Mobility-Aware Fusion for Robust Short-Term Spectrum Prediction in Mobile Autonomous Systems
IEEE Access · 2026 · 14 · 114937-114946
DOI: 10.1109/access.2026.3717013 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Nitin Rakheja. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Themen: Mobilisation
Abstract (Originalsprache)
Mobile autonomous systems rely on wireless communication for redundancy and interaction with external infrastructure while operating under continuous motion. Unlike human-driven platforms, their motion is continuously observable, introducing structure into channel variation that standard prediction approaches ignore. Existing methods treat short-term spectrum prediction as a temporal problem, yet in mobile autonomous systems part of the received signal strength (RSSI) variation is induced by motion itself. This work studies short-term RSSI prediction in mobile autonomous systems. A baseline Long Short-Term Memory (LSTM) model trained on historical RSSI is compared against a mobility-aware formulation that incorporates position through a simple propagation model. The two estimates are integrated using a residual-based weighting mechanism that adjusts their relative contribution based on recent prediction behavior. The framework is evaluated in a controlled simulation environment capturing stochastic motion, distance-dependent path loss, and measurement noise across multiple noise regimes and repeated Monte Carlo trials, and validated on real-world LTE measurements across six independent sessions. The mobility-aware formulation consistently reduces prediction error and exhibits lower variance than the baseline. The improvement persists as noise increases, though its relative magnitude diminishes at higher noise levels. Real-world validation confirms that the finding generalizes beyond simulation, with a mean improvement of 48.5% across six sessions ( $p=0.036$ , Cohen’s $d=1.276$ ). These results suggest that mobility is not merely a source of variation but a structured signal that improves short-term prediction without increasing model complexity. More stable channel estimates provide a basis for anticipatory link adaptation in heterogeneous environments.
Zitieren
APA
Rakheja, N. (2026). Mobility-Aware Fusion for Robust Short-Term Spectrum Prediction in Mobile Autonomous Systems. IEEE Access, 14, 114937-114946. https://doi.org/10.1109/access.2026.3717013
Vancouver
Rakheja N. Mobility-Aware Fusion for Robust Short-Term Spectrum Prediction in Mobile Autonomous Systems. IEEE Access. 2026;14:114937-114946. doi:10.1109/access.2026.3717013
BibTeX
@article{Rakheja2026,
author = {Rakheja, Nitin},
title = {Mobility-Aware Fusion for Robust Short-Term Spectrum Prediction in Mobile Autonomous Systems},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {114937--114946},
doi = {10.1109/access.2026.3717013}
}
EUVIMED