Analysing network-wide anomalies using dependency graphs and baseline

In this paper, it is presented a solution that performs the alarm correlation based in the dependencies between the network devices using baseline or digital signature of network segment (DSNS). The main objective of this work is to provide a network-wide view, showing its global status instead of just presenting the various alarms generated individually for different network elements. A dependency graph is used in order to correlate these alarms, increasing the semantic content of resulting notifications. Besides, we also present a visual approach for the network administrator, which reflects the network global status, enabling network administrators to visualize the anomaly propagation in the network. Tests were performed in the State University of Londrina network and the results of the experiments have been widely satisfactory.