Robust Rauch--Tung--Striebel Smoothers Based on Generalized Statistical Measure Under Cyberattacks

Area of Science:

• Control Systems Engineering
• Cybersecurity
• Signal Processing

Background:

• State estimation (SE) in non-Gaussian (NG) noise systems is difficult.
• Hybrid cyberattacks further complicate SE, often causing detector failures.

Purpose of the Study:

• To address the fixed-interval smoothing problem for NG systems under hybrid cyberattacks.
• To improve SE accuracy and detector reliability in compromised systems.

Main Methods:

• Proposed a flag-bit-based detection mechanism using a marking signal in the measurement equation.
• Derived robust forward filtering and backward smoothing using new cost functions based on a generalized statistical measure (GSM).
• Developed a new robust Rauch-Tung-Striebel smoother.

Main Results:

• The proposed detector demonstrated a lower false detection rate.
• The novel smoother achieved improved estimation accuracy compared to existing methods.
• Sufficient conditions for the convergence of forward and backward passes were rigorously established.

Conclusions:

• The developed flag-bit detector and robust smoother effectively handle non-Gaussian noise and hybrid cyberattacks.
• The approach provides theoretical guarantees for optimality and uniqueness in state estimation.
• Simulations confirm the practical advantages under diverse attack and noise conditions.