In-Orbit Collaborative Computing Method for Real-Time Tasks in Low Earth Orbit Satellite Networks

With the rapid evolution of Low Earth Orbit (LEO) satellite Internet, onboard intelligent applications impose increasingly stringent requirements on real-time task processing capability. However, a systematic theoretical basis for determining the minimum collaborative computing resources required to satisfy given reliability constraints is still lacking. To address this issue, this paper proposes a modeling and analytical framework for inorbit collaborative computing of real-time tasks. A dual-layer satellite architecture with functional separation is considered, where one layer is responsible for remote sensing and task triggering, while the other layer provides distributed computing resources for collaborative processing. By integrating a marked Poisson point process, a coupling model between collaborative node scale and spatial beam coverage is established. Based on Amdahl's law, a parallel processing model is then developed to characterize the relationship between collaborative node count and task latency. Closed-form expressions for the minimum required number of collaborative nodes and the corresponding optimal satellite search radius under quality-of-service constraints are derived. Monte Carlo simulations validate the accuracy of the proposed model. The results quantitatively characterize the relationship between task reliability requirements and satellite parameters, providing theoretical support for resource planning and rapid resource bounding in satellite networks.