Hierarchical Learning and Computing over Space-Ground Integrated Networks
Jingyang Zhu, Yuanming Shi, Yong Zhou, Chunxiao Jiang, Linling Kuang
TL;DR
The paper addresses the challenge of globally training AI models for remote IoT devices without incurring huge data transfer or privacy risks. It introduces a hierarchical learning framework over space-ground integrated networks (SGINs) that uses LEO mega-constellations for local model collection and GEO satellites for global aggregation, formalized as a global objective $f(x)=\sum_{n=1}^J \lambda_n f_n(x)$. The energy-efficient routing problem for model aggregation is cast as a Directed Steiner Tree (DST) problem on space topology snapshots; a topology-aware routing algorithm (TAEER) combines Dijkstra shortest paths and Chu-Liu-Edmonds minimum spanning arborescence to efficiently connect terminals to a root, with extensions to outages via a weighted objective. Simulations over Walker-Star and Walker-Delta constellations demonstrate significant energy savings and robust performance, with convergence approaching centralized training as communication rounds increase, validating practical applicability for remote, privacy-sensitive learning tasks.
Abstract
Space-ground integrated networks hold great promise for providing global connectivity, particularly in remote areas where large amounts of valuable data are generated by Internet of Things (IoT) devices, but lacking terrestrial communication infrastructure. The massive data is conventionally transferred to the cloud server for centralized artificial intelligence (AI) models training, raising huge communication overhead and privacy concerns. To address this, we propose a hierarchical learning and computing framework, which leverages the lowlatency characteristic of low-earth-orbit (LEO) satellites and the global coverage of geostationary-earth-orbit (GEO) satellites, to provide global aggregation services for locally trained models on ground IoT devices. Due to the time-varying nature of satellite network topology and the energy constraints of LEO satellites, efficiently aggregating the received local models from ground devices on LEO satellites is highly challenging. By leveraging the predictability of inter-satellite connectivity, modeling the space network as a directed graph, we formulate a network energy minimization problem for model aggregation, which turns out to be a Directed Steiner Tree (DST) problem. We propose a topologyaware energy-efficient routing (TAEER) algorithm to solve the DST problem by finding a minimum spanning arborescence on a substitute directed graph. Extensive simulations under realworld space-ground integrated network settings demonstrate that the proposed TAEER algorithm significantly reduces energy consumption and outperforms benchmarks.