Watermarking-Based Protection Strategy Against Stealthy Sensor Attacks on Zonotopic Set-Membership State Estimation.

Area of Science:

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

Background:

• Cyber-physical systems (CPSs) are vulnerable to stealthy false data injection (FDI) sensor attacks.
• Traditional state estimation methods can be compromised by sophisticated attackers.
• Zonotopic set-membership state estimation (ZSMSE) offers a robust approach but requires protection against FDI.

Purpose of the Study:

• To investigate the vulnerability of ZSMSE to stealthy FDI sensor attacks in CPSs.
• To define conditions under which FDI attacks can bypass detectors and compromise ZSMSE.
• To propose a watermarking-based strategy to defend ZSMSE against such attacks.

Main Methods:

• Development of a ZSMSE approach enclosing the system state within a parallelotope.
• Formal definition of stealthiness for FDI attacks and vulnerability for ZSMSE.
• Derivation of necessary and sufficient conditions for ZSMSE vulnerability.
• Design of a watermarking matrix for attack stealthiness disruption and ZSMSE convergence.

Main Results:

• Characterization of FDI attack stealthiness and ZSMSE vulnerability.
• Identification of conditions making ZSMSE susceptible to attacks.
• Demonstration of a watermarking strategy effectively countering stealthy FDI attacks.
• Validation of the proposed strategy through illustrative examples.

Conclusions:

• ZSMSE in CPSs is susceptible to stealthy FDI attacks.
• The proposed watermarking strategy effectively mitigates FDI attacks while ensuring ZSMSE convergence.
• The findings provide a robust defense mechanism for secure state estimation in CPSs.