[ns] error models

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

Hi all,
   I am trying to model a wireless network and am trying to use the two-state 
error model to use a markovian model for the error distributions on the 
channel. I find that irrespective of the TCP variant being used, once the 
channel enters a bad state, there is no getting back to the good state..I was 
puzzled by this behavior and on looking through the contributed modules page, 
found a module by Alhussein abouzeid which describes this same problem and 
provides a new version for the files errmodel.cc and the corresponding .h and 
the tcl files..
But, their modifications were done to the 2.1b4a version of ns and I am 
currently using 2.1b6. My questions are
a)Is the 2.1b6 implementaion of ns also flawed in the same way as the 2.1b4a 
model, in that it does not update the channel state when there is no packet 
reception event?
b)If so, what is the fix for this? If someone can give me some suggestions on 
this, it would be great..

Thanks.
Jeyandran