Distributed Filtering Over Sensor Networks With Byzantine Attacks: A Token Bucket Protocol.

Area of Science:

• Control Systems Engineering
• Network Security
• Signal Processing

Background:

• Sensor networks are vulnerable to Byzantine attacks, compromising data integrity.
• Time-varying state-saturated systems require robust filtering techniques.
• Token Bucket Protocol (TBP) regulates data transmission with stochastic packet sizes.

Purpose of the Study:

• To investigate distributed filtering for state-saturated systems under Byzantine attacks.
• To develop a robust filtering strategy using TBP to mitigate attacks.
• To establish an upper bound for filtering error covariance (FEC) and optimize filter gains.

Main Methods:

• Formulation of a Byzantine attack model targeting measurement signals.
• Utilization of TBP to manage data transmission based on token availability.
• Minimization of the filtering error covariance (FEC) upper bound to compute filter gains.
• Matrix-based theoretical analysis to ensure boundedness of filtering error dynamics.

Main Results:

• An effective method for constructing an upper bound of the filtering error covariance (FEC) was developed.
• Suitable filter gains were computed by minimizing the derived FEC bound.
• The boundedness of the filtering error dynamics was rigorously proven.
• Numerical simulations confirmed the proposed algorithm's effectiveness.

Conclusions:

• The proposed distributed filtering algorithm effectively handles Byzantine attacks in sensor networks.
• The TBP integration ensures reliable data transmission under varying network conditions.
• The developed theoretical framework guarantees the robustness and stability of the filtering error dynamics.