Application Error Detection in Networks by Protocol Behavior Model

The identification of causes of errors in network systems is difficult due to their inherent complexity. Network administrators usually rely on available information sources to analyze the current situation and identify possible problems. Even though they are able to identify the symptoms seen in the past and thus can apply their experience gathered from the solved cases the time needed to identify and correct the errors is considerable. The automation of the troubleshooting process is a way to reduce the time spent on individual cases. In this paper, the model that can be used to automate the diagnostic process of network communication is presented. The model is based on building the finite automaton to describe protocol behavior in various situations. The unknown communication is checked against the model to identify error states and associated descriptions of causes. The tool prototype was implemented in order to demonstrate the proposed method via a set of experiments.