Skip to content
Open access

Edge-Oriented Tree-Guided Temporal Fusion for GNSS Spoofing Detection in UAV-Class Consumer Receivers

Nov 2026 · IEEE transactions on consumer electronics · Vol 72, pp. 12500-12509 · 0 citations · 27 references

Abstract

Low-cost Global Navigation Satellite System (GNSS) receivers are increasingly used in consumer uncrewed aerial vehicles (UAVs), compact autonomous platforms, and location-aware consumer devices, where spoofed observables can affect localisation, return-to-home logic, route execution, and geofencing. Existing anti-spoofing methods often depend on auxiliary sensors, multi-antenna front-ends, specialised receiver access, or platform-specific fusion, which restricts their practicality on cost- and resource-constrained consumer platforms. To address this gap, this paper proposes TGTF-NET, an edge-oriented Tree-Guided Temporal Fusion Network for receiver-centric GNSS spoofing detection from synchronised eight-channel receiver observations under uncrewed aerial system (UAS)-emulated conditions. The design integrates shared bidirectional recurrent temporal encoding, channel-validity-aware attention and max pooling, differentiable threshold-bank evidence that approximates tree-style splits, residual temporal-summary fusion, and auxiliary channel-level supervision within a receiver-side detector with validation-selected thresholding. The method is evaluated on the curated Mendeley v3 UAS spoofing dataset using mission-aware segmentation, classical and neural baselines, repeated-seed stability, ablation, robustness, calibration, and cross-segment validation. On the held-out test split, TGTF-NET achieves an F1-score of 0.9505, recall of 0.9808, receiver operating characteristic area under the curve (ROC AUC) of 0.9555 and precision–recall area under the curve (PR AUC) of 0.9883 at a validation-selected operating threshold. Across repeated seeds, it obtains a mean F1-score of $0.9377~\pm ~0.0074$ . Within the evaluated protocol, TGTF-NET gives the highest recall and F1-score among the thresholded models, while gradient boosting retains higher ROC AUC and PR AUC. The results support TGTF-NET as a receiver-side anti-spoofing component for UAV-class consumer GNSS devices and indicate its relevance to location-oriented security in future edge-enabled and Beyond-6G systems that depend on trustworthy positioning, with remaining work on calibration, false-alarm reduction, embedded-hardware validation, and cross-segment generalisation.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.