Multi-fractal Modeling of Network Video Traffic and Performance Analysis

Abstract

As the network scale expanding and network business demand increasing sharply, network behavior prediction problems are constantly emerging, such as network detection need to complete the description of the modern network traffic characteristics, to set up the mathematical model, aimed at more efficient use of network resources and ensure the implementation of network QoS. Firstly, described the multi-fractal model of network video traffic and analyzed the influence factors of the simulation sequence of the model. Secondly, designed the algorithm and utilized Haar wavelet to express the simulation sequences of the multi-fractal model and analyzed those long range dependence (LRD), the simulation sequence of multi-fractal model with Haar wavelet is the most close to real video traffic. Thirdly, proposed a controlling method of the LRD of multi-fractal model, relation of edge distribution and the relevance function of the coefficient from the point of theory view. Finally, the early scale coefficients are modeled with AR and the connection is constructed on the short range dependence (SRD) of the early scale coefficients and LRD of finally traffic sequence, realized the precise control sequence on LRD. Experiments shown the stability of multi-fractal model and the consistency of LRD are improved.