Iterative Diffusion-Based Distributed Cubature Gaussian Mixture Filter for Multisensor Estimation.

Area of Science:

• * Signal Processing
• * Estimation Theory
• * Sensor Networks

Background:

• * Multisensor systems require robust data fusion techniques for accurate state estimation.
• * Gaussian mixture models are effective for representing uncertainties in complex systems.
• * Existing distributed filters often lack efficiency in information exchange and convergence.

Purpose of the Study:

• * To propose a novel distributed cubature Gaussian mixture filter with iterative diffusion (DCGMF-ID).
• * To enhance multisensor estimation and information fusion capabilities.
• * To analyze the convergence properties and information-theoretic interpretation of the proposed filter.

Main Methods:

• * Utilizing a high-degree cubature Kalman filter for accurate Gaussian mixture component estimation at each sensor node.
• * Implementing an iterative diffusion scheme for fusing mean and covariance of Gaussian components.
• * Employing multiple iterative information exchanges among neighboring sensor nodes for enhanced fusion.

Main Results:

• * The DCGMF-ID demonstrates superior performance compared to noniterative diffusion methods.
• * Iterative information exchange significantly benefits the fusion process.
• * The proposed filter outperforms DCGMF-AC and DCGMF-ICI under limited iteration conditions.

Conclusions:

• * The DCGMF-ID provides an effective approach for multisensor estimation and information fusion.
• * Iterative diffusion enhances the accuracy and convergence of distributed filtering.
• * The method offers advantages in scenarios with computational constraints.