1、Report No. 27 A Technical ReportonNovember 1993 Echo CancellingPrepared byT1A1.6 Working Groupon SpecializedSignal ProcessingCommittee T1 is sponsored by the Alliance for Telecommunciations Industry Solutions(formerly the Exchange Carriers Standards Association)Accredited by American National Standa
2、rds InstituteCopyright 1993 by Alliance for Telecommunications IndustrySolutions All rights reserved.No part of this publication may be reproduced in any form, in an electronic retrieval system or otherwise, without the prior written permission of the publisher.A Technical Report onECHO CANCELLINGAB
3、STRACTThis technical report summarizes T1s current views on the principles of the operationof echo cancellers and their applications. It includes a historical background for the useof echo control devices in the Public Switched Telecommunication Network (PSTN). Italso reviews application rules and o
4、perational constraints for public/private networktransmission planning. The report discusses the relationship among planners ofPSTNs, modem manufacturers, private network planners, and end users regarding thecontrol of echo (from sources inside and outside the PSTN) and the associated terminalconsid
5、erations.This report reviews the effect of echo cancellers on voiceband data, especially theirinteraction with facsimile transmission. It also discusses additional issues, such as theeffect of high signal levels. Service considerations are reviewed from the perspective ofpotential network evolutions
6、. These considerations include communication withswitches, bit transparency, effect of nonlinearities in tail circuits, voice compressionbetween cancellers, and the tandeming of echo cancellers. Other topics discussed areconvergence speed, acoustic echo control, comfort noise, and the specialconside
7、rations needed when Digital Circuit Multiplication Equipment (DCME) andPacket Circuit Multiplication Equipment (PCME) are used.Document T1A1/93-012Prepared byT1A1.6Working Group on Specialized Signal ProcessingForewordThis technical report was written as part of the activities of T1 Subcommittee T1A
8、1.6under project T1Y1 28, entitled “Digital Techniques in Voice Networks.” The editorialcommittee consisted of the following members:Maurice GivensTellabsMostafa Hashem Sherif (Committee Chair)AT they are expectedto replace echo suppressors in modern telecommunication networks. Echo cancellersare in
9、creasingly present on nearly every long distance connection and may beencountered singly or in tandem on a given connection.This technical report represents the current T1 views on network echo cancellers.These views may change in the future.As a result of the recent reorganization of the Internatio
10、nal Telecommunication Union(ITU), the CCITT was restructured and its name changed to the TelecommunicationStandardization Sector (ITU-T). To avoid confusion, the old nomenclature has beenretained throughout this report.1.2 PurposeThe purpose of this technical report is to:1. explain the general prin
11、ciples of the operation of echo cancellers;2. identify a limited set of application rules and the constraints under which echocancellers operate;3. explain the relationship among the roles of the planners of a Public SwitchedTelecommunications Network (PSTN), modem manufacturers, private networkplan
12、ners, and end users regarding the control of echo (from sources inside andoutside the PSTN) and the associated terminal design considerations;4. identify how echo cancellers may affect the perceived quality of speech, the qualityof voiceband data, as well as the performance of various signal process
13、ing systems(such as digital and packetized circuit multiplication systems);5. identify both public and private network changes that may require additional studyof echo cancellers, to fully understand how these changes may impact thefunctionality of present echo cancellers; and6. explain how new serv
14、ices, if accepted for implementation, could have anevolutionary impact on echo canceller design.1.3 ApplicationThis technical report is limited to echo cancellers that have a digital input and digitalsubtractors. CCITT refers to these as Type C echo cancellers.In writing this report, the purpose of
15、Committee T1A1.6 was not to develop a standard,but to make the technical information on echo cancellers available to thetelecommunications industry. Accordingly, the use of “should” or “recommend”TECHNICAL REPORT NO. 272throughout the report is not to be construed as implying any requirements. Reade
16、rsmay wish to contact Committee T1A1.6 for any updates prior to using the information inthis report.2. Normative references and definitionsThe following standards are referenced in this report. At the time of this publication, theeditions indicated were valid. All standards are subject to revision,
17、and readers areencouraged to investigate the possibility of applying the most recent editions of thestandards indicated below.ANSI T1.303-1989, Digital Processing of Voice-Band Signals Algorithm for 24, 32, and40 kbit/s ADPCMANSI T1.508-1992, Network Performance - Loss Plan for Evolving Digital Netw
18、orksCCITT Recommendation G.114, Mean one-way propagation time, Blue Book, VolumeIII, Fascicle III.1, pp. 84-94CCITT Recommendation G.131, Influence of national systems on stability, talker echo,and listener echo in international connections, Blue Book, Volume III, Fascicle III.1, pp.143-155CCITT Rec
19、ommendation G.161, Echo-suppressors suitable for circuits having eithershort or long propagation times, Blue Book, Volume III, Fascicle III.1, pp. 178 Also Vol.III, Green Book, Geneva, 1977CCITT Recommendation G.164, Echo suppressors, Blue Book, Volume III, FascicleIII.1, pp. 186-221CCITT Recommenda
20、tion G.165, Echo cancellers, Blue Book, Volume III, Fascicle III.1,pp. 225-243CCITT Recommendation G.711, Pulse Code Modulation (PCM) of Voice Frequencies,Blue Book, Volume III, Fascicle III.3, pp. 175-184CCITT Recommendation P.31, Transmission characteristics for digital telephones,Blue Book, Volum
21、e V, pp. 55-56CCITT Recommendation T.30, Procedures for document facsimile transmission in thegeneral switched telephone network, Blue Book, Volume VII, Fascicle VII.3, pp. 77-174CCITT Recommendation V.21, 300 bits per second duplex modem standardized foruse in the general switched telephone network
22、, with V-series type interfaces withprovision for statistical multiplexing, Blue Book, Volume VIII, Fascicle VIII.1, pp. 65-69CCITT Recommendation V.23, 600/1200-baud modem standardized for use on thegeneral switched telephone network, Blue Book, Volume VIII, Fascicle VIII.1, pp. 98-104TECHNICAL REP
23、ORT NO. 273CCITT Recommendation V.26 ter, 2400 bits per second duplex modem using the echocancellation technique standardized for use on the general switched telephone networkand on point-to-point 2-wire leased telephone-type circuits, Blue Book, Volume VIII,Fascicle VIII.1, pp. 165-180CCITT Recomme
24、ndation V.27 ter, 4800/2400 bits pers second modem standardized foruse in the general switched telephone network, Blue Book, Volume VIII, Fascicle VIII.1,pp. 199-211CCITT Recommendation V.29, 9600 bits per second modem standardized for use onpoint-to-point 4-wire leased telephone-type circuits, Blue
25、 Book, Volume VIII, FascicleVIII.1, pp. 215-228CCITT Recommendation V.32, A family of 2-wire, duplex modems operating at datasignalling rates of up to 9600 bit/s for use on the general switched telephone networkand on leased telephone-type circuits, Blue Book, Volume VIII, Fascicle VIII.1, pp. 234-2
26、51Communications Canada Terminal Attachment Program, Certifications Standard CS-03, Issue 7, Release 3, February 1990The Office of the Federal Register National Archives and Records Administration, Codeof Federal Regulations Title 47, Part 68 Connection of Terminal Equipment to theTelephone Network,
27、 October 1, 1992.3. Historical background3.1 PSTN transmission planningIn the telephone network, an access line provides the communication path between acustomer premises and the local switching system. Typically, all or a portion of theaccess line is composed of a 2-wire facility. However, transmis
28、sion facilities betweenswitching systems are almost always 4-wire. Wherever a conversion from 4-wire to 2-wire takes place within a connection, a perfect impedance match cannot be achievedand thus a return signal, referred to as echo, results. Echoes of sufficient amplitudesand delay are objectionab
29、le. Therefore, one of the major concerns of the PSTNplanners is to ensure adequate echo control to provide satisfactory transmissionperformance.For low-delay connections, echo is controlled by the insertion of appropriatetransmission path losses, as defined in ANSI T1.508. Longer delay connections n
30、eedecho control devices. It is the PSTN planners role to design PSTNs so that the echocontrol devices installed provide adequate control of the echo from the 4-wire-to-2-wireconversions in the PSTN, and to ensure that the customer obtains satisfactorytransmission performance.In the past, echo suppre
31、ssors were used to control echo in long distance networks.Today, however, the echo canceller is the device of choice. The following twosubsections summarize the reasons for the use of echo cancellers instead of echosuppressors in modern telephone networks.1, 2, 3, 4, 5TECHNICAL REPORT NO. 2743.2 Ech
32、o suppressorsThe principle of echo suppressors is well-known; it is summarized as follows: Whenspeech is detected on the receive path, a very high attenuation is inserted in the sendpath. When double-talk is detected, the send path is closed and a receive loss isinserted in the receive path. Thus, d
33、uring double-talk, there is no echo suppression,but the echo is much more attenuated than the direct speech. Other refinements arepossible, as indicated in CCITT Recommendation G.164.Many problems can occur in the operation of echo suppressors; this is because thedecision as to which end is talking
34、and which is listening is based essentially on thesignal levels. If the level of the echo is high and the level of direct speech is low,speech could be mutilated and/or it could be difficult to distinguish between single-talkand double-talk. This could also be the case at the beginning or at the end
35、 of a speechburst.The problems are compounded on long-delay transmission paths, because the patternof conversation is usually changed to account for the circuit delay. In addition, thecascading of echo suppressors is not recommended. In the case of voiceband data, a2100-Hz tone is specified to permi
36、t disabling of the echo suppressor before thebeginning of data transmission; this is for two reasons:Gb7G20 to avoid insertion losses for modems with a secondary channel;G20Gb7G20 to avoid delays due to hangover at turnarounds, thereby increasing the throughput.Facsimile is a special case. Even if a
37、n echo suppressor is disabled by 2100-Hz tone, itmay be re-enabled during a facsimile transmission. The tone disabler hangover time ofan echo suppressor is specified as 250 150 ms in 5.7 of CCITT RecommendationG.164 (Blue Book). Therefore, periods of silence greater than 100 ms at the echosuppressor
38、 may cause the echo suppressor to be re-enabled, while periods greaterthan 400 ms cause it to be re-enabled. During a facsimile call, there are a number ofsilent periods that may be long enough to permit the re-enabling of the echosuppressor. In addition, some facsimile manufacturers have chosen to
39、exceed thesignal separation intervals specified in CCITT Recommendation T.30; therefore, echosuppressors may be re-enabled.Enabled echo suppressors may distort the facsimile signals. One type of distortion isthe truncation of fast turnaround signals. Typically, the echo suppressor operates in asingl
40、e-talk mode, so that when a signal arrives at the receive port, the suppressionswitch is activated and remains in that state until no signal arrives for a certain time.The recommended hangover time associated with each state transition is in the rangeof 24 to 36 ms1, as specified in Table 4/G.164. T
41、he suppression hangover time guardsagainst echo stored in the local echo path.Now, CCITT Recommendation T.30 specifies that the guard time between V.21 andV.29 transmission should be 75 20 ms. If a return signal from the local facsimileG20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G
42、20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G20G201Analogue echo suppressors conforming to CCITT Recommendation G.161 are still in service; they have a suppressionhangover time of 40 to 75 ms (see Table 4/G.161). Accordingly, a signal may be mutilat
43、ed if it reaches the transmit portbefore 40 to 75 ms. In 1992, the CCITT declared Recommendation G.161 to be obsolete.TECHNICAL REPORT NO. 275machine, within a V.21 message-response sequence or a V.21/V.29 sequence such asa confirmation to receive (CFR) followed by training, reaches the echo suppres
44、sortransmit port within 24 to 36 ms of the termination of the signal at the receive port, thepersistence of echo suppression insertion losses or open-circuit condition mayintroduce an attenuation. As a result, the echo suppressor mutilates the initial portionof that fast turnaround signal. When this
45、 signal is part of the training/training checksignal, training might be disrupted and rate fallback ensues, or in a worse case, the callis terminated.Similarly, an enabled echo suppressor may block a low-level secondary channel signal.If the level of that signal is high enough, the suppressor may en
46、ter the double-talkmode, in which a receive loss is inserted. The result is a reduction in the levels of boththe transmit and the receive signals, if echo suppressors are at both ends of theconnection and are both in the double-talk mode.Finally, for certain combinations of propagation times and ins
47、ertion losses, listener echomay cause the 2100-Hz tone to persist long enough to disable the echo suppressors.This echo may then contribute to the degradation of the image quality by reducing thesignal-to-noise ratio during page transmission.3.3 Echo cancellersEcho cancellers are devices that use ad
48、aptive signal processing to reduce or eliminateechoes. Echo cancellers are placed in the 4-wire portion of a circuit, and reduce (orcancel) the echo by subtracting an estimate of the echo from the returned echo signal.Echo cancellers may operate on a single circuit or on a multiplexed facility, e.g.
49、, echocancellers operate on a 64-kbit/s speech facility that is multiplexed into a primary ratelink.Echo cancellers are designed to:Gb7G20 cancel linear echo path signals;G20Gb7G20 refrain from cancelling the echo when requested to do so by an in-band disablingsignal;G20Gb7G20 return to an operational mode after being disabled when the in-band signal powerlevel drops below a specified level for a specified period of time. This design allowssome networks to transport voiceband data on the same speech channels. It alsoallows the echo canceller to re-enable during a voice call a
