Re: [ns] Error Model in Wireless Network

[Huan Pham <huan@cs.jcu.edu.au>]

Hi Jahn and NS users,

I currently have a similar concern as Leandropereiramaciel's one. So thank you, Jahn, for your reply
to his message.

But I think you may not have answered his questions. I think what he is simulating can probably be a
Fixed Wireless network. The propogation models available in NS are not of much interest in fixed
wireless, as users are not moving. On the other hand, for Fixed wireless, we care more about the
channel variable error rate, and possible variable physical bandwidth which change with the weather
conditions.

I think the variable error rate problem can be partly solved by using multi-state error model,
although it does not do all what we need, for example it's impossible if I would like to simulate the
error rate increasing gradually from 1% to 10% (packets) during a period from  0 to 10 seconds.

The second problem, (the changes of physical bandwidth) does not seem to have an answer yet. I mean,
can we set a link to have "multi-state" bandwidth or can we change the channel bandwidth "on the
flight".

Your further clarification to my concerns will be very appreciated.

Thanks



jahn-isi wrote:

> For modeling the transmission error over wireless channel, ns provides three propagation
> models such as free-space, two-ray, and showdowing which compute the signal power
> when a packet arrive at the receiver. Based on the left signal power, ns can decide a packet's
> loss. For the detail of this large-scale fading effect, you may refer to "Wireless Communications"
> written by Rappaport. Also search for the ns mail archive for small-scale fading models.
> Ratis reported that he implemented the Ricean model for ns.
>
> Ns also gives another mechanism to infuse errors over wireless channels based on some
> analytical equation such as unifrom, multistate, and etc as like we do over wireline networks.
> What you are asking is about the parameters related to these analytical equations to generate
> errors.
>
> Briefly, the rate indicates a probability that a packet, byte, or bit can be corrupted depending
> on what unit you specified. The bandwidth is a parameter necessary for computing the byte
> duration when the unit is time. I guess that this bandwidth should be the same as one specified
> for physical wireless channels. For more information, please refer to errmodel.cc.
>
> -jahn
>
> -----Original Message-----
> From: owner-ns-users@ISI.EDU [mailto:owner-ns-users@ISI.EDU]On Behalf
> Of leandropereiramaciel
> Sent: Saturday, May 26, 2001 12:03 AM
> To: ns-users@ISI.EDU
> Subject: [ns] Error Model in Wireless Network
>
> Hi, all,
>
> I would like to know how to simulate a wireless network
> where I can vary the following parameters:
>
> > the transmission error of the channel - I'm using the
> new function introduced in March, 2001. I understand
> that i just need to adjust the "rate_" parameter (in
> ErrorModel) to achieve the goal.
>
> > the bandwidth of the channel - How I adjust the
> bandwith? The right parameter to adjust is
> "Phy/WirelessPhy set bandwidth"? If so, the "ErrorModel
> set bandwidth" also has to have the same value?
>
> > the size of the file being transmited - I need to
> simulate a file being transmited, like a HTTP file. Is
> there a protocol already done or i need to construct my
> own?
>
> Are my assumptions correct? Any suggestion is welcome.
>
> Thanks...
>
>
> __________________________________________________________________________
> Acesso pelo menor pre? do mercado! R$ 14,90 nos 3 primeiros meses!
> ASSINE AGORA! http://www.bol.com.br/acessobol/

--

Huan Pham - PhD Student
School of Information Technology
James Cook University, QLD4811, Australia
Phone: 61-7-4781 6909    Mobile: 61-418-875084
Fax: 61-7-4781 4029
Homepage: http://www.cs.jcu.edu.au/~huan