Distributed correntropy Kalman filtering over sensor networks with FlexRay-based protocols

This paper focuses on distributed correntropy filtering for discrete-time stochastic systems subject to non-Gaussian noises over a sensor network. A novel filtering algorithm derived from the Gaussian maximum correntropy criterion is developed to handle non-Gaussian noises and possibly occurred outliers. Furthermore, the FlexRay mechanism consisting of both Round-Robin protocol and try-once-discard protocol is convened to realise high-efficient schedule of communication resources. And scalar filter gains are derived by using fixed-point iterative update rules, resulting in enhanced performance against non-Gaussian noises. A distributed approach to determine an optimal weight matrix is developed by using the upper bounds of the filtering error covariance, balancing the trade-off between estimated states and measurements at different scales. Finally, the efficacy of the developed filtering algorithm are validated through numerical simulations, demonstrating its significant advantage in real-world engineering applications.