Abstract: The security of cyber-physical systems is first and foremost a safety problem, yet it is typically handled as a traditional security problem, which means that solutions are based on defending against threats and are often implemented too late. This approach neglects to take into consideration the context in which the system is intended to operate, thus system safety may be compromised. This paper presents a systems-theoretic analysis approach that combines stakeholder perspectives with a modified version of Systems-Theoretic Accident Model and Process (STAMP) that allows decision-makers to strategically enhance the safety, resilience, and security of a cyber-physical system against potential threats. This methodology allows the capture of vital mission-specific information in a model, which then allows analysts to identify and mitigate vulnerabilities in the locations most critical to mission success. We present an overview of the general approach followed by a real example using an unmanned aerial vehicle conducting a reconnaissance mission.
Introduction
As Cyber-Physical Systems (CPS)—such as autonomous vehicles, industrial control systems, and unmanned aerial vehicles (UAVs)—become increasingly complex and interconnected, traditional security paradigms are proving insufficient. Current security practices often mirror standard IT security, focusing on a reactive "bottom-up" approach that identifies and patches individual software vulnerabilities. However, in the context of CPS, a security breach is frequently a safety failure in disguise. This paper argues for a shift toward a top-down, mission-centric approach that treats security as a fundamental subset of system safety.
The Limitations of Traditional CPS Security
Traditional security methodologies for CPS suffer from three critical shortcomings:
- Mission Agnosticism: Standard methods often fail to distinguish between a vulnerability that could crash the entire system and one that has no impact on the specific mission objectives. Without mission context, resource allocation for security is often inefficient.
- Late-Stage Implementation: Security is frequently treated as an afterthought, "bolted on" after the design phase is complete. This leads to significantly higher costs and lower overall effectiveness.
- The Safety-Security Blur: In physical systems, the distinction between a malicious exploit and a random component failure is often irrelevant to the outcome. Both can lead to the same catastrophic physical consequences, yet they are often managed by different teams using different tools.
As shown in Figure 1, the ability to impact system cost and performance is highest during the initial concept and requirements phases. By the time a system reaches the build or operate phase, the cost of design changes skyrockets while the effectiveness of those changes diminishes.
Proposed Methodology: The War Room and STPA-Sec
The authors propose a two-phase methodology designed to elicit security requirements early in the lifecycle, grounded in the specific operational needs of the mission.
Phase 1: The War Room
The "War Room" is a structured elicitation exercise that brings together a diverse group of stakeholders, including designers, commanders, operators, and maintenance personnel. The objective is to define the Concept of Operations (ConOps) and identify "unacceptable losses." By focusing on what must not happen for the mission to succeed, the analysis remains grounded in operational reality rather than abstract technical threats.
Phase 2: STPA-Sec Modeling
The outputs from the War Room are integrated into a modified version of Systems-Theoretic Process Analysis for Security (STPA-Sec). Based on Nancy Leveson’s STAMP model, STPA-Sec treats security as a control problem rather than a component-failure problem.
The process follows a rigorous logical chain (see Figure 2):
- Identify Unacceptable Losses: Defining high-level mission failures (e.g., loss of a UAV in hostile territory).
- Identify Hazards: Determining the system states that lead to those losses.
- Analyze Control Actions: Identifying the specific commands—whether human or automated—that could trigger these hazards.
- Derive Constraints: Formulating specific requirements to prevent unsafe or insecure control actions.
Case Study: UAV Reconnaissance
The methodology was validated through a qualitative case study of a small UAV tasked with a tactical reconnaissance mission. By applying the STPA-Sec framework, the authors identified three primary categories of unacceptable loss: loss of resources (L1), loss of sensitive technology or information (L2), and loss of control leading to collateral damage (L3).
From these losses, the team derived specific hazards, such as the UAV straying into an inappropriate area (H3). This led to the creation of concrete security constraints. For example, one derived constraint (CA 1.1) requires that the mission planner define an area of interest that is congruent with the area of operations, preventing navigation errors that could be exploited to cause a mission-critical hazard.
Conclusion and Future Work
By shifting the focus from individual threats to mission-level outcomes, this systems-theoretic approach allows for the creation of more resilient and cost-effective CPS. While the current process is largely manual and qualitative, it provides a necessary foundation for future advancements. Future research directions include the automation of "War Room" data processing, the use of formal methods to verify derived constraints, and the testing of specific technical solutions—such as encryption or redundant control loops—to ensure they satisfy mission-centric requirements.
Summary updated