Abstract
Functioning Packet Highways are encountering increasing loads from wired access networks, which in turn are transitioning to optical technologies for connecting future end devices. Numerous studies of data in multiservice networks confirm their non-stationarity. This article is dedicated to analyzing network traffic using unit root tests. The uniqueness of this work lies in the use of real measured data. Applying the obtained results in practice can aid in quality of service (QoS) management. Identifying network traffic stationarity enables the selection of the optimal forecasting model, ensuring more accurate load prediction and effective QoS optimization. The novelty of this study lies in the application of unit root tests to real-world traffic data collected via Wireshark, providing a statistically grounded approach to evaluating network nonstatianarity. The study proposes a comprehensive approach that includes classical and advanced stationarity tests, analysis by sampling frequencies, as well as an assessment of the impact on forecasting model selection.