[e2e] Re: crippled Internet

Colin Perkins csp at ISI.EDU
Thu Apr 26 07:04:34 PDT 2001


--> Randy Bush writes:
>> ]    According to G.114, one-way delays up to 150 ms should cause no problems
>> ] with typical voice communications.  From 150 to 400, loss/noise/etc. become
>> ] a bigger issue and "long-delay" echo cancelers are needed.  (But some form
>> ] of echo canceler/suppression is needed on ALL delays beyond 25 ms.)
>> ] ...
>
>i admit i am blind and stupid, but i just can't wrap my head around how this
>applies to voip.  echo cancellation?
>
>> ]   So I'd summarize by saying that human factors in dealing with one-way
>> ] delay sets in at about 150-250 ms.
>
>again, in my limited experience, the speed of light in fiber and copper are
>dominate the mode of the measured delays.  though i could imagine that
>codecs and the like will make their contribution.

And the packetisation delay, since you can't send a packet until you've
accumulated a full frame of audio (this gives 20-30ms of unavoidable delay
at the sender). 

Right now, I see around 100 ms round trip time between the US East coast
and London (19 hops, via PSInet, Teleglobe and JANET). Making a few rash
assumptions gives 50 ms one way, add 30 ms for packetisation, and that 
still leaves 70 ms for de-jitter, codecs, etc., if you think 150 ms one
way delay is the minimum acceptable.

Seems do-able to me... 

>> Given the sort of digital processing against echos that is now routinely
>> done for good speakerphones and the CPU cycles available in modern
>> personal computers, I wonder if the old worries about echo cancelers are
>> irrelevant.  There's also the fact that everything is digital in VoIP.
>
>it seems more relevant to me that voip is two *separated* one way paths.
>what's to echo?

Even if there's nothing to echo, human perception becomes a factor once the
round trip time gets close to human response time. 

Colin



More information about the end2end-interest mailing list