Adaptive Predictive Routing for Dense LEO Constellations

Abstract

The rapid deployment of Low Earth Orbit (LEO) mega-constellations introduces challenges such as dynamic topology changes, congestion, and routing instability. Continuous orbital motion causes frequent inter-satellite link variations, reducing the effectiveness of traditional reactive routing protocols. This paper proposes an Adaptive Predictive Routing (APR) framework that combines orbital-aware predictive routing with real-time adaptive rerouting to improve latency, route stability, and scalability in dense LEO networks. Compared with predictive-only routing, the proposed APR framework reduces average end-to-end delay by approximately 18% and increases Packet Delivery Ratio (PDR) by about 7% under dynamic traffic conditions.