Adaptive Kalman Filter for Dynamic Systems Localization with Skewed Heavy-Tailed Noise

Nowadays, the wireless localization (WL) has emerged as a popular solution to the indoor localization problem due to its low power consumption and ease of implementation. However, the accuracy of WL is compromised by dynamic uncertainties and skewed and/or heavy-tailed measurement noise, primarily caused by non-line-of-sight (NLOS) errors and multipath effects. In this paper, we introduce a fading factor and the shape parameter mixture (SPM) distribution to mitigate dynamic uncertainties and non-Gaussian measurement noise, respectively. Building on this, we derive an adaptive Kalman filter based on the fading factor within the variational Bayesian framework. The effectiveness of the proposed algorithm is validated through extensive simulations and experiments.