Selection of a stealthy and harmful attack function in discrete event systems

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Discrete Event Systems Analysis

Background:

• Partially-observed discrete event systems are vulnerable to attacks targeting state estimation.
• Attackers can manipulate sensor readings to mislead operators about the system's true state.
• Ensuring accurate state estimation is critical for system operation and security.

Purpose of the Study:

• To analyze the problem of joint state estimation under attack in discrete event systems.
• To develop methods for computing a stealthy joint subestimator that enables undetectable attacks.
• To demonstrate how to select a stealthy and harmful attack function.

Main Methods:

• Utilizing an automaton known as a joint estimator.
• Developing a method to compute a supremal stealthy joint subestimator.
• Defining and applying the concept of a harmful attack function.

Main Results:

• A method for computing a supremal stealthy joint subestimator is presented.
• The subestimator guarantees attacker stealthiness regardless of future system evolution.
• A technique for selecting a stealthy and harmful attack function is demonstrated.

Conclusions:

• The proposed joint estimator framework allows for the computation of stealthy subestimators.
• This enables the design of attacks that are both undetectable and misleading.
• The findings have implications for securing discrete event systems against sophisticated attacks.