Abstract:
With the rapid development of satellite constellations, space-based edge computing, and terrestrial computing networks, Terrestrial-Space Interconnected Collaborative Computing (TSICC) has attracted increasing research attention in the fields of space computing and space artificial intelligence. Computing resources across terrestrial and space domains exhibit distinctive characteristics. Conventional terrestrial-centric computing architectures may face challenges in efficiently supporting global-scale applications involving Earth observation and the intelligent processing of massive spatial data. This paper proposes a working definition and establishes explicit conceptual boundaries for Terrestrial-Space Interconnected Collaborative Computing (TSICC), based on a structured comparison with closely related paradigms including satellite edge computing, satellite-terrestrial integrated computing, space-air-ground integrated networks, space data centers, satellite computing networks, and cloud-edge-end collaboration. It then presents a five-layer TSICC architecture and three representative operation paradigms, and analyzes key scientific challenges for the development of TSICC from the perspectives of scheduling algorithms, network protocols, operating systems, and AI models. In addition, representative application scenarios, including emergency communication, environmental monitoring, remote sensing, and global information services, are discussed. Finally, several future research directions are outlined to support the interconnection and collaborative utilization of terrestrial and space computing resources, the development of terrestrial-space collaborative intelligence, and related standardization efforts. |