Adaptive event-triggered robust distributed filter for nonlinear systems with non-stationary heavy-tailed noise

This paper investigates a novel adaptive event-triggered robust distributed filtering approach for nonlinear sensor networks under communication congestion and nonstationary heavy-tailed noise. To simultaneously ensure high estimation accuracy and low communication overhead, a hybrid model-based adaptive event-triggered mechanism is developed, enabling the dynamic adjustment of triggering thresholds and filter gains based on noise characteristics. A novel distributed filter based on this mechanism is then derived using a sequential fast covariance fusion scheme. Notably, to tackle the nonlinear integration challenge introduced by Student’s t-distribution weighting in the proposed algorithm, a new numerical integration method is proposed by combining cubature rule with Gauss–Laguerre quadrature, achieving high-accuracy approximation. Subsequently, the algorithm’s boundedness is analyzed, and sufficient conditions for mean-square exponential stability are provided. Finally, tracking simulations involving multiple unmanned aerial vehicles validate the effectiveness and superiority of the proposed method.