Thanks for the tip.
I've even found an example on a previous msg of this mailling list:

"  ...
  set udp0 [new Agent/UDP]
  $ns attach-agent $n(1) $udp0
  set null0 [new Agent/Null]
  $ns attach-agent $n(2) $null0
  $ns connect $udp0 $null0

  # create 16 Paroto On/Off source traffic
  for {set i 0} { $i < 16} {incr i} {
          set par($i) [new Application/Traffic/Pareto]
          $par($i) set rate_ [expr 64*4]k
          $par($i) set packetsize_ 210
          $par($i) set burst_time_ 500ms
          $par($i) set idle_time_ 50ms
          $par($i) attach-agent $udp0
          $ns at 0.5 "$par($i) start"
  }
"
thanks,
Pedro Fortuna
INESC Porto

On 8/4/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
>
> If you want to simulate VoIP traffic you could also use the Paretto
> generator, I think it will simulate bursty VoIP more precisely than
> Exponential.
>
> Douglas Nascimento - UFRGS
>
> Quoting Pedro Fortuna <[EMAIL PROTECTED]>:
>
> >
> > I assume that there's a correlation, because two people do not talk
> > exactly at the same time. They usually talk and then listen. Thus, one
> > flow periods of idleness usually correlate to the other flow's periods
> > of burstiness. Maybe the best way to model that is with a single
> > on/off source, but I don't know the details.
> >
> > I followed a simpler approach, where I setup a flow in each direction,
> > with the same burst and idle mean values and also with the same data
> > rates. With this kind of strategy, you'll have very simetric data
> > passing in the pipes. This means that in such kind of conversations,
> > both in the short and in the long run, people will talk aproximately
> > the same time... which sounds more like a political debate than a
> > regular phone conversation :-)
> >
> > On 8/4/06, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote:
> >> Is it necessary to consider the correlation between UL and DL voice 
> >> traffic?
> >>
> >> Or we simply assume UL and DL are independent traffic stream both
> >> generated by
> >> Application/Traffic/Exponential?
> >>
> >> I am simulating VoIP in 802.11 networks, where UL and DL compete
> >> for the same
> >> channel.
> >>
> >> Thanks
> >>
> >>
> >> Quoting Pedro Fortuna <[EMAIL PROTECTED]>:
> >>
> >> >
> >> > Hello,
> >> >
> >> > I've been using Application/Traffic/Exponential as a traffic generator
> >> > on top of Agent/UDP. It's not perfect, because, for instance, I
> >> > specified 8Kbit data rate, but often I only see 5 or 6Kbit/s passing
> >> > through... but maybe i'm doing something wrong....
> >> >
> >> > Pedro Fortuna
> >> > INESC Porto
> >> >
> >> > On 8/4/06, Matthew Jakeman <[EMAIL PROTECTED]> wrote:
> >> > >
> >> > > VoIP is basically just UDP packets encapsulating RTP packets
> >> with the voice
> >> > > data inside, all you should need to do to simulate a VoIP
> >> stream is set the
> >> > > correct packet size and frequency that the packets are sent out and 
> >> > > that
> >> > > would simulate a stream, all of which can be done easily inside your 
> >> > > tcl
> >> > > script!
> >> > >
> >> > > Matt
> >> > >
> >> > > On Thursday 03 August 2006 16:41, lekkie omotayo wrote:
> >> > > > Hi,
> >> > > >
> >> > > > Does anybody know an existing VoIP simulated Application
> >> existing on NS2?
> >> > > >
> >> > > > I am ready to develop it myself, however, most of my questions has no
> >> > > > been answered.
> >> > >
> >> > >
> >> >
> >> >
> >> > --
> >> > Cumprimentos,
> >> > Pedro Fortuna
> >> >
> >> >
> >>
> >>
> >>
> >>
> >
> >
> > --
> > Cumprimentos,
> > Pedro Fortuna
> >
> >
>
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
>
>


-- 
Cumprimentos,
Pedro Fortuna

Reply via email to