Publication Details
Using Rule-Based Decision Trees for Automatic Passive Diagnostics of the Network Problems
Network diagnostics, passive network monitoring, rule-based diagnostics, fault tree analysis, event-based diagnostics, decision trees.
Network troubleshooting often requires a detailed analysis that may involve network packet capturing and a manual analysis using tools such as Wireshark. This is time-consuming and requires deep knowledge of communication protocols. Therefore this domain is a suitable candidate for the deployment of an expert system. In this paper, we consider a rule-based system integrating the expert knowledge that performs an automatic root cause analysis of network problems identifiable from network communications. The system is open, thus it is possible to add new rules as needed, e.g., for specific and recurring cases of a target environment. The rules are evaluated in a tree-based fashion, which enables us to collect additional information during the problem search to better explain the possible causes. We successfully deployed the tool as part of a commercial tool for network monitoring.
@ARTICLE{FITPUB12221, author = "Martin Holkovi\v{c} and Ond\v{r}ej Ry\v{s}av\'{y}", title = "Using Rule-Based Decision Trees for Automatic Passive Diagnostics of the Network Problems", pages = "1--10", booktitle = "International Journal On Advances in Networks and Services", journal = "International Journal on Advances in Networks and Services", volume = 2020, number = 1, year = 2020, ISSN = "1942-2644", language = "english", url = "https://www.fit.vut.cz/research/publication/12221" }