Network Modeling for the Internet

Structural Breaks



Leader Vladas Pipiras (University of North Carolina, Chapel Hill), pipiras@email.unc.edu
 Meeting Tuesday 3:00 - 4:00 pm, room 201
Members Robert Buche (North Carolina State University), rtbuche@unity.ncsu.edu
Fred Godtliebsen (University of Tromso, Norway), godtlieb@email.unc.edu
Cheolwoo Park (SAMSI), cwpark@email.unc.edu
Stilian Stoev (Boston University), sstoev@bu.edu
Murad Taqqu (Boston University), murad@bu.edu
Outline Objectives An increasing number of publications in the time series literature suggest that evidence of long-range dependence is an artifact of structural breaks. Structural breaks (also: level shifts, regime switching, structural instability) is one type of non-stationarity of a time series. The explanation for long-range dependence through structural breaks is particularly prevalent in Econometrics where evidence for long-range dependence was found in the time series of volatility of stock prices, inflation rates and other economic indicators.
The goal of this working group is to explore the idea of structural breaks in the context of Internet traffic modeling where evidence of long-range dependence is also ubiquitous. It appears that these ideas have not been examined yet. The work of the group will focus around the following themes:
1. definition and types of structural breaks, models (some mixture, Markov-switching and other models),
2. structural breaks versus long-range dependence (possibly versus other phenomena),
3. statistical tests to discriminate between structural breaks and long-range dependence,
4. applications to Internet traces.
References http://www.unc.edu/~pipiras/





Project Web Site


Internet Program Home Page

SAMSI Home Page

 

© 2002, Statistical and Applied Mathematical Sciences Institute. All rights reserved.