Long-term forecasting method and device of network flow

Abstract

A long-term prediction method and apparatus of network traffic. The method comprises: initially setting the length of a dynamic sliding window; performing cyclic learning on given historical data according to the initially set length of the dynamic sliding window, and selecting an optimal sliding window, a linear trend attribute coefficient and a fluctuation series parameter set; and predicting future data traffic of a given step number according to the optimal sliding window, the linear trend attribute coefficient and the fluctuation series parameter set. The embodiments of the present invention optimize a model parameter of a fluctuation sequence according to a trend feature of traffic, thereby improving the prediction precision of unstable (long-term) traffic.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a method and device for long-term forecasting of network traffic. Background technique [0002] With the expansion of the network scale and the continuous growth of business types, the network traffic data in the real network environment has shown the typical characteristics of a non-linear, multi-dimensional dynamic system, such as fractal, long-term correlation, self- Similarity, suddenness, etc. Stochastic models such as Markov and Poisson cannot effectively describe these characteristics of traffic. That is to say, traditional stochastic models can only deal with short-range dependence of network services, such as Poisson process, Markov process, AR (Auto Regressive), MA (Moving Average), ARMA (Auto Regressive Moving Average) And the ARIMA (Auto Regressive Integrated Moving Average) process, etc., cannot deal with the long-range dependence (long-range dependence) of ...

AI Technical Summary

Problems solved by technology

[0004] During the research and practice of the existing technology, the inventors of the present invention found that in the existing implementation, the traditional FARIMA model does not describe the trend characteristics of the flow rate, and the estimated model parameters are not optimal parameters. The accuracy is not high enough, especially for non-stationary (long-term) traffic forecasting effect is worse, this is because, in the long-term traffic forecasting, it is easy to cause traffic changes due to network route changes or topology changes

Images