Abstract
"The article uses nonparametric criteria to assess the trend of network access traffic. Practice shows that most real processes do not satisfy the properties of stationarity. They are characterized by a trend, systematic changes in variance, the presence of periodic fluctuations and the presence of changing interdependencies between the levels of the time series. A review of some foreign articles devoted to this topic has been carried out. Also, the Wireshark sniffer program is used, the task of which is to capture packets passing through the network. Empirical data were obtained on the access network segment between LAN and ISP. In the article, using nonparametric tests (the series criterion, the «ascending and descending» series criterion, the Cox-Stewart sign criterion and the «Caterpillar» method), the hypothesis of the presence of a trend in the measured time series on the access network was tested. The decomposition of the time series into its components is performed. Keywords: access network, time series, trend, stationarity, nonparametric criteria."