according to i.1) over time measured at customer B side, with two intermittent Internet download
44、s also on the B side. According to the formula for the serialization delay, one Internet packet (1 500 Bytes at IP layer - app. 1 696 Bytes on the physical layer in this case) will have a serialization delay of 2,94 ms (downstream bitrate B: 4 608 kbit/s). Since the maximal delay variation measured
45、is much higher (nearly 12 ms, taken the difference of the maximum IPDV and the IPDV without Internet download), it can be assumed, that the playout buffer of the network equipment involved in the prioritisation towards the DSL-line holds up to 4 IP-packets. With a better implementation of the priori
46、tisation algorithm, IPDV could be reduced by 9 ms. 051015202530350 2 4 6 8 10121416182022242628303234363840424446Time sDelay variation(IPDV)msFigure 4: Delayvariation (IPDV) of a call B to A with parallel download at A and upload at B ETSI ETSI TR 102 720 V1.1.1 (2009-10) 10This graph shows the dela
47、y variation (IPDV according to i1) over time measured at customer A side, with a continuous download at the A side (starting at 4 s) and two intermittent uploads at the B side. The serialisation delay of one internet packet (1 500 Bytes at IP layer - app. 1 696 Bytes on the physical layer in this ca
48、se) on the A side will be 2,12 ms (downstream bitrate A: 6 400 kbit/s). Since the maximal delay variation measured is much higher (about 8 ms, taken the difference of the maximum IPDV without upload and the IPDV without up- and download), the conclusion from the previous measurement is confirmed: Th
49、e playout buffer of the network equipment involved in the prioritisation towards the DSL-line holds up to 4 IP-packets. The serialisation delay of one internet packet (1 500 Bytes at IP layer app. 1 696 Bytes on the physical layer in this case) on the B side will be 23,5 ms (upstream bitrate B: 576 kbit/s). This is the delay variation measured during the two intermittent periods of upstream traffic (taken the difference of maximum IPDV during up- and download and maximum IPDV during download). This means that the playout buffer of the customer
