An Availability-Aware Adaptive Hybrid Localisation Framework with Structurally Invariant Decision Logic for GPS-Degraded Emergency Environments

Pedestrian localisation in emergency environments is challenged by degraded Global Positioning System (GPS) availability, mobile-anchor uncertainty, radio-frequency interference, non-line-of-sight distortion, and inertial drift. Reporting only valid-estimate error may conceal localisation outages and recovery burden. This study proposes Hybrid S9, an adaptive, device-centric localisation framework with a structurally invariant hierarchical controller that selects among GPS, Bluetooth Low Energy (BLE)-assisted Extended Kalman Filter (EKF) correction, zero-velocity update (ZUPT), kinematic coasting, and Particle Filter (PF) recovery according to measurement quality, uncertainty, and mobility state. The framework is evaluated in Normal, Urban Canyon, and Search and Rescue regimes using valid-estimate mean absolute error (MAE), availability, penalised mean error (PME), and operation cost. In the degraded regimes, Hybrid S9 maintains near-complete localisation availability while reducing operation cost. Dedicated severe-degradation experiments verify PF activation, posterior state and covariance transfer, and controlled de-escalation.