Critical Path Prioritization Dashboard for Alert-driven Attack Graphs

Abstract

Although intrusion alerts can provide threat intelligence regarding attacker strategies, extracting such intelligence via existing tools is expensive and time-consuming. Earlier work has proposed SAGE, which generates attack graphs from intrusion alerts using unsupervised sequential machine learning. This paper proposes a querying and prioritization-enabled visual analytics dashboard for SAGE. The dashboard has three main components: (i) a Graph Explorer that presents a global view of all attacker strategies, (ii) a Timeline Viewer that correlates attacker actions chronologically, and (iii) a Recommender Matrix that highlights prevalent critical alerts via a MITRE ATT&CK-inspired attack stage matrix. We describe the utility of the proposed dashboard using intrusion alerts collected from a distributed multi-stage team-based attack scenario. We evaluate the utility of the dashboard through a user study. Based on the responses of a small set of security practitioners, we find that the dashboard is useful in depicting attacker strategies and attack progression, but can be improved in terms of usability.