Trust-Aware Adaptive Event-Triggered Unscented Kalman Filtering for Target Tracking in Mobile Wireless Sensor Networks

To mitigate the degradation of target tracking performance caused by modeling errors and measurement anomalies in resource-constrained wireless sensor network (WSN), this paper proposes a trust-aware adaptive event-triggered unscented Kalman filtering method. First, a trust-aware adaptive triggering mechanism based on a time-varying received signal strength (RSS) response radius is designed, enabling the mobile target to dynamically schedule and activate an approximately prescribed number of trusted responding anchors for data transmission, thereby adapting to their spatial distribution. Second, a K-means-based dimensionality reduction robust square-root unscented Kalman fusion filtering (RSUKF) algorithm is developed. This algorithm compensates for uncertainties induced by modeling errors through weighted averaging of multiple sigma points generated via uniform random sampling. Furthermore, it employs the dissimilarity between multiple local posterior estimates and the fused prior estimate as features for two-cluster K-means clustering, facilitating the identification of trustworthy local posterior estimates with low dissimilarity for participation in the weighted fusion process. Finally, numerical simulation results demonstrate that the proposed method not only ensures an approximately prescribed number of trusted responding anchors but also significantly improves the stability, robustness, and accuracy of target tracking.