Multifractal Signal Models Withapplication to Network

In this paper, we develop a new multiscale modeling framework for characterizing positive-valued data with long-range-dependent correlations. Using the Haar wavelet transform and a special multiplicative structure on the wavelet coeecients to ensure non-negative results, the model provides a rapid O(N) algorithm for synthesizing N-point data sets. We study the second-order and mul-tifractal properties of the model and derive a scheme for matching it to real data observations. To demonstrate its eeectiveness, we apply the model to TCP network traac synthesis. The exibility and accuracy of the model and tting procedure result in a close match to the real data statistics (variance-time plots) and queuing behavior.