g. u. bEkTEMySovA, J. b. ibrAEvA
International University of Information Technologies, Almaty, Kazakhstan
RESEARCH OF A TIME SERIES FOR STATIONARITY
Real MPEG packet intensity network traffic has a non-uniform rate of packet arrival at serving
network devices. With the development of the concept of the Internet of Things ( IoT), the relevance of
network management tasks is growing every day. The study allows you to get meaningful answers to the
questions of interest to the researcher. The use of various tests (criteria) in the study of a series is relevant
because they allow us to reveal their structure.
With the increase in the variety of network applications, new data transfer protocols, properties and
features that are associated with its non-stationarity began to appear in the behavior of traffic. This article
considers a time series with real data taken on the highway of the city of Almaty.
To study the time series for stationarity, the Kolmogorov-Smirnov, Shapiro-Wilk, Darbin, David-
Hartley-Pearson, Anderson-Darling criteria, QQ plot, Variance Ratio test for Random Walk data in the
Matlab numerical simulation environment were applied. Estimates are obtained for studying the series in
terms of Pearson and Poisson distributions. Numerical characteristics of the form of distributions are also
determined: asymmetry and kurtosis.
The article carried out simplified checks of the measured time series.
key words: time series, network traffic, data analysis, stationarity test criteria, normal distribution
law.
Достарыңызбен бөлісу: |