Czech Researchers Innovate Cybersecurity – Detecting Network Threats in Advance

Researchers Ing. Josef Koumar and Ing. Jaroslav Pešek from the Faculty of Information Technology at the Czech Technical University in Prague (FIT CTU), together with Ing. Kamil Jeřábek, Ph.D., and Ing. Jiří Setinský from the Faculty of Information Technology at the Brno University of Technology (FIT VUT), combined their efforts and presented an innovation workshop focused on detecting network anomalies using forecasting methods within the interdisciplinary Network for Cybersecurity (NeCS) PhD School. The main goal of this joint initiative was to introduce a new methodology that allows for the early detection of deviations in network traffic and to prevent potential threats.

Forecasting is based on processing historical data and continuously updating the model, which enables the prediction of the most likely future development of network traffic in monitored networks. In cybersecurity, this prediction is used to quickly identify unexpected changes — whether it’s attempts to breach a network or other types of cyberattacks, which manifest as atypical load or altered characteristics of transmitted data and are difficult to detect.

„Our primary goal was to demonstrate that statistical modeling and forecasting, combined with modern data processing tools, can effectively detect even the subtlest anomalies,” says Ing. Jaroslav Pešek from FIT CTU.

„Thanks to this method, we can predict in advance where non-standard network behavior occurs and react in time to potential cybersecurity incidents,” adds Ing. Josef Koumar from FIT CTU.

Workshop participants were able to practically test the forecasting method on real network traffic samples to better understand how it can predict unexpected changes. Both students and experts learned ways to integrate this predictive technology into standard IT security tools, thereby significantly enhancing their effectiveness.

„The workshop was very useful because it covered the necessary fundamental knowledge, data explanations, and practical experience with coding and training models. It was easy to keep up because the organizers paid great attention to the participants,” says Shiva Azizzadeh from Radboud University.

The importance of methods like forecasting is further confirmed by the fact that the related research paper has been accepted at the internationally recognized NOMS 2025 conference, where its results will be presented to a broader expert audience from around the world.

The person responsible for the content of this page: Bc. Veronika Dvořáková