Thomas Karagiannis and Michalis Faloutsos
Over the last few years, the network community has started to rely heavily on the use of novel concepts such as fractals, self-similarity, long-range dependence, power-laws. Especially evidence of fractals, self-similarity and long-range dependence in network traffic have been widely observed. Despite their wide use, there is still much confusion regarding the identification of such phenomena in real network traffic data. For one, the Hurst exponent can not be calculated in a definitive way, it can only be estimated. Second, there are several different methods to estimate the Hurst exponent, but they often produce conflicting estimates. It is not clear which of the estimators provides the most accurate estimation. In this extended abstract, we make a first step towards a systematic approach in estimating self-similarity and long-range dependence. We present SELFIS, a javabased tool that will automate the self-similarity analysis. To our knowledge, our software tool is the first attempt to create a stand-alone, free, open-source platform to facilitate self-similarity analysis. We show the use of our tool and describe the methodologies that currently incorporates in real Internet data. Finally, we present an intuitive approach to validate the existence of long-range dependence.
In 1st Workshop on Fractals and Self-Similarity in Data Mining: Issues and Approaches (IN KDD)