Routing in High-Velocity LEO Satellite Constellations

This direction is merged into another direction: dynamic network topology (dynamic topology management). Its last version stays readable and citable here.

Version 1 of 139 papers43 conceptsfrozen 2026-10-04

Traditional satellite communications relied on Geostationary (GEO) satellites—massive, stationary mirrors in the sky. But the shift toward Low Earth Orbit (LEO) mega-constellations like Starlink and Kuiper has fundamentally changed the networking landscape. Unlike GEO satellites, LEO nodes move at speeds exceeding 27,000 km/h, completing an orbit in roughly 90 minutes.

In this environment, the network topology is not a fixed grid but a swirling, high-velocity dance. Links between satellites and ground stations appear and disappear in windows of only a few minutes, and the relative positions of neighboring satellites shift every few seconds. This extreme mobility renders traditional terrestrial routing protocols—which assume a stable physical layer—ineffective.

The Failure of Static Snapshots

The most intuitive way to handle a moving network is to take a "snapshot" of the topology, calculate the best paths, and update the routing tables. However, as Hao Jin and Yuan Liao argue in Towards Routing and Edge Computing in Satellite-Terrestrial Networks, discretizing time into static snapshots often fails to capture the dynamics that occur during the routing process itself. By the time a path is calculated and a packet is sent, the physical link may have already moved or severed.

To solve this, researchers are moving toward temporal partitioning. Jiahe Huang and Yifei Zhu describe a system in SpaceMeta that uses time-division slots to schedule handovers. Instead of reacting to a broken link, the network uses the deterministic nature of orbital mechanics to predict exactly when a handover must occur, allowing for seamless transitions between orbiting servers.

Decentralized Intelligence and Learning

Because LEO constellations can involve thousands of satellites, a centralized controller (like those used in traditional Software-Defined Networking) becomes a bottleneck. The propagation delay between a satellite and a ground-based controller is often longer than the lifespan of the network state the controller is trying to manage.

The emerging solution is to push intelligence to the "orbital edge." Wanja de Sombre and Arash Asadi introduced SKYLINK, a decentralized link management system detailed in Scalable and Resilient Link Management in LEO Satellite Network. SKYLINK uses a Multi-Armed Bandit (MAB) learning framework, allowing each satellite to autonomously learn which neighbors provide the most reliable paths based on local context. This eliminates the need for global topology updates and allows the network to adapt to failures in real-time.

Similarly, Yong Lü and colleagues proposed Distributed Traffic Balancing Routing (DTBR) in Distributed Traffic Balancing Routing for LEO Satellite Networks. DTBR combines a virtual topology model with geographically-aware cost factors, enabling satellites to balance traffic loads locally without waiting for instructions from the ground.

Geometric and Stochastic Abstractions

Calculating the shortest path in a 10,000-node constellation is computationally expensive. To speed this up, researchers are using geometric abstractions to simplify the search space. Menglan Hu and Chenxin Wang developed Landmark-based Geometric Segment Routing (LGSR), which abstracts the massive constellation into "skeleton graphs." As described in A Lightweight and Scalable Design of Segment Routing in Broadband LEO Constellations, this approach achieves a 60x speedup in route calculation by focusing on key "landmark" satellites rather than every individual node.

For even more radical scaling, some researchers abandon the idea of a fixed graph entirely. In Ultra Reliable Low Latency Routing in LEO Satellite Constellations, Ruibo Wang and Mohamed-Slim Alouini apply Stochastic Geometry (SG). By treating the constellation as a random field of points rather than a rigid grid, they can derive analytical bounds for reliability and latency, creating routing strategies that remain robust even as the specific "points" in the field move at high velocities.

Bridging the Terrestrial Gap

The ultimate goal of these advancements is to make the satellite hop invisible to the end-user. Current "terrestrial-first" protocols like TCP or standard RTC (Real-Time Communication) tools struggle with the jitter and frequent handoffs of LEO. T3P addresses this by using transport-layer tuning to handle the performance "tails" created by orbital dynamics.

By combining predictive orbital models, decentralized machine learning, and geometric simplifications, researchers are transforming LEO constellations from a collection of fast-moving targets into a stable, high-speed backbone for global 6G connectivity. As Chang Liu demonstrates with the GT-SAC framework in Joint Cache Placement and Routing, treating the moving constellation as a dynamic graph allows us to not just route data, but to intelligently place content and computation exactly where it needs to be, even when the "where" is moving at seven kilometers per second.

Papers in this version

History & citationVersion 1 · 2026-10-04

Merged into another direction · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_routing_in_high_velocity_leo_satellite_c_8a7091f2,
  title = {Routing in High-Velocity LEO Satellite Constellations},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/routing-in-high-velocity-leo-satellite-constellations/rdr_8a7091f216ab47d59b4a5ffd46ef78a3},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_routing_in_high_velocity_leo_satellite_c_8a7091f2_v1,
  title = {Routing in High-Velocity LEO Satellite Constellations},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/routing-in-high-velocity-leo-satellite-constellations/rdr_8a7091f216ab47d59b4a5ffd46ef78a3/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}