Network flow modelling method and system based on Weibull distribution

Abstract

The invention relates to a network flow modelling method and system based on Weibull distribution. The method comprises the following steps of (1), generating a pre-determined number of Weibull noise time sequences at; (2), pre-processing the generated Weibull noise time sequences at, removing singular points, and performing zero equalization treatment so as to obtain Weibull time sequences epsilont; (3), estimating a fractional difference coefficient d by adopting a search method; (4), according to the fractional difference coefficient d, differentially filtering the Weibull time sequences epsilont obtained in the step (2) by using a fractional difference formula so as to obtain time sequences Wt; (5), performing ARMA process of the time sequences Wt so as to obtain FARIMA time sequences Xt; and (6), verifying that the generated FARIMA time sequences Xt can describe the self-similarity and the burstiness of the network flow. Gaussian distribution is replaced by Weibull distribution in the network flow modelling method; therefore, the generated FARIMA model is more precise than the traditional network flow model; the FARIMA model accords with a practical network flow waveform diagram better; and the self-similarity degree is relatively high.

Description

technical field [0001] The invention relates to the field of network communication, in particular to a Weibull distribution-based network flow modeling method and system. Background technique [0002] Network traffic modeling plays a very important role in the design, evaluation and optimization of computer networks. Among them, the statistical distribution theory is used as the basis for the study of network services, which promotes the development of network traffic services. The self-similarity and burstiness of network services have a huge impact on network analysis, design, control and management, and network service Quality presents a tougher challenge. Network performance evaluation has become an important research part of network design and control, especially with the development of the network, the services carried on the network have also expanded from voice to video, multimedia, etc. Therefore, establishing a reasonable network traffic model plays a vital role ...

AI Technical Summary

Problems solved by technology

However, this model is flawed to some extent. For example, under a single Hurst parameter, it is difficult for this model to describe the correlation of the entire time series. Moreover, when modeling actual network traffic, This kind of self-similar process is relatively one-sided

Images