Architecture and Method for On-Board Heterogeneous Intelligent Computing Based on High-Speed Interconnection

The ultra-high speed and massive access requirements of native intelligence 6G have placed extremely high demands on the computing and storage capabilities of space-based cloud infrastructure. However, due to constrained onboard resources and high cosmic radiation, the reliability of space-based cloud systems faces significant challenges.To address this issues, this paper first analyzed the reliability of onboard intelligent computing within a resource-constrained onboard environment. Next, a high-speed, interconnected heterogeneous onboard intelligent computing architecture was proposed, and a multi-modehigh-reliability onboard computing method was designed based on this architecture. Finally, the reliability of the proposed computing architecture and scheme was validated through theoretical analysis.