[e2e] typical network syllabus

[Jon Crowcroft]

In message <3BDF08DC.8138A4B1 at cs.columbia.edu>, "Henning G. Schulzrinne" typed:

 >>Thanks for the list. I'm not sure a single 'advanced' course is
 >>necessarily the best option, given the diversity of topics and
 >>approaches.

 >>At Columbia, we've been splitting this up into multiple courses,

also, you have to decide if you are addressing
hardware engineers, computer scientists, or applicaiton programmers - for
example I've found that courses on protocols for CS work well when
oriented aroudn the Steven's books (unix net programmign and the
illustrated series), but these are no use to EE students whose programmign
may be thinner

whereas courses on network design are better oriented around Keshav's book
and suit EE folks, but less so CS

a middle ground for EECS is to orient thigns around Peterson/Davies
systems book

both multimedia and transmission, coding/framing 
can be split either way - you can go for implementation
or information theoretic, or you can go for h/w, circuits, dsp, control
theory, FFT/DCT, etc etc - too much applied math , you lose the CS people,
too much C, you lose the EE...

security is also tricky - generally , bellovin, or ross anderson or
schneier's books do nicely for anyone.....but of course only real math
people should go near novel crypto:-)

routing is another area where you can take a classical distributed
algorithm approach, or you can do it by code/example, and simulation ...

the master's in data comms and telecomms courses at UCL CS and EE reflect
this dichotomoy, with a somewhat different mix of approaches, but the same
list of topics...

cheers
jon