Supervisory Nonlinear State Observers for Adversarial Sparse Attacks.

Area of Science:

• Control Systems Engineering
• Cyber-Physical Systems Security
• Estimation Theory

Background:

• Secure state estimation is crucial for continuous-time linear systems.
• Existing methods struggle with dynamic and unpredictable sensor attacks.
• Adversaries can alter the set of attacked sensors, posing a significant challenge.

Purpose of the Study:

• To develop a robust state estimation method for linear systems under sparse, time-varying sensor attacks.
• To design a supervisory observer capable of adapting to evolving adversarial strategies.
• To guarantee the convergence of the state estimation in the presence of attacks and disturbances.

Main Methods:

• A novel supervisory state observer is proposed.
• The observer utilizes a bank of candidate nonlinear subobservers.
• A monitoring function and switching logic dynamically select the active subobserver.

Main Results:

• The supervisory observer ensures asymptotic convergence to a neighborhood of the true system state.
• The method is effective even when the set of attacked sensors changes over time.
• Theoretical results are substantiated through a simulation example.

Conclusions:

• The proposed supervisory observer provides a robust solution for secure state estimation against dynamic sensor attacks.
• This approach enhances the resilience of control systems in adversarial environments.
• The method offers a promising direction for securing cyber-physical systems.