Fairness-Aware Beam Routing Automation for Low Earth Orbit Satellite Constellations

Abstract

Large low Earth orbit (LEO) satellite constellations rely on inter-satellite links and distributed ground gateways to deliver broadband services at scale. Designing routing and gateway selection strategies that balance throughput, latency, and user-centric fairness remains challenging due to the size and dynamic connectivity of these networks. This paper studies beamlevel gateway routing in large LEO constellations using a scalable multi-objective evolutionary heuristic and an exact mixed-integer linear programming (MILP) benchmark. We propose an NSGA-II formulation that integrates shortest-path inter-satellite routing, explicit beam-level proportional fairness, and a novel objective minimizing inter-satellite link incidence. A controlled gateway-splitting heuristic is introduced to bridge the gap between unsplittable beam routing and flow-based optimization. Simulation results with real Starlink-based data show that the proposed approach achieves competitive throughput and fairness while remaining computationally tractable at large scales, highlighting key trade-offs between heuristic and exact routing models.