Fast Clustering of Self-Similar Network Traffic Using Wavelet
In this paper we have given our proposed approximation methods for similarity search in large temporal databases using wavelet transformation based featured signals and time warping distance algorithm. Our main goal is to propose efficient methods to speed up the mining of matched sequences especially when the sequences are of random lengths and traditional distance metrics like Euclidean distance fail to achieve the desire goals. We proposed two methods for truncation of databases for optimizing search procedures for similarities using the concept of wavelet based featured time warping. In our first model, we utilize the maxima, minima and average features of wavelet based compressed signals and in second model, features of wavelet transformation using average of approximation coefficients at the coarsest scale and maxima of maxima and minima of minima of detail coefficients at all scales. We show by carrying out extensive experiments that our proposed methods are very effective and ensure the nonoccurrence of false dismissals and minimal false alarms with least compromise over accuracy.
Keywords: Wavelets, Multiresolution analysis, Dimensionality reduction, Network Traffic, Time warping, Data mining
Download Full-Text
