Robust Indoor-Outdoor Seamless Positioning Using Adaptive Cubature Particle Filter and Variational Bayesian in 5G-IoT Networks

Accurate positioning in 5G-enabled IoT systems remains a critical challenge, particularly in indoor–outdoor scenarios characterized by nonlinear system dynamics and non-Gaussian noise. In this letter, we propose a Variational Bayesian Adaptive Cubature Particle Filter (VB-CK-APF) to deliver robust and seamless positioning performance under such complex conditions. A novel adaptive weight adjustment mechanism is proposed that dynamically modulates particle weights according to real-time measurement uncertainty and system variability. This approach effectively addresses particle degeneracy and preserves sample diversity, which are common shortcomings of conventional particle filtering techniques. To rigorously implement the proposed framework, the method incorporates the Cubature Kalman Filter to construct an efficient proposal distribution and employs variational Bayesian inference for the estimation of unknown, time-varying measurement noise covariance. The integration of these components enables the filter to adapt to complex and evolving environments while maintaining high positioning accuracy. Extensive simulations and real-world experiments demonstrate that the VB-CK-APF algorithm consistently achieves positioning errors below 10 cm, even in challenging indoor–outdoor transition scenarios.