Re: [ns] Loss models in wirelss

[jeyandran venugopal <jvenugop@students.uiuc.edu>]

Hi,
  This might not answer your question. But, I am arising a related
question, in case , miraculously some one decides to reply to this mail, 
he/she might answer this related question also..

I tried using the markov model to introduce a two state error model
during reception of the packet at the mac layer..but, unfortunately, when
I simulate a FTP/TCP eventhough the througput decreases and increases when
the average time spent in the bad state is decreased and increased
respectively, there is no difference when I try to use either a
reno/newreno instead of tahoe..By tracing the cwnd I found that the
pattern is exactly the same for all the variants of tcp and hence there is
no difference whatsoever when using different variants of tcp in the way
the traffic is sent..(once the transition to the bad state occurs, the
cwnd never recovers...it stays at 1 after that)....

Can someone explain this behavior when trying to use the two state error
model as implemented in ns2.1b6?

Thanks.
Jeyandran

On Mon, 20 Nov 2000, Pablo Souki Mauricio Jim�nez wrote:

> We have looked through the mailing list and every
> single person who has asked this has gone unreplied
> to, so here it goes again:
> 
> We're trying to simulate the loss patterns of wireless
> models, and up til now have been doing it using Markov
> models on wired links. We were wondering if there's a
> way of inserting loss patterns into wireless links, or
> if that is not exactly the intention of the wireless
> model in ns.
> 
> Thank you very much.
> 
> Mauricio & Pablo.
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Calendar - Get organized for the holidays!
> http://calendar.yahoo.com/
>