Abstract

Carefully logging network activity is essential to meet the requirements of high security and optimal resource availability. However, detecting break-in attempts within this activity is a difficult task. Making the distinction between misuse and normal use is hard, and identifying intrusions that use novel attacks is fundamentally difficult.

In this paper, we introduce a visual approach for analyzing network activity. This approach differs from anomaly and misuse detection because it considers human factors to support the exploration of network traffic. Our prototype application is based on an unsupervised neural network and consequently does not rely on any prior knowledge of the data being analyzed. We use self-organizing maps to project the network events on a space appropriate for visualization, and achieve their exploration using a map metaphor. The approach we present can be used to analyze past and present activities, as well as to show trends in the events.

To demonstrate the usability of our tools, we describe the investigation of a dataset containing common intrusion patterns. We also discuss some weaknesses of current intrusion detection systems and propose a new paradigm for monitoring network activity that enables the discovery of new, sophisticated, and structured attacks.

• View the full text of this paper in HTML form and PDF form.

• If you need the latest Adobe Acrobat Reader, you can download it from Adobe's site.

• To become a USENIX Member, please see our Membership Information.