Internet Traffic Modeling

Problem
This project introduces techniques for the modeling of WWW traffic on the Internet. This project also will provide a general overview of HTTP and NNTP and the interaction between these applications and TCP/IP.

Requirement

• UNIX C or PERL (preferably both).
• Basic TCP/IP.

Project Procedure

• Collect TCP/IP traffic of the desired application using tcpdump. You will be given a 50MB tcpdump sample file instead of collecting, because you are required to have ROOT password to do so.
• Identify and parse specific parameters using PERL.
  According to HTTP and NNTP transaction, there are essential parameters that constitute WWW traffic such as Round Trip Time and Number of In-lines. PERL is a very useful shell script when dealing with a string of ASCII text(not for CGI at this project). With this tool, gather the statistics from the tcpdump sample file.
• Find the best fit distribution of each parameter.
• Generate artificial traffic according to the distributions with preferable language(if time allows).
• Compare the artificially generated traffic with the real traffic(if time allows).

Background

• Statistical characterization of a World Wide Web browsing session by Sunil Khunte and John Limb available at ftp://ftp.cc.gatech.edu/pub/coc/tech_report/1997/GIT_CC_97_17.ps.Z
• Empirical model of WWW document arrival at access link by Shuang Deng available at IEEE ICC ’96
• Explaining World Wide Web traffic self-similarity by M. Crovella and A. Bestavros available at http://www.cs.bu.edu/~best/res/papers/Home.html#Reports

• Write a 3 - 5 page project report including the following items: Describe the way to identify each parameter in PERL script. Explain each parameter and list its mean, standard deviation and best fit distribution. List the limitations and your own methods and suggestions to overcome these limitations.
• Hand in the raw PERL script and the code for generated WWW traffic.