Self-Similar Network Traffic Modelling Using Fractal Point Process-Markovian Approach

Abstract

AbstractSeveral recent Internet traffic measurement studies reported that traffic in modern high-speed networks is a self-similar process. If the stochastic self-similar network traffic models do not accurately represent the real traffic, then the network performance may be over estimated or underestimated, and it causes degradation of Internet router performance. Therefore, it is decisive for an appropriate design of a router. In this paper, we investigate mean waiting time and tail probability of network router with pseudo self-similar traffic input. We use Fractal Point Process (FPP) as input process since it emulates self-similar traffic. However, this process is asymptotic in nature and has less effective in queueing based performance analysis. Therefore, for queueing analysis Markov modulated Poisson process (MMPP) is fitted for FPP. FPP involves another parameter Fractal Onset Time (FOT) besides Hurst parameter. Effect of FOT on tail probability and mean waiting time is examined.KeywordsNetwork routerSelf-similarFractal point processMarkov modulated Poisson processMean waiting timeTail probabilityTraffic intensity