Course Title:  Statistical Analysis and Modelling of 
Internet Traffic Data


Fall Semester 2001


Taught By:  J. S. Marron


Description:  The analysis and modelling of internet traffic 
data represents an important major challenge for engineers, 
for computer scientists, for statisticians and for 
probabilists.  Really new ideas and models are needed 
because heavy tailed distributions and long range dependence 
(both appearing at a number of different points) render 
standard methods, such as classical queueing theory, 
unusable.  While the intellectaul challenges are great, the
problem is also of central importance because the present 
protocols were not designed with today's massive scale of 
the world wide web in mind, which results in large 
inefficiencies.  This course considers a variety of 
mnethods for understanding and modelling internet traffic 
at a variety of levels, from individual TCP traces, to 
monitoring traffic on a main link.  An important underlying
concept is cross scale views of data.  Novel graphical 
views of data play an important role.


Text Book:  none


Prerequisites:  One year of probability and statistics, at
the undergraduate level.