SkyRipple: Decentralized Routing for Multi-Shell Low-Earth-Orbit Satellite Constellation Networks

Low Earth Orbit (LEO) satellite constellations are rapidly emerging as critical infrastructure for global broadband connectivity. Yet, as multi-shell constellations like Starlink scale to tens of thousands of satellites, routing across dynamic topologies and heterogeneous gateway stations (GSs) presents a significant challenge. This paper introduces SkyRipple, a distributed routing framework designed for GS-augmented multi-shell LEO constellations, addressing both topology synchronization and feeder–inter-satellite bandwidth asymmetry. We first propose a localized flooding mechanism that achieves near-instant topology convergence with minimal signaling overhead. Building on this, SkyRipple combines geographical heuristics with adaptive load-aware metrics to jointly optimize latency and link utilization. Leveraging the multi-antenna architecture and inherent redundancy of GSs, we show that routing efficiency remains largely insensitive to constellation scale. Moreover, SkyRipple effectively mitigates congestion caused by feeder-link bottlenecks and inter-satellite-link failures, significantly enhancing routing stability, scalability, and fairness—paving the way toward resilient, mega-scale communication fabrics in next-generation LEO networks.