Energy-Efficient Cell-Free Massive MIMO in Ultra-Dense LEO Networks: A Semi-Centralized Clustering Approach

We propose a novel semi-centralized architecture tailored for Ultra-Dense Low Earth Orbit (UD-LEO) satellite communication (SatCom) networks to enhance scalability and energy efficiency across various operational domains. By employing a clustering methodology that assigns each user to a single cluster, we localize critical tasks for UD-LEO SatCom systems such as synchronization, power allocation, and resource management. This approach simplifies the computational demands and reduces backhaul rates while presenting an effective strategy for user scheduling. It tackles synchronization challenges in high-dynamics SatCom environments and reduces reliance on extensive Inter-Satellite Link (ISL) communications. Utilizing a consensus synchronization algorithm within clusters, the proposed scheme can maintain network coherence under fluctuating conditions. Our analysis accounts for imperfect Channel State Information (CSI) at the satellites and statistical channel knowledge at the ground users, as well as unresolved synchronization discrepancies. Analyzed and tested with a Time Division Duplexing (TDD) framework under Rician fading, our approach demonstrates significant potential for improving energy efficiency and adaptability for future SatCom systems.