ПРОГНОЗИРОВАНИЕ СЕТЕВОГО ТРАФИКА

Authors

Zhanybek Sharshenaliev Institute of Automation and Information Technologies of the National Academy of Sciences of the Kyrgyz Republic, Bishkek
Sh. A Mirzakulova
S. U. Isakova

Keywords:

: time series, neural network, perceptron, multilayer network, forecast

Abstract

A modern heterogeneous network generates network traffic with a complex (heterogeneous) structure. A study of the actually measured data shows that they do not have a uniform intensity of packet arrival to serving network devices. Moreover, models built on the basis of data characterizing one object for a series of consecutive moments of time (time series) have the non-stationary property, which means that their structure is multicomponent. Therefore, today network management tasks are based, among other things, on predicted future data to make the right decision. To identify and quantify the components of a complex structure - the presence / absence of a trend, periodicity, random component is the main task of the analysis of the time series. To identify a nonlinear function and carry out its forecasting, neural network algorithms with deep learning are very successfully implemented today.

References

Николаева С.Г. Нейронные сети. Реализация в Matlab. Учебное пособие. Нейронные сети. Реализация в Matlab: учебное пособие / С.Г. Николаева. – Казань: Казанский государственный энергетический университет, 2015. – 92 с.

Исакова С.У., Мирзакулова Ш.А. Статистическая оценка основных тенденций измеренного временного ряда. – Вестник КРСУ, 2018. – Том 18. – №12. –С. 51–54.

Шаршеналиев Ж.Ш., Мирзакулова Ш.А., Юсупова Г.М. Оценка спектра мощности временного ряда // Материалы научной конференции ИИВТ КН МОН РК «Инновационные IT и Smart-технологии» – Алматы, 03. 2019. – С. 332–337.

Серіков Т.Ғ., Мирзакулова Ш.А., Юсупова Г.М., Авелбекова С.Ш., Сабитова А.Ж. Анализ временного ряда методом ССА // Вестник ПГУ, 2019. – № 3 –

С. 328–339.