1、ITU-T RECMN*G=7b4 APPENDIXrI 95 4862593 Ob339Ob 899 W INTERNATIONAL TELECOMMUNICATION UNION ITU-T TELECOMMUNICATION STANDARDIZATION SECTOR OF ITU GENERAL ASPECTS OF DIGITAL TRANSMISSION SYSTEMS Appendix I Rec. G.764 (1 1/95) PACKETIZATION GU IDE Appendix1 to ITU-T Recommendation G.764 (Previously CC
2、ITT Recommendation”) ITU-T RECMN*G*764 APPENDIX*I 95 4862573 Ob13907 725 FOREWORD The ITU-T (Telecommunication Standardization Sector) is a permanent organ of the International Telecommunication Union (). The ITU-T is responsible for studying technical, operating and tariff questions and issuing Rec
3、ommen- dations on them with a view to standardizing telecommunications on a worldwide basis. The World Telecommunication Standardization Conference (WTSC), which meets every four years, establishes the topics for study by the -T Study Groups which, in their turn, produce Recommendations on these top
4、ics. The approval of Recommendations by the Members of the ITU-T is covered by the procedure laid down in WTSC Resolution No. 1 (Helsinki, March 1-12, 1993). Appendix I to IT-T Recommendation G.764 was prepared by ITU-T Study Group 15 (1993-1996) and was approved under the WTSC Resolution No. 1 proc
5、edure on the 13th of November 1995. NOTE In this Appendix, the expression “Administration” is used for conciseness to indicate both a telecommunication administration and a recognized operating agency. O ITU 1996 All rights reserved. No part of this publication may be reproduced or utilized in any f
6、orm or by any means, electronic or mechanical, including photocopying and microfilm, without permission in writing from the ITU. ITU-T RECMN*G*769 APPENDIX*I 95 48b259L OhLL908 hhL CONTENTS I . 1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 I . 12 1.13 Introduction Summary of design issues Historical b
7、ackground . Reconstitution of speech signals . Delay equalization 1.5.1 Blind delay 1.5.2 Absolute time stamp 1.5.3 Relative time stamp . Robustness to errors 1.6.1 Sensitivity of speech to bit errors 1.6.2 CRC calculation over part of the frame 1.6.3 CRC calculation over whole frames . Congestion c
8、ontrol 1.7.1 Global control . 1.7.2 Local control . Packet loss 1.8.1 Case of speech . 1.8.2 Case or voiceband data . 1.8.3 Fill-in strategies (speech) Choice or packet size 1.9.1 Speech considerations . 1.9.2 Bit error considerations . 1.9.3 Integrated traffic 1.9.4 Recommendation G.7 64 Compressio
9、n issues . 1.10.1 Speech coding algorithms . 1.10.2 Digital speech interpolation Channel-oriented signalling Extensions . 1.12.1 Facsimile . 1.12.2 Speechhide0 synchronization . 1.12.3 Interface between the PSTN and LANs 1.12.4 Extension to new algorithms . Summary . Page 1 1 2 2 2 3 3 4 4 5 5 5 5 6
10、 6 8 8 9 9 10 10 10 10 11 11 11 11 12 12 12 13 13 14 14 References . 14 Recommendation G.764 . (1V95) i ITU-T RECMN*G-764 APPENDIX*I 95 iBb2.591 Ob11909 5TB ITU-T RECMN*G.?b4 APPENDIX+I 95 4862591 ObLLSLO 2LT Appendiu I to Recommendation G.764 Packetization Guide (Geneva, 1995) (This appendix does n
11、ot form an integral part of this Recommendation) 1.1 Introduction This appendix summarizes the current views on speech packetization in Study Group 15 of the IT-T during the study period 1992-1996 and in CClT SG XV during the study period 1988-1992. These views may change in the future. This appendi
12、x does not address the following topics: 1) the different treatments within the network that may be accorded to packets depending on their assignment to priority classes, in general; however, it is agreed that the network should give priority to speech over digital data to reduce delay and delay var
13、iability, and to decouple bursty traffic from reai- time traffic; 2) quality of service for different traffk classes. The purpose of this appendix is to: 0 explain the various issues that affect the packetization of speech; 0 provide an overview of the techniques and considerations in the transport
14、of packetized speech, that are used in Recommendation G.764: 0 disseminate information on the various topics of concern to designers, implementors of packetized speech equipment and to the service providers that use them. 1.2 Historical background Traditionally voice services have been implemented i
15、n the Public Switched Telephone Network (PSTN) (also denoted as Wide Area Network or WAN) using a circuit-oriented approach. The growth of packet transport techniques e.g. Recommendation X.25K.75, Internet, Wideband Packet Technology, Frame Relay and the Asynchronous Transfer Mode (ATM) has stimulat
16、ed research in new techniques for the transport of speech. Packetized systems can exploit the bursty nature of traffic to multiplex different types of traffk (e.g. voice, data, video) of many users so that they can share transmission bandwidth and switching resources dynamically. Packetization facil
17、itates the integration of the different types of traffic to allow more efficient utilization of the available bandwidth and switching resources. Packetization offers more flexibility than circuit-oriented approaches because the packet header contains the necessary control information that identifies
18、, for example, the type of traffic and, where appropriate, the coding scheme. Work in speech packetization began in the CC since the middle of the 1984-1988 study period in Working Party xvm/8 and continued in Working Party XV/2 during the 1989-1992 study period. It is now continuing, in the ITU-T f
19、or wideband packet and ATM networks. The goal is to provide a uniform basis for speech packetization, with or without speech compression and speech interpolation, to facilitate the interworking of equipment from various vendors in telecommunications applications. The work in the CCI?T has lead to th
20、e Voice Packetization Protocol of Recommendation G.764 and its extensions in Recommendation G.765. These protocols are compatible at the link layer with the ISDN protocols LAPD and LAPF specified in Recommendations 4.921 and 4.922, respectively. Recommendation G.764 . (1Y95) 1 ITU-T RECMNUG.764 APPE
21、NDIXxI 95 4862593 Ob33933 356 1.3 Summary of design issues Using the OS1 protocol stack as a reference, the major design considerations in devising a speech packetization protocol are: 1) Layer I (physical layer) - The issue is whether the physical interface will conform to that of public telephone
22、networks (Recommendation G.703lG.704) or to other local area networks, such as IEEE 802.2, 802.3 or 802.9, etc. Layer 2 (link layer) - Some of the issues are: a) whether the logical layer will be compatible with ISDN (LAPDLAPF) protocols or will have the same structure as those for LANs; b) how to d
23、eal with the loss of frames; c) robustness to errors. Layer 3 (procedures to deal with digitized voice and voiceband data traffk) - Issues are: a) b) Higher-layer issues involve the speech coder and the type of compression used. 2) 3) the delay variability for speech packets; the transport of channe
24、l-associated signalling. 4) A speech packetization protocol has the following requirements: 0 The speech must be reconstnicted at the receiving end from packets arriving at irregular intervals (or, in some architectures, out of order). The protocol must be robust against line errors. It must offer a
25、n easy method for congestion control in the network. It must specify procedures at the terminating end to recover from packet loss or excessive delay. It must carry channel-associated signalling. If digital speech interpolation is used to eliminate silence intervals, it shall specify the level at wh
26、ich noise is re-injected at the terminating end. 0 0 0 0 1.4 Reconstitution of speech signais To achieve good speech quality, the terminating end must reconstitute a continuous speech stream and play it out at regular intervals despite varying packet arrival times. This involves two aspects: 1) 2) d
27、elay equalization. preserving the relative timing of information within one speech burst: and In Recommendation G.764, a packet sequence number is used to encode the relative timing of the information within one speech burst. The first packet of a speech burst always has the sequence number of O: su
28、bsequent packets in the same burst have the numbers from 1 to 15, rolling back to 1. The terminating endpoints use the packet sequence number to: 1) 2) determine the first packet of a speech burst; and to detect packet loss. The determination of the first packet is useful for delay equalization and
29、may be needed for some speech coding algorithms such as Recommendation G.728. Delay equalization is discussed in the next subclause. 1.5 Delay equalization Delays in packet communication consist of two components: a fixed delay and a variable delay i. The fixed delay arises from signal propagation o
30、n the transmission links, and from fixed processing delays in the originating and terminating endpoints and within the network. The effects of variations in the propagation delay for a given path are assumed to be negligible. 2 Recommendation G.764 (1V95) ITU-T RECMN*G.764 APPENDIXSI 95 4862593 Ob33
31、932 O92 For speech packetization, the fixed processing delays consist of the following components: packetization delay during which the speech samples are buffered for further processing; if digitai speech interpolation is used to remove silent intervals, the hang-over time of the speech detector 2;
32、 the end-teend algorithmic delay due to the encoding and decoding of speech - this delay depends on the coding scheme; for example, it is 125 p for PCM, 250 p for the Adaptive Differential Pulse Coded Modulation (ADPCM) algorithms of Recommendations G.726 and G.727 while it is less than 2 ms for the
33、 Low-Delay Code Excited Linear Predictor (LD-CELP) algorithm of Recomendation G.728; any added delay at the terminating end to mask the timing jitter resulting from the variability in the delay - this added delay is denoted as build-out. Variable delays result primarily from the queueing and process
34、ing of packets. They depend on the characteristics of the route of each packet: the number of hops (nodes), the type and speed of each link and the traffic intensity. Speech traffic requires low and uniform delay. Recommendation G. 114 (1993) discusses the effect of end-to-end delays on the quality
35、of a conversation. Delay variations that may, be acceptable for digital data transmission, usually affect the gaps between words and syllables and are troublesome for conversational speech. Available data suggest that variation in the interval between speech bursts should be less than 200 ms to avoi
36、d subjective degradation of speech quality. In applications such as videotelephony where both audio and video information are transmitted and must remain synchronized, the effect of the variable delay on this synchronization must also be taken into account. There are several methods to mask the vari
37、ability of the delay in the network. These techniques include: 1) blind delay; 2) absolute time stamp; and 3) relative time stamp. The effect of all these methods is to increase the effective end-to-end buffering delay and, therefore, the total end-to-end delay. 1.5.1 Blind delay In the blind delay
38、method, a fixed buffer delay is always added at the terminating end at the fust packet of a speech burst. This delay corresponds to the maximum variable delay expected. The advantage of the blind delay scheme is its simplicity which makes it a good candidate when the transmission speech is such that
39、 the variable delays are of the order of a fraction of a millisecond (e.g. local area networks or broadband networks at 150 Mbit/s). In these cases, a fixed build-out delay on the order of 10 ms will be adequate to eliminate end-to-end delay jitter 3. Over long haul connections in the PSTN. the sche
40、me may require too large a delay that the total end-to-end delay would exceed the performance limits for network delays specified in Recommendation G. 114. For example, if the first packet has already experienced the worst case delay variation, the total variable delay to be added will be twice the
41、worst case value l. Because this approach does not mask the delay variability completely, the gaps between the words may be varying which may degrade the subjective quality of speech. Voiceband traffic, included demodulated facsimile, may be perturbed; total delays larger than 500 ms cause premature
42、 disconnections of facsimile calls, especially when echo is present 4. 1.5.2 Absolute time stamp This is the method used in datagram networks. The packet header includes a field for a time stamp that represents real time. The time stamp has a resolution sufficient to allow accurate detection of pack
43、et jitter, and to cover the worst case transit time of a packet across the network. Thus, packets that arrive out of sequence may be correctiyordered and buffered at the receiver using time stamp information 5. This datagram scheme also requires clock synchronization between the transmitter and rece
44、iver so that the delay of each incoming packet can be compared to the previous ones, assuming that the network delay is fixed. Recommendation 6.764 + (1V95) 3 ITU-T RECMNrG.764 APPENDIXtI 95 4862.591 Ob11913 T29 1.5.3 Relative time stamp In the relative time stamp method, an estimate of the play-out
45、 time is obtained for the first packet of a speech burst, for all signalling packets and for the first packet after a missing packet. This time is then used to adjust the delay of all remaining packets of this burst conveyed on that virtual circuit. The accumulated variable delay experienced by a pa
46、cket is recorded in the time stamp field of the packet header i. Each network node adds to the time stamp the amount of time it took to serve a packet before sending it, using its local clock as reference. The maximum allowable variable delay for a virtual circuit is specified as the build-out. The
47、build-out is defined for a given virtual circuit. Once an estimate of the play-out time has been made, subsequent packets are placed in the order of their sequence number in the play-out buffer and then held for the following duration: time before play-out = build-out delay -time stamp value The ter
48、minating endpoint must, therefore, store the speech packets that arrive before their scheduled play-out time and then play them at regular intervals. Packets whose time stamp field exceeds the build-out delay are considered late and are dropped. The relative time stamp method is less complicated tha
49、n the absolute time stamp method of 1.5.2 when virtual circuits are used to guarantee that the packets remain in sequence. The scheme does not rely on clock synchronization between the endpoints and places the timing function in layer 3. Furthermore, the delay measurement in each packet can be used to detect network congestion and then to invoke overload management strategies. In a multisite conference, the introduction of build-out delay improves the conversational dynamics because it ensures that the playback of the speech to all parties is synchronized.
