Multimodal RF Fingerprinting for IoT Devices in Satellite-Based Sensing

The rapid expansion of the Internet of Things (IoT) presents critical challenges in device authentication, network security, and wide-area visibility. While terrestrial solutions have been extensively explored, IoT visibility via Non-Terrestrial Network (NTN) platforms remains underdeveloped, despite the significance of NTN in regions lacking terrestrial communication infrastructure. To address this gap, and accounting for the complexities of satellite communication channel, this work proposes a framework that enables signal-based RF fingerprinting for IoT device classification via satellites by extracting key features from the received signals. The proposed framework integrates MUSIC-based Direction of Arrival (DoA) estimation, a Support Vector Machine (SVM) classifier, and signal processing techniques to extract key RF features, including DoA, modulation type, frequency, and Received Signal Strength Indicator (RSSI). These features are subsequently clustered using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to classify unique transmitters. The results demonstrate high classification accuracy, even under low Signal-to-Noise Ratio (SNR) conditions, providing a scalable solution for IoT device monitoring and spectrum awareness in satellite-based communications.