Deep-Ensemble-Learning-Based GPS Spoofing Detection for Cellular-Connected UAVs

Unmanned aerial vehicles (UAVs) are an emerging technology in the 5G-and-beyond systems with promise of assisting cellular communications and supporting IoT deployment remote density areas. Safe secure navigation is essential for UAV autonomous deployment. Indeed, opensource simulator can use commercial software-defined radio tools to generate fake global positioning system (GPS) signals spoof GPS receiver calculate wrong locations, deviating from planned trajectory. Fortunately, existing mobile provide additional cellular-connected UAVs verify locations spoofing detection, but it needs at least three base stations (BSs) same time. In this article, we propose a novel deep-ensemble-learning-based, mobile-network-assisted monitoring tracking detection. The proposed method uses path losses between BSs communication indicate trajectory deviation caused by spoofing. To increase detection accuracy, statistics methods adopted remove environmental impacts on losses. addition, deep ensemble learning deployed edge cloud servers multilayer perceptron (MLP) neural networks analyze statistical features making final decision, which has no requirements energy consumption UAVs. experimental results show effectiveness our detecting spoofing, achieving above 97% accuracy rate under two BSs, while still achieve 83% only one BS.