Collaborative Orbital Edge Intelligence: A Decentralized Paradigm for Energy-Efficient Computing in Space

Abstract

In recent years, Low Earth Orbit (LEO) satellites have been increasingly deployed to enable connectivity in remote and disaster-prone areas. Researchers have proposed Orbital Edge Computing, which adds computational intelligence to LEO satellites to process data on orbit, providing edge intelligence close to space data sources. Existing work on Orbital Edge Computing typically assumes centralized control without collaboration among satellites from different providers, leading to limited connectivity, higher latency, and increased satellite battery depletion. They have not fully explored decentralized inter-satellite collaboration for energy-efficient intelligence under heterogeneous LEO constellations. This paper introduces a novel paradigm, Collaborative Orbital Edge Intelligence (COEI), that leverages decentralized collaboration among LEO satellites to enable energy-efficient on-orbit processing of space data. We describe the overall system architecture of COEI, including the issues related to networking, computing, and power management. COEI can help create a multi-party, multi-orbit megaconstellation of satellites that delivers better service quality by providing benefits, including global resilient connectivity, real-time intelligence, and energy-efficient services. To demonstrate the COEI benefits, we conduct a case study on decentralized energy-aware satellite task offloading to maximize the task success rate while minimizing the sum of the maximum battery depth-of-discharge across all satellites. Finally, we outline future directions for COEI that offer opportunities for further investigation.