Location-Prior-Aided Grant-Free Detection in Satellite IoT Networks: A Hierarchical Cooperative Architecture With Spatial Correlation

Satellite IoT has emerged as a promising solution to extend connectivity beyond terrestrial networks, particularly in remote and underserved regions. However, grant-free random access (GF-RA) in satellite networks faces two key challenges: severe interference from massive uncoordinated terminals and limited computational resources on satellite platforms. In contrast to terrestrial multipath channels, the dominant line-of-sight propagation in satellite-terrestrial links induces deterministic spatial correlations between terminal positions and angles-of-arrival (AoAs). To leverage this property, a hierarchical satellite-terrestrial cooperative processing architecture is proposed, where computationally intensive terminal positioning tasks are offloaded to the terrestrial center, and satellites focus on real-time joint activity detection and channel estimation (JADCE). In this framework, we propose a location-prior-aided JADCE algorithm that incorporates terminal position information into the message-passing process to prune low-probability propagation paths and reduce computational complexity. The algorithm further integrates an expectation-maximization procedure to refine AoA estimates and optimize model hyperparameters. Furthermore, a region-based location management scheme is introduced to enable efficient terminal location estimation and storage. Based on this scheme, region affiliation probabilities are computed by evaluating the expectation of the posterior distribution conditioned on AoA estimates. Simulation results demonstrate significant improvements over benchmark schemes in terms of terminal activity detection and channel estimation accuracy.