Optimal Stealthy Attacks Against Remote State Estimation in Cyber-Physical Systems: A More Fine-Grained Stealthiness

Area of Science:

• Cyber-Physical Systems Security
• Control Theory
• Optimization

Background:

• Cyber-physical systems (CPS) are vulnerable to stealthy attacks.
• Existing stealthiness quantification (ϵ, δ)-stealthiness leads to complex, nonconvex optimization problems.
• Previous methods often require output matrices with full row rank.

Purpose of the Study:

• To address the stealthy attack design problem in CPS under (ϵ, δ)-stealthiness constraints.
• To simplify the optimization problem by transforming it into one with equality constraints.
• To develop an analytical solution for optimal attack strategies without requiring full row rank output matrices.

Main Methods:

• Analysis of the relationship between stealthiness parameters (ϵ, δ).
• Transformation of the nonconvex optimization problem into an equality-constrained problem.
• Derivation of analytical solutions for optimal attack parameters under a general linear attack model.

Main Results:

• A simplified, one-step solution process for the stealthy attack design problem.
• Achieved superior attack performance with enhanced stealthiness compared to existing methods.
• Presented completely analytical forms of optimal attack parameters, removing the full row rank constraint.

Conclusions:

• The proposed method effectively solves the (ϵ, δ)-stealthy attack design problem in CPS.
• The analytical solutions offer improved attack performance and stealthiness.
• The removal of the full row rank constraint broadens the applicability of the method.