A Correct Implementation of a Two State Error Model

[AL-HUSSEIN ABOU-ZEID <hussein@maxwell.ee.washington.edu>]

In reference to quite a large number of questions and bug reports on the
two state error model implementation, I have modified the 2-state error
model to work in a manner expected by most of us (people trying to
simulate a fading situation where the channel transitions between a good
and a  bad state, spending exponentially distributed times in each of the
states, and the channel transitions are independent of packet
transmissions).

For now, the packet is dropped with probability 1 in the bad state and
with probability 0 in the good state, but you can easily modify this.

Please refer to the URL
http://students.washington.edu/~abouzeid/research.htm
to download the changed files, read more about how the 2 state model is
different from the original ns implementation, and download a tcl script
file using such a model.

Hoping this contribution will be added to the list of contributed modules,
and hopefully, be part of the next ns version.

Best regards,

Alhussein A. Abouzeid
Fundamentals of Networking Laboratory (FuNLab),
Department of Electrical Engineering,
University of Washington, Seattle.