Looking for a Black Cat in a Dark Room: Security Visualization for Cyber-Physical System Design and Analysis

Georgios Bakirtzis, Brandon J. Simon, Cody Harrison Fleming, Carl R. Elks

TL;DR

This paper introduces an open-source security analyst dashboard designed to bridge the gap between system designers and security experts during the early stages of Cyber-Physical System (CPS) development. By integrating top-down systems-theoretic analysis with bottom-up vulnerability data, the tool allows teams to trace how a single software weakness could lead to catastrophic physical failures, like a drone losing flight control.

What Problem Does This Paper Address?

In the world of Cyber-Physical Systems—think autonomous vehicles, medical devices, or power grids—security is often treated as a "bolt-on" feature added late in the game. However, a security breach in a CPS isn't just a data leak; it’s a potential safety hazard.

The authors identify a major "complexity gap": during the early design phase, there is an overwhelming amount of data but very little specific information (like exact software versions or patches). Traditional security tools are built for "realized" systems that are already running. Furthermore, system designers (who care about safety and requirements) and security analysts (who care about exploits and vulnerabilities) often speak different languages. This paper aims to provide a "common language" through visualization to find the metaphorical "black cat" (a hidden vulnerability) in the "dark room" of complex system design.

Methods

The researchers developed a dashboard consisting of three interconnected interactive panes:

  1. System Topology ($\Sigma$): A graph representing the physical and digital components of the system. It highlights the "attack surface" (entry points for intruders) in red and traces potential "exploit chains" in yellow.
  2. Specification ($S$): A hierarchical view that maps mission-level requirements (e.g., "Safe Flight") to functional requirements and finally to specific hardware/software components. This allows users to see exactly which high-level goals are threatened if a specific component is compromised.
  3. Attack Vector Space ($AV$): A massive graph of known vulnerabilities and attack patterns. To make this readable, the authors used abstraction techniques to group thousands of specific vulnerabilities into broader classes.

The tool supports "What-if" analysis, allowing designers to swap out components (like changing a radio protocol) and immediately see how the system's attack surface and safety risks change.

Datasets

The dashboard pulls from several industry-standard "open datasets" to populate its attack vector space:

  • MITRE CAPEC: Common Attack Pattern Enumeration and Classification.
  • MITRE CWE: Common Weakness Enumeration (used to abstract specific bugs into classes).
  • NVD/CVE: The National Vulnerability Database and Common Vulnerabilities and Exposures.

By focusing on CWEs and CAPECs rather than the 114,000+ individual CVE entries, the authors reduced the data noise to a manageable level (~1,200 entries) without losing the ability to trace back to specific evidence.

Key Insights and Ideas

  • Traceability is King: The most powerful feature is the ability to click a low-level software weakness and see a highlighted path leading all the way up to a "Mission Failure." This makes the abstract risk of a bug tangible to a project manager or safety engineer.
  • Abstraction as a Filter: By using computational linguistics to map system models to the CWE/CAPEC hierarchies, the tool automatically filters out irrelevant attacks, leaving only what is "physically" possible within that specific system design.
  • The "Bucket" Workflow: The dashboard includes a "bucket" where analysts can collect interesting vulnerabilities to investigate further or include in a final report, facilitating better communication between technical and non-technical stakeholders.

Key Results

The authors validated their tool through a detailed use case involving an Unmanned Aerial System (UAS).

  • Qualitative Success: They demonstrated how a Systems Engineer (who knows the mission) and a Security Analyst (who knows the exploits) could use the tool together.
  • Discovery: In the case study, the team discovered that all radio modules shared a common vulnerability (ZigBee protocol). By visualizing this, they could see that a single exploit could cause the loss of three critical mission requirements simultaneously.
  • Design Iteration: The tool allowed them to simulate removing the ZigBee attribute, showing that while the radio modules were no longer part of the "attack surface," they still had associated vulnerabilities that needed hardening.

Limitations and Future Work

The authors acknowledge a few hurdles:

  • Expert Availability: Because CPS security is such a niche field, conducting a large-scale user study with qualified professionals has been difficult.
  • Design vs. Operation: The current tool is a "research tool" focused on the design phase. It does not yet integrate real-time monitoring data from systems that are already deployed.
  • Graph Complexity: While abstraction helps, the attack vector graphs can still become "hairballs" of data. The authors suggest integrating alternative rendering algorithms (like BioFabric) to improve navigation in future versions.

Summary updated

Abstract

Today, there is a plethora of software security tools employing visualizations that enable the creation of useful and effective interactive security analyst dashboards. Such dashboards can assist the analyst to understand the data at hand and, consequently, to conceive more targeted preemption and mitigation security strategies. Despite the recent advances, model-based security analysis is lacking tools that employ effective dashboards-to manage potential attack vectors, system components, and requirements. This problem is further exacerbated because model-based security analysis produces significantly larger result spaces than security analysis applied to realized systems-where platform specific information, software versions, and system element dependencies are known. Therefore, there is a need to manage the analysis complexity in model-based security through better visualization techniques. Towards that goal, we propose an interactive security analysis dashboard that provides different views largely centered around the system, its requirements, and its associated attack vector space. This tool makes it possible to start analysis earlier in the system lifecycle. We apply this tool in a significant area of engineering design-the design of cyber-physical systems-where security violations can lead to safety hazards.
BibTeX
@inproceedings{bakirtzis2018looking,
  author = {Georgios Bakirtzis and Brandon J. Simon and Cody Harrison Fleming and Carl R. Elks},
  title = {{Looking for a Black Cat in a Dark Room: Security Visualization for Cyber-Physical System Design and Analysis}},
  booktitle = {VizSEC},
  year = {2018},
  url = {https://openreview.net/forum?id=BNuzuj5pYve},
}

Loading related content...

Figure