Adaptive Inter-Satellite Power Transfer for Sustainable Space Computing

Abstract

With the advent of the 6G era, Low Earth Orbit (LEO) satellite networks deliver low latency, high throughput in space edge computing services, yet they face severe energy constraints and battery aging. This challenge worsens when periodic solar power harvesting fails to match fluctuating task demand, driving deep discharge during eclipse and leaving energy reserves insufficient at dawn. Existing data-finds-the-energy schemes offload tasks to energy-rich nodes, at the cost of extra transmission energy and resource underutilization. To tackle this dilemma, we derive the underlying physical models to analyze the feasibility and potential of inter-orbital wireless power transfer, and introduce Prometheus, the first framework to reverse the traditional model by enabling energy-finds-the-data through power beaming from Medium Earth Orbit (MEO) satellites. Within our Prometheus, we formulate the MEO-to-LEO energy provisioning problem, and propose the Sustainable and Task-Aware Energy Transfer (STA-ET) approach, including an energy-sensitive zone partition algorithm to reduce problem scale and a task-aware energy routing algorithm to determine the energy transfer strategy from MEO to LEO satellites. We then prove the convergence for the partition algorithm and the zone-wise optimality of the energy routing algorithm. Finally, we implement Prometheus and evaluate the STA-ET approach based on real-world traces. Extensive experiments show that STA-ET significantly extends battery life and improves resource utilization compared to state-of-the-art schemes.