Program for the ``QoS'' class, fall 2002.

I hope all students have been able to do the programming for the
homework for Thu 09/19. I will, however, give an outline of how I
would tackle this problem on Thursday 09/19 and give students more
time to do the programming.

I have heard some students are running scared because of the math in
the Parekh bounds. Indeed! Preparing that material is taking even more
of my time then I had anticipated. However: Teaching a PhD level course
in ``QoS in the Internet'' and not covering the Parekh bounds runs the
risk of later standing for joker. (Both you and me. You when you are
interviewing for a job. I when you are interviewing for a job.) So, I
have no choice but to finish the topic.

But, I am thinking of changing the order a bit. 
Talk more about NS,
Talk about the TCP dynamics,
RED (Random Early Detection),
ECN (Explicit Congestion Notification)
The square root law (more than handwaving but less than the 
full mathematical treatment),
TCP-friendly behavior,
Equation-driven TCP.

With the remainder of the Parekh bounds in-between.

With the exception of the proof of the square root law, the topics
above are (at least in the math) easier than the Parekh bounds.

(However: this looks like already too ambitious a program, and I would also 
like to discuss RSVP, Intserv, Diffserv. Choices will have to be made).

I would like to have every student do an individual project, using NS 
to analyze some interesting QoS-related problem. Students who have an
idea for a project they would like to do, feel free to stop by.

Teun Ott.