SATSense: Multi-Satellite Collaborative Framework for Spectrum Sensing

TL;DR

This work tackles dynamic spectrum sharing for LEO satellite networks by proposing SATSense, a multi-satellite collaborative spectrum sensing framework. It blends graph learning to model inter-satellite data correlations (GLSS), multi-coset sub-Nyquist sampling with a deep autoencoder for data compression, and a contrastive learning–augmented autoencoder to mitigate packet loss during satellite-to-ground transmission. Empirical results show that GLSS outperforms baselines in spectrum-sensing accuracy, CAE offers superior raw-data recovery under packet loss and preserves timeliness, and the overall SATSense design effectively navigates RF-heterogeneity, data-volume bottlenecks, and transmission unreliability in LEO CSS scenarios. The approach demonstrates the potential for fast, accurate spectrum sensing across wide bands and multiple satellites, enabling robust dynamic spectrum sharing in non-terrestrial networks.

Abstract

Low Earth Orbit satellite Internet has recently been deployed, providing worldwide service with non-terrestrial networks. With the large-scale deployment of both non-terrestrial and terrestrial networks, limited spectrum resources will not be allocated enough. Consequently, dynamic spectrum sharing is crucial for their coexistence in the same spectrum, where accurate spectrum sensing is essential. However, spectrum sensing in space is more challenging than in terrestrial networks due to variable channel conditions, making single-satellite sensing unstable. Therefore, we first attempt to design a collaborative sensing scheme utilizing diverse data from multiple satellites. However, it is non-trivial to achieve this collaboration due to heterogeneous channel quality, considerable raw sampling data, and packet loss. To address the above challenges, we first establish connections between the satellites by modeling their sensing data as a graph and devising a graph neural network-based algorithm to achieve effective spectrum sensing. Meanwhile, we establish a joint sub-Nyquist sampling and autoencoder data compression framework to reduce the amount of transmitted sensing data. Finally, we propose a contrastive learning-based mechanism compensates for missing packets. Extensive experiments demonstrate that our proposed strategy can achieve efficient spectrum sensing performance and outperform the conventional deep learning algorithm in spectrum sensing accuracy.

Figures (16)

• Figure 1: A scenario of collaborative spectrum sensing with multiple LEO satellites.
• Figure 2: The curve showing the variation over time of SNR and the corresponding spectrum sensing accuracy for two satellites.
• Figure 3: The Pearson coefficient of sensing data under different Doppler shifts.
• Figure 4: The measured downlink transmission rate using the GS of Starlink.
• Figure 5: The overview of SATSense framework.