Experimental Evaluation On Methods Of The Hurst Parameter Estimation Using Fractional Gaussian Noise

Analysis to the network traffic is still the key field relating to network measurement, network dynamics, call admission control, and network performance analysis. With the rapid development of communication networks, the popularization of network and various new applications make network traffic increasing rapidly such as Video On Demand (VOD) and Voice Over Internet Protocol (VOIP) etc.. Thus, network traffic comes to show properties differing from the traditional network traffic, such as Poisson model. Accordingly, new challenge has been brought to the analysis and study of the wired and wireless network traffic. At the same time, new fields have also been introduced which will absorb swarm of attention and investigators. One of the effective methods for studying self-similar traffic is to build models that could describe network characters more reality, and be applied to simulation research.In Chapter 1, we introduce research background of self-similar traffic and the status at home and abroad. In Chapter 2, first, the traditional typical traffic models have been introduced. Second, the mathematical meaning of the self-similarity and Long-range Dependence has been provided. Chapter 3 presents several common FGN random data generation algorithms. Several algorithms for the simulation and analysis in MATLAB environment are also given. In Chapter 4, we give experimental evaluations on 8 commonly used methods of the Hurst parameter. Firstly, we examine the ability of these methods to distinguish self-similar series. Then, we elaborate the evaluation methods of accuracy, consistency and the operation speed. Finally, we discuss the influence factors of estimation, such as periodicity signals, Gaussian stochastic signals and relational structure. In Chapter 5, the FGN random data generation and estimation of the Hurst parameter system is generated in the MATLAB GUIDE development platform. This will help us in the generation storage and statistical analysis of FGN random data and estimation Hurst parameter.The main contributions of this paper are: (1) several algorithms for the simulation and analysis in MATLAB environment are given; (2) experimental evaluation of the Hurst parameter estimation; (3) we lucubrate the factors of influence in estimation Hurst parameter; (4) we have given a system of FGN data generation and Hurst parameter estimation.