From Emerging LEO Satellite Constellations to the Space Cloud: Emulation Platforms and Orchestration Methods

In the rapidly advancing field of satellite communications, mega-constellations of Low Earth Orbit (LEO) satellites are gaining significant attention from the academic and industrial sectors. Managing these expanding constellations has become increasingly complex, and integrating them with classical cellular networks presents new automation challenges. We envision a Space Cloud in which Multiaccess Edge Computing (MEC) services are deployed within cross-liked space networks to address emerging Non-Terrestrial Networks (NTNs) latency demands. Integrating computation services in orbit will be instrumental in unlocking a Space Cloud that reduces the need to route computation requests to the Internet backbone. This study’s first contribution is MeteorNet, an open-source constellation and edge computing emulation platform aimed at assessing the expected performance of future Space Clouds. MeteorNet realistically replicates the behavior of edge computing in a synthetic satellite constellation network hosting onboard containerized servers. The second contribution comprises two innovative edge orchestration strategies based on fuzzy logic and reinforcement learning. These strategies leverage historical data on task loads and processing failures to control the activation of on-orbit edge servers, ensuring efficient resource utilization. A Pareto-efficient analysis of multiple Key Performance Indicators (KPIs) using MeteorNet proves the approach’s feasibility in space missions with energy constraints and limited computation resources.