Distributed Finite-Horizon Extended Kalman Filtering for Uncertain Nonlinear Systems

Area of Science:

• Control Systems Engineering
• Signal Processing
• Networked Systems

Background:

• State estimation is crucial for discrete nonlinear systems, especially in sensor networks.
• Existing methods often struggle with model uncertainties and computational demands.
• Distributed filtering offers a decentralized approach to enhance robustness and scalability.

Purpose of the Study:

• To develop a novel robust distributed extended Kalman filter for discrete nonlinear systems.
• To address norm-bounded uncertainties in system models and their approximations.
• To ensure reliable state estimation performance in sensor network environments.

Main Methods:

• A robust distributed extended Kalman filter is designed, incorporating neighbor information for each sensor.
• The filter optimizes the upper bound of the estimation error covariance.
• A user-friendly sufficient condition is derived based on system properties, avoiding step-by-step covariance calculations.

Main Results:

• The proposed filter effectively handles norm-bounded uncertainties in system models and Taylor expansions.
• Estimation performance is guaranteed through the developed algorithm.
• Simulation results validate the effectiveness and robustness of the filtering approach.

Conclusions:

• The novel robust distributed extended Kalman filter provides an effective solution for state estimation in discrete nonlinear sensor networks.
• The algorithm's design simplifies implementation by relying on system properties rather than iterative calculations.
• This work contributes to improved state estimation in complex, uncertain networked systems.