Maximum Correntropy Filtering for Complex Networks With Uncertain Dynamical Bias: Enabling Componentwise...

Area of Science:

• Control Systems Engineering
• Signal Processing
• Networked Systems

Background:

• Nonlinear complex networks often face challenges from non-Gaussian noise and uncertain biases.
• Efficient resource utilization is crucial in networked systems with limited bandwidth and energy.

Purpose of the Study:

• To develop a dynamic event-triggered recursive filtering scheme for nonlinear complex networks.
• To attenuate the effects of non-Gaussian noises using the maximum correntropy criterion.
• To address uncertain dynamical bias in filtering applications.

Main Methods:

• A componentwise dynamic event-triggered transmission (DETT) protocol was adopted for efficient data sharing.
• A novel correntropy-based performance index (CBPI) was proposed to quantify impacts of DETT, nonlinearity, and bias.
• Filter gain was designed by maximizing the CBPI, incorporating derived upper bounds on prediction error and noise covariance.

Main Results:

• The proposed correntropy-based performance index effectively integrates system uncertainties and the DETT mechanism.
• The dynamic event-triggered maximum correntropy filter (MCF) demonstrates significant noise attenuation capabilities.
• An illustrative example validated the feasibility and effectiveness of the developed MCF scheme.

Conclusions:

• The developed dynamic event-triggered MCF scheme offers a robust solution for nonlinear complex networks under non-Gaussian noise.
• The componentwise DETT protocol enhances resource efficiency without compromising filtering performance.
• The proposed CBPI provides a valuable metric for designing event-triggered filters in complex systems.