Distributed Recursive Filtering Over Sensor Networks Under Random Access Protocol: When State Saturation Meets...

Area of Science:

• Control Systems Engineering
• Networked Systems
• Signal Processing

Background:

• Distributed filtering is crucial for sensor networks, but faces challenges like state saturation and measurement censoring.
• Existing methods struggle with the complexity of time-varying systems and the need for efficient data transmission.

Purpose of the Study:

• To develop a novel distributed filtering algorithm for state-saturated, time-varying systems with Tobit-modeled censored measurements.
• To ensure guaranteed upper bounds on filtering error covariances under a random access protocol (RAP).
• To design filter parameters that effectively handle both measurement censoring and state saturation.

Main Methods:

• Implementation of a random access protocol (RAP) to manage sensor node transmissions and reduce communication burden.
• Utilizing matrix difference equations to derive and minimize upper bounds on filtering error covariances.
• Employing matrix simplification techniques to address sparsity in network topology.

Main Results:

• The proposed algorithm successfully guarantees upper bounds on filtering error covariances for the studied systems.
• Filter parameters were effectively designed to manage state saturation and Tobit measurement censoring.
• The simulation example validated the practical applicability of the distributed filtering approach.

Conclusions:

• The developed distributed filtering approach effectively addresses state saturation and measurement censoring in time-varying sensor networks.
• The use of RAP and matrix-based methods provides a robust framework for guaranteed filtering performance.
• The algorithm demonstrates significant potential for real-world applications in networked control systems.