Covariance-Based Estimation for Clustered Sensor Networks Subject to Random Deception Attacks

Area of Science:

• Control Systems Engineering
• Signal Processing
• Networked Systems

Background:

• Distributed fusion estimation is crucial for networked systems.
• Random deception attacks pose a significant threat to data integrity.
• Existing methods may lack robustness in clustered sensor networks.

Purpose of the Study:

• To develop a cluster-based distributed fusion estimation algorithm for discrete-time stochastic signals.
• To address the challenge of random deception attacks in networked sensor systems.
• To enhance filtering and fixed-point smoothing performance under adversarial conditions.

Main Methods:

• A two-stage cluster-based approach is proposed.
• Local estimators are designed using an innovation approach with least-squares.
• Fusion estimators are generated at a global center by combining local estimates.
• Covariance-based design utilizes first and second-order moments.

Main Results:

• The proposed method effectively performs distributed fusion estimation (filtering and smoothing).
• The algorithms are robust to random deception attacks with known probabilities.
• Performance analysis demonstrates the impact of attack success probability.

Conclusions:

• The cluster-based fusion estimation strategy provides a robust solution for networked systems facing deception attacks.
• The method offers improved estimation accuracy without requiring full signal evolution models.
• This approach enhances the reliability of signal processing in compromised environments.