Inter-Satellite Routing Algorithms for Dual-Layer Low Earth Orbit Satellite Internet

In recent years, low-cost low Earth orbit (LEO) satellites provide shorter-latency and higher-throughput services when compared to the medium Earth orbit (MEO) and geostationary Earth orbit (GEO) satellites. However, existing research on multi-layer satellite constellations and inter-layer routing typically focuses on larger-span constellations and routing among LEO/MEO/GEO. Research on the multi-layer LEO satellite networks is currently in its early stages. Existing routing algorithms designed so far have not fully leveraged the advantages of cross-layer links. Therefore, this paper proposes a method for establishing inter-layer links and rules for a dual-layer LEO satellite network, along with a routing algorithm for such a dual-layer constellation. Firstly, a uniform distribution duallayer constellation configuration for LEO satellites is constructed, selecting the cross-layer linking satellites in LEOs. Secondly, to meet the requirement for long-term stable connections, a method for rapidly switching inter-layer links based on the periodic relative motion of satellites between layers is proposed, ensuring continuous and reliable transmission of inter-layer links within the satellite network. Furthermore, based on the rules for establishing inter-layer links and the dynamic relationships within the satellite network, time slot information for the dual-layer satellite network is established. Utilizing the Dijkstra’s algorithm, the shortest transmission path from source to destination nodes in the dual-layer satellite network model is computed to determine the effective routing within the satellite network. Finally, a dual-layer constellation simulation model is built based on the parameters of the Kuiper constellation, and simulation experiments are conducted. Simulation results demonstrate that the dual-layer constellation routing algorithm (DCRA) reduces the round trip time (RTT) by 15.3%. At distance of 14,000 km between the source and the destination nodes, the average RTT is reduced by approximately 22.28% compared to the single-layer routing algorithm (SLRA).