Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > VOIP > Another question: some theory.

Reply
Thread Tools

Another question: some theory.

 
 
Jack L.
Guest
Posts: n/a
 
      02-23-2004
VoIP-developers know the problems VoIP applications suffer during a
conversation, eg. delay, echo in the far-end and loss of data packets.
Numerous mechanisms have been developed to remove some of them, but in the
end, the Internet cannot guarantee data packets to arrive within a specific
time which means that a VoIP conversation (currently) won't be as good as
with PSTN networks.

Would it make sense to research on whether opening X channels to the other
side and thereby sending the SAME packet X times at the same time will
ensure that data arrives to the destination? We assume that both sides have
broadband connection so we have enough bandwidth to "waste". The idea is
that we hope the packets will be routed differently so even if one packet
disappears on one of the channels, the same packet will still arrive to the
destination as the routers (hopefully) let it take another path.

The idea probably sounds crazy but just come up with your thoughts and
ideas; I am looking for a subject to my final thesis to write about.
Actually, any inspirations to a subject is highly welcome. Thanks!

--
Mvh / Best regards,
Jack, Copenhagen

The email address is for real.



 
Reply With Quote
 
 
 
 
Hank Karl
Guest
Posts: n/a
 
      02-24-2004
There are a lot of issues that can cause VoIP quality problems.
Packet loss is one; Jitter and delay are others.

Broadband connections don't always have bandwidth to waste--ADSL may
be 1.5M down, but only 128K up. If you are using G.711, you may need
over 90K of the bandwidth for voice (considering the TCP/IP overhead).
If you choose the worst voice packet size, you will have a total
message size of 49 bytes. This can be a problem if your link uses ATM
(like ADSL normally does). ATM requires two 53 byte cells to send a
49 byte message (ATM uses a 5 byte header and 48 byte payload per
cell), in which case the ADSL bandwidth utilization just about
doubles..

So if you are losing packets because any link is congested, sending
the same packet four times will be counter-productive.

Assuming that you can get the four different channels to be routed
differently, sending the same packet four times over four different
links will give you delay problems, you would be setting your delay to
that of the slowest link.

For more information on VoIP voice quality, see

http://www.voiptroubleshooter.com

and http://www.telchemy.com/techref.html

Regards,
Hank

On Tue, 24 Feb 2004 00:51:53 +0100, "Jack L." <(E-Mail Removed)>
wrote:

>VoIP-developers know the problems VoIP applications suffer during a
>conversation, eg. delay, echo in the far-end and loss of data packets.
>Numerous mechanisms have been developed to remove some of them, but in the
>end, the Internet cannot guarantee data packets to arrive within a specific
>time which means that a VoIP conversation (currently) won't be as good as
>with PSTN networks.
>
>Would it make sense to research on whether opening X channels to the other
>side and thereby sending the SAME packet X times at the same time will
>ensure that data arrives to the destination? We assume that both sides have
>broadband connection so we have enough bandwidth to "waste". The idea is
>that we hope the packets will be routed differently so even if one packet
>disappears on one of the channels, the same packet will still arrive to the
>destination as the routers (hopefully) let it take another path.
>
>The idea probably sounds crazy but just come up with your thoughts and
>ideas; I am looking for a subject to my final thesis to write about.
>Actually, any inspirations to a subject is highly welcome. Thanks!


 
Reply With Quote
 
 
 
 
Andreas Sikkema
Guest
Posts: n/a
 
      02-24-2004
Hank Karl <(E-Mail Removed)> wrote in
news:(E-Mail Removed):

> (considering the TCP/IP overhead).


I think you mean UDP/IP overhead

--
Andreas
 
Reply With Quote
 
shope
Guest
Posts: n/a
 
      02-24-2004
"Jack L." <(E-Mail Removed)> wrote in message
news:Omw_b.96441$(E-Mail Removed) k...
> VoIP-developers know the problems VoIP applications suffer during a
> conversation, eg. delay, echo in the far-end and loss of data packets.
> Numerous mechanisms have been developed to remove some of them, but in the
> end, the Internet cannot guarantee data packets to arrive within a

specific
> time which means that a VoIP conversation (currently) won't be as good as
> with PSTN networks.
>
> Would it make sense to research on whether opening X channels to the other
> side and thereby sending the SAME packet X times at the same time will
> ensure that data arrives to the destination? We assume that both sides

have
> broadband connection so we have enough bandwidth to "waste". The idea is
> that we hope the packets will be routed differently so even if one packet
> disappears on one of the channels, the same packet will still arrive to

the
> destination as the routers (hopefully) let it take another path.


the problem is that packets normally arent diverse routed at the places
where losses are common - over the 1st and last hops and at peering points.

Even if this "helps" with your individual stream - imagine the effect on a
Voip based carrier where most of the traffic is voip - you are making 4
copies of the data.

You need to think about what you mean by "better" - if more packets have a
copy arrive, then you have lower loss rate, but what about the other
characteristics? if your paths are not stable and you are pushing load on
some links higher then i suspect that jitter is going to get worse not
better - or maybe better on average, but worse in the worst case. No idea
what effect that has on voip but i doubt it is good.

finally - the end point has to sort out the mess. Just using the 1st copy
packet is probably going to work.
>
> The idea probably sounds crazy but just come up with your thoughts and
> ideas; I am looking for a subject to my final thesis to write about.
> Actually, any inspirations to a subject is highly welcome. Thanks!


you may not know but voip already allows for error correction in the packets
(cisco phones for example can mask a 30 mSec error) -
>
> --
> Mvh / Best regards,
> Jack, Copenhagen
>
> The email address is for real.

--
Regards

Stephen Hope - remove xx from email to reply


 
Reply With Quote
 
Aaron Leonard
Guest
Posts: n/a
 
      02-24-2004
Good thought. Look into rtp redundancy.

---

~ VoIP-developers know the problems VoIP applications suffer during a
~ conversation, eg. delay, echo in the far-end and loss of data packets.
~ Numerous mechanisms have been developed to remove some of them, but in the
~ end, the Internet cannot guarantee data packets to arrive within a specific
~ time which means that a VoIP conversation (currently) won't be as good as
~ with PSTN networks.
~
~ Would it make sense to research on whether opening X channels to the other
~ side and thereby sending the SAME packet X times at the same time will
~ ensure that data arrives to the destination? We assume that both sides have
~ broadband connection so we have enough bandwidth to "waste". The idea is
~ that we hope the packets will be routed differently so even if one packet
~ disappears on one of the channels, the same packet will still arrive to the
~ destination as the routers (hopefully) let it take another path.
~
~ The idea probably sounds crazy but just come up with your thoughts and
~ ideas; I am looking for a subject to my final thesis to write about.
~ Actually, any inspirations to a subject is highly welcome. Thanks!

 
Reply With Quote
 
Jack L.
Guest
Posts: n/a
 
      02-24-2004
Jack L. wrote:

> The idea probably sounds crazy but just come up with your thoughts and
> ideas; I am looking for a subject to my final thesis to write about.
> Actually, any inspirations to a subject is highly welcome.
> Thanks!


Thank you so much for sharing your thoughts on this idea - the arguments
sound convincing enough not to put further effort on it.

--
Mvh / Best regards,
Jack, Copenhagen

The email address is for real.



 
Reply With Quote
 
Hank Karl
Guest
Posts: n/a
 
      02-26-2004
On 24 Feb 2004 15:23:08 GMT, Andreas Sikkema <(E-Mail Removed)> wrote:

>Hank Karl <(E-Mail Removed)> wrote in
>news:(E-Mail Removed) :
>
>> (considering the TCP/IP overhead).

>
>I think you mean UDP/IP overhead


Ok, I slipped. But there is also RTP, RTCP, and possibly PPPoEoA to
consider.
 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
SetAuthCookie works some times and fails some times? =?Utf-8?B?bWF2cmlja18xMDE=?= ASP .Net 0 03-23-2006 09:24 PM
How can I restrict that the some ID can only login once in the some time ad ASP .Net 2 08-12-2005 09:14 PM
Need some help for some perl homework.... Perl 0 02-25-2004 01:45 AM
Forms Authentication question: How to have some pages open and some requiring forms authentication Eric ASP .Net 2 02-13-2004 02:14 PM
Some questions regarding 070-305 and hopefully some right answers. Needs correction... wink, wink ;-) Daniel Walzenbach MCSD 1 11-10-2003 12:25 AM



Advertisments