End-to-end network traffic modeling method and system based on Bayesian theory

Abstract

The invention provides an end-to-end network traffic modeling method based on the Bayesian theory. The method comprises the following steps: in a network topology structure, obtaining initial values of end-to-end network traffic at preset continuous multiple time slots; converting the end-to-end network traffic at the preset continuous multiple time slots into a random process obeying normal distribution, and obtaining parameter and probability of the end-to-end network traffic at each time slot; based on the Bayesian theory, constructing an end-to-end network traffic model at the current time slot according to the parameter and probability of the end-to-end network traffic at each time slot before the current time slot in the preset continuous multiple time slots; and defining a model deviation function to enable deviation between an estimated value of end-to-end network traffic at the current time slot and a true value thereof to be minimum, and constructing an optimization model of the end-to-end network traffic model at the current time slot according to the model deviation function and the end-to-end network traffic model at the current time slot.

Description

technical field [0001] The invention relates to the field of network prediction under a large-scale network environment, in particular to a Bayesian theory-based end-to-end network flow modeling method and system. Background technique [0002] With the rapid development of network technology and new applications, network traffic exhibits new characteristics. This brings new challenges to network engineering. It is very important to accurately characterize and model network traffic to improve network performance. The characteristics of network traffic, such as self-similarity, autocorrelation, heavy-tailed distribution, etc., have an important impact on network optimization and routing optimization. End-to-end network traffic represents network-wide behavior from a global perspective. Therefore, the modeling of end-to-end network traffic has received extensive attention from researchers, operators, and developers from all over the world. [0003] End-to-end traffic behavi...

AI Technical Summary

Problems solved by technology

However, these methods need to acquire additional information from link loads or a priori information about end-to-end network traffic, increasing the complexity and overhead of model parameter computation

Other methods, such as time-frequency domain analysis and neural network, can construct models to represent end-to-end network traffic, but it is difficult to accurately capture and obtain the characteristics of end-to-end network traffic, and cannot build accurate and appropriate network traffic model

Images