0% found this document useful (0 votes)
67 views17 pages

ETE405:: Lecture 9: IP Telephony In-Depth Analysis

The document discusses several key issues related to IP telephony: 1) It describes the three main types of delay in telephony networks: propagation, serialization, and handling delay. Propagation delay is caused by the speed that signals travel through fiber or copper. 2) Voice compression standards like G.711, G.722, G.723.1, and G.728 are discussed which allow voice to be encoded into different bit rates. 3) Packet loss is common in data networks and voice protocols need mechanisms to handle variations in packet arrival times and potential lost packets.

Uploaded by

zicobaby
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
67 views17 pages

ETE405:: Lecture 9: IP Telephony In-Depth Analysis

The document discusses several key issues related to IP telephony: 1) It describes the three main types of delay in telephony networks: propagation, serialization, and handling delay. Propagation delay is caused by the speed that signals travel through fiber or copper. 2) Voice compression standards like G.711, G.722, G.723.1, and G.728 are discussed which allow voice to be encoded into different bit rates. 3) Packet loss is common in data networks and voice protocols need mechanisms to handle variations in packet arrival times and potential lost packets.

Uploaded by

zicobaby
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

Page 120-135

ETE405 :: Lecture 9 Chapter 8 IP telephony in-Depth Analysis

Introduction
To create a proper network design, it is important to know all the caveats and inner workings of networking technology This chapter e!plains many of the issues facing "oice over #$ %"o#$& The following issues are covered in this chapter:

'elay(latency )itter 'igital sampling "oice compression Echo $acket loss "oice activity detection 'igital*to*analog conversion Tandem encoding Transport protocols 'ial*plan design

2/22/10

Delay/Latency
oIP delay or latency is characteri!ed as the a"ount o# ti"e it ta$es #or speech to e%it the spea$er&s "outh and reach the listener&s ear' (hree types o# delay are inherent in today&s telephony net)or$s* +1, propagation delay* caused -y the speed o# light in #i-er or copper--ased net)or$s' +2, seriali!ation delay +3, handling delay.also called processing delay .de#ines "any di##erent causes o# delay +actual pac$eti!ation/ co"pression/ and pac$et s)itching, and is caused -y de0ices that #or)ard the #ra"e through the net)or$'

Propagation Delay
Light tra0els through a 0acuu" at a speed o# 181/000 "iles per second/ and electrons tra0el through copper or #i-er at appro%i"ately 125/000 "iles per second' A #i-er net)or$ stretching hal#)ay around the )orld +13/000 "iles, induces a one-)ay delay o# a-out 20 "illiseconds +20 "s,' Although this delay is al"ost i"percepti-le to the hu"an ear/ propagation delays in con3unction )ith handling delays can cause noticea-le speech degradation'

4andling Delay
As "entioned pre0iously/ de0ices that #or)ard the #ra"e through the net)or$ cause handling delay' 4andling delays can i"pact traditional phone net)or$s/ -ut these delays are a larger issue in pac$eti!ed en0iron"ents'

5ueuing Delay
A pac$et--ased net)or$ e%periences delay #or other reasons' ()o o# these are the ti"e necessary to "o0e the actual pac$et to the output 6ueue +pac$et s)itching, and 6ueuing delay' 7hen pac$ets are held in a 6ueue -ecause o# congestion on an out-ound inter#ace/ the result is 6ueuing delay' 5ueuing delay occurs )hen "ore pac$ets are sent out than the inter#ace can handle at a gi0en inter0al'

8itter
9i"ply stated/ 3itter is the 0ariation o# pac$et interarri0al ti"e' 8itter is one issue that e%ists only in pac$et--ased net)or$s' 7hile in a pac$et 0oice en0iron"ent/ the sender is e%pected to relia-ly trans"it 0oice pac$ets at a regular inter0al +#or e%a"ple/ send one #ra"e e0ery 20 "s,' (hese 0oice pac$ets can -e delayed throughout the pac$et net)or$ and not arri0e at that sa"e regular inter0al at the recei0ing station +#or e%a"ple/ they "ight not -e recei0ed e0ery 20 "s,' (he di##erence -et)een )hen the pac$et is e%pected and )hen it is actually recei0ed is 3itter'

ariation o# Pac$et Arri0al (i"e +8itter,

you can see that the a"ount o# ti"e it ta$es #or pac$ets A and : to send and recei0e is e6ual +D1;D2,' Pac$et C encounters delay in the net)or$/ ho)e0er/ and is recei0ed a#ter it is e%pected' (his is )hy a 3itter -u##er / )hich conceals interarri0al pac$et delay 0ariation/ is necessary'

<ean =pinion 9core


>ou can test 0oice 6uality in t)o )ays* su-3ecti0ely and o-3ecti0ely' 4u"ans per#or" su-3ecti0e 0oice testing/ )hereas co"puters. )hich are less li$ely to -e ?#ooled? -y co"pression sche"es that can ?tric$? the hu"an ear.per#or" o-3ecti0e 0oice testing' Codecs are de0eloped and tuned -ased on su-3ecti0e "easure"ents o# 0oice 6uality' 9tandard o-3ecti0e 6uality "easure"ents/ such as total har"onic distortion and signal-to-noise ratios/ do not correlate )ell to a hu"an&s perception o# 0oice 6uality/ )hich in the end is usually the goal o# "ost 0oice co"pression techni6ues'

A co""on su-3ecti0e -ench"ar$ #or 6uanti#ying the per#or"ance o# the speech codec is the "ean opinion score +<=9,' <=9 tests are gi0en to a group o# listeners' :ecause 0oice 6uality and sound in general are su-3ecti0e to listeners/ it is i"portant to get a )ide range o# listeners and sa"ple "aterial )hen conducting a <=9 test' (he listeners gi0e each sa"ple o# speech "aterial a rating o# 1 +-ad, to 5 +e%cellent,' (he scores are then a0eraged to get the "ean opinion score' <=9 testing also is used to co"pare ho) )ell a particular codec )or$s under 0arying circu"stances/ including di##ering -ac$ground noise le0els/ "ultiple encodes and decodes/ and so on' >ou can then use this data to co"pare against other codecs'

I(@-( codec <=9 9coring

"oice +ompression

()o -asic 0ariations o# 1A B-ps PC< are co""only used* C-la) and a-la)' (he "ethods are si"ilar in that they -oth use logarith"ic co"pression to achie0e 12 to 13 -its o# linear PC< 6uality in 8 -its/ -ut they are di##erent in relati0ely "inor co"pression details +C-la) has a slight ad0antage in lo)-le0el/ signal-to-noise ratio per#or"ance,' @sage is historically along country and regional -oundaries/ )ith Dorth A"erica using C-la) and Europe using a-la) "odulation' It is i"portant to note that )hen "a$ing a long-distance call/ any re6uired C-la) to a-la) con0ersion is the responsi-ility o# the C-la) country'

2/22/10

"oice +oding ,tandards

(he "ost popular 0oice coding standards #or telephony and pac$et 0oice include*

F'211.Descri-es the 1A B-ps PC< 0oice coding techni6ue outlined earlierG F'211-encoded 0oice is already in the correct #or"at #or digital 0oice deli0ery in the pu-lic phone net)or$ or through Pri0ate :ranch eHchanges +P:Hs,' F'221.Descri-es ADPC< coding at A0/ 32/ 2A/ and 11 B-psG you also can interchange ADPC< 0oice -et)een pac$et 0oice and pu-lic phone or P:H net)or$s/ pro0ided that the latter has ADPC< capa-ility' F'228.Descri-es a 11 B-ps lo)-delay 0ariation o# CELP 0oice co"pression'

2/22/10

F'22I.Descri-es CELP co"pression that ena-les 0oice to -e coded into 8 B-ps strea"sG t)o 0ariations o# this standard +F'22I and F'22I Anne% A, di##er largely in co"putational co"ple%ity/ and -oth generally pro0ide speech 6uality as good as that o# 32 B-ps ADPC<' F'223'1.Descri-es a co"pression techni6ue that you can use to co"press speech or other audio signal co"ponents o# "ulti"edia ser0ice at a lo) -it rate/ as part o# the o0erall 4'32A #a"ily o# standards' ()o -it rates are associated )ith this coder* 5'3 and 1'3 B-ps' (he higher -it rate is -ased on <P-<L5 technology and pro0ides greater 6uality' (he lo)er -it rate is -ased on CELP/ pro0ides good 6uality/ and a##ords syste" designers )ith additional #le%i-ility'

2/22/10

Echo
In a traditional toll net)or$/ echo is nor"ally caused -y a "is"atch in i"pedance #ro" the #our-)ire net)or$ s)itch con0ersion to the t)o-)ire local loop' Echo/ in the standard Pu-lic 9)itched (elephone Det)or$ +P9(D,/ is regulated )ith echo cancellers and a tight control on i"pedance "is"atches at the co""on re#lection points'

Echo Caused -y I"pedance <is"atch

2/22/10

Pac$et Loss

Pac$et loss in data net)or$s is -oth co""on and e%pected' <any data protocols/ in #act/ use pac$et loss so that they $no) the condition o# the net)or$ and can reduce the nu"-er o# pac$ets they are sending'

You might also like