Abstract
Alert correlation is a process that analyzes the raw alerts produced by one or more intrusion detection systems, reduces nonrelevant ones, groups together alerts based on similarity and causality relationships between them and finally makes aconcise and meaningful view of occurring or attempted intrusions. Unfortunately, most correlation approaches use just a few components that aim only specific correlation issues and so cause reduction in correlation rate. This paper uses a general correlation model that has already been presented in [9] and is consisted of a comprehensive set of components. Then some changes are applied in the component that is related to multi-step attack scenario to detect them better and so to improve semantic level of alerts. The results of experiments with DARPA 2000 data set obviously show the effectiveness of the proposed approach. DOI: http://dx.doi.org/10.11591/ijece.v3i4.2771
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have