Wavelet Analysis of Long-Range-Dependent Traffic：（小波分析Long-Range-Dependent流量）.pdfVIP

2 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 44, NO. 1, JANUARY 1998 Wavelet Analysis of Long-Range-Dependent Traffic Patrice Abry and Darryl Veitch Abstract— A wavelet-based tool for the analysis of long-range with tails which decay slower than the exponential decay dependence and a related semi-parametric estimator of the Hurst familiar from Markovian models [11], [13], [27], [28], [32]. In parameter is introduced. The estimator is shown to be unbiased the absence, however, of a complete program of data analysis, under very general conditions, and efficient under Gaussian assumptions. It can be implemented very efficiently allowing informed model selection, parameter estimation, and finally the direct analysis of very large data sets, and is highly robust model verification against relevant performance criteria, the against the presence of deterministic trends, as well as allowing connection between LRD and performance metrics cannot be their detection and identification. Statistical, computational, and fully understood. numerical comparisons are made against traditional estimators Among the outstanding issues in the list above, there are including that of Whittle. The estimator is used to perform a thorough analysis of the long-range dependence in Ethernet two of particular importance which we contribute to in the traffic traces. New features are found with important implications present paper. The first is the primordial problem of data for the choice of valid models for performance evaluation. A analysis/parameter estimation in