COEI paradigm enables energy-efficient decentralized computing on LEO satellites

A new research paper introduces Collaborative Orbital Edge Intelligence (COEI), a decentralized paradigm for energy-efficient computing in Low Earth Orbit (LEO) satellite constellations. Authored by Yuvraj Sahni, Jiannong Cao, and Fu Xiao, the paper argues that existing Orbital Edge Computing approaches rely on centralized control without cross-provider satellite collaboration, leading to limited connectivity, higher latency, and accelerated battery depletion. COEI leverages decentralized inter-satellite collaboration to enable energy-efficient on-orbit processing of space data, paving the way for multi-party, multi-orbit megaconstellations.

COEI's architecture addresses networking, computing, and power management challenges across heterogeneous satellite fleets. The researchers validate the paradigm through a case study on decentralized energy-aware task offloading, aiming to maximize task success rate while minimizing the sum of maximum battery depth-of-discharge across all satellites. Results suggest COEI can deliver global resilient connectivity, real-time intelligence, and energy-efficient services for remote and disaster-prone regions. The paper, published on arXiv (2608.00487) and accepted for IEEE Network, outlines future directions including further optimization of collaborative intelligence under real-world constellation constraints.