Beyond the Paper Orbit: Why Most LEO Routing Research Fails the Reality Test

In 2026, the success of a satellite constellation depends on how it manages its “logical brain.” While academia and industry often seem to disagree on clustering, the reality is that the method’s value depends entirely on its architecture.

The following guide breaks down the state of LEO routing, distinguishing between the failed theories of the past and the high-speed realities of today.

In a network of 10,000+ satellites moving at 7.5 km/s, routing is a battle against computational exhaustion. If every satellite spent its battery power calculating global paths, the constellation would go dark in weeks.

1. The Clustering Paradox: “Good” vs. “Bad”

Clustering is currently the most debated topic in satellite research. To understand why it’s both an “Industrial Standard” and a “Research Trap,” we must look at how it’s controlled.

❌ The “Bad”: Decentralized Ad-Hoc Clustering

This model, borrowed from terrestrial sensor networks, fails in space due to the Energy Hole Problem.

• Unpredictable Elections: Satellites “vote” for a Cluster Head (CH). In a high-speed LEO environment, these clusters break and re-form so often that the “election” messages consume more bandwidth than actual data.
• The “Head” Burnout: The elected CH must handle 10x the processing load. Without perfect coordination, the CH’s battery dies, creating a dead zone in the constellation.
• Instability: If a CH fails or moves behind the Earth’s horizon, the entire sub-network loses its connection during the “re-election” phase.

✅ The “Good”: SDN-Managed Hierarchical Clustering

This is the 2026 Industrial Standard used by major players like Starlink and Project Kuiper.

• The “Ground Brain”: A Software-Defined Networking (SDN) controller on Earth uses its unlimited power to pre-calculate cluster roles based on the predictable orbital clock.
• Deterministic Role Rotation: The ground station tells a satellite exactly when it will become a leader and for how long. This ensures that no single node “burns out,” distributing the energy load perfectly across the shell.
• Scalability: Managing 120 “clusters” is mathematically easier than managing 6,000 individual satellites.

---

2. Borrowing from Terrestrial: MANET & VANET

Since satellites are essentially “vehicles” in the sky, we can adapt protocols from Mobile (MANET) and Vehicular (VANET) Ad-Hoc Networks.

🏆 The “Practical” Hall of Fame

• GPSR+ (Geographic): A VANET favorite. It is stateless, meaning satellites don’t need a map. They simply pass the packet to the neighbor physically closest to the destination’s coordinates. It’s the most resilient for military or emergency use.
• G-AODV (Adaptive): A MANET protocol that only finds a path when you have data to send. The “G” (Global) version uses orbital prediction to ensure the chosen path won’t break 2 seconds later.
• Virtual Node (VN): This creates a static grid over the Earth. Even as physical satellites fly by at 27,000 km/h, the “routing address” stays fixed over a city, preventing IP address resets.

⚠️ The “Inefficient” Hall of Shame

• Standard OSPF: Terrestrial protocols that “flood” the network with updates. In LEO, this creates a Broadcast Storm that consumes 100\% of the bandwidth just for background chatter.
• Pure DSR: Storing the entire 15-hop path in the packet header. In space, this “overhead” leaves no room for the actual message.

---

3. Summary of Routing Performance (2026)

---

Why 2026 is Different

In 2026, the “Seam”—where satellites moving North meet those moving South—has been solved by Hybrid Routing. Modern satellites use Clustering for 90% of the flight but switch to Geographic “Greedy” Forwarding the moment they hit the chaotic Seam area.  This provides the order of a hierarchy with the survival instincts of a lone wolf.

Modern satellite networks uses Clustering. This doesn’t mean that the paperware clustering-based routing algorithm, such as  LEACH will work on LEO, in fact, they will never work in any practical sensor network itself.