Author(s): Beirão, Fábio Duarte
Date: 2010
Persistent ID: http://hdl.handle.net/10400.6/3718
Origin: uBibliorum
Project/scholarship: info:eu-repo/grantAgreement/FCT/5876-PPCDTI/PTDC%2FEIA%2F73072%2F2006/PT;
Subject(s): Analysis of traffic behaviour; Auto-correlation estimator; Average and standard deviation; Entropy estimator; Hurst parameter; Modular approach; Random capture generator; Real time analysis; Statistical traffic analysis; Network traffic
Traffic monitoring and analysis is of critical importance for managing and designing modern computer networks, and constitutes nowadays a very active research field. In most of their studies, researchers use techniques and tools that follow a statistical approach to obtain a deeper knowledge about the traffic behaviour. Network administrators also find great value in statistical analysis tools. Many of those tools return similar metrics calculated for common properties of network packets. This dissertation presents NetOdyssey, a framework for the statistical analysis of network traffic. One of the crucial points of differentiation of NetOdyssey from other analysis frameworks is the windowed analysis philosophy behind NetOdyssey. This windowed analysis philosophy allows researchers who seek for a deeper knowledge about networks, to look at traffic as if looking through a window. This approach is crucial in order to avoid the biasing effects of statistically looking at the traffic as a whole. Small fluctuations and irregularities in the network can now be analyzed, because one is always looking through window which has a fixed size: either in number of observations or in the temporal duration of those observations. NetOdyssey is able to capture live traffic from a network card or from a pre-collected trace, thus allowing for real-time analysis or delayed and repetitive analysis. NetOdyssey has a modular architecture making it possible for researchers with reduced programming capabilities to create analysis modules which can be tweaked and easily shared among those who utilize this framework. These modules were thought so that their implementation is optimized according to the windowed analysis philosophy behind NetOdyssey. This optimization makes the analysis process independent from the size of the analysis window, because it only contemplates the observations coming in and going out of this window. Besides presenting this framework, its architecture and validation, the present Dissertation also presents four different analysis modules: Average and Standard deviation, Entropy, Auto-Correlation and Hurst Parameter estimators. Each of this modules is presented and validated throughout the present dissertation.
Fundação para a Ciência e a Tecnologia (FCT)
