In the competitive landscape of space-based internet, the geometry of a constellation is just as important as the rockets that launch them. During the AIORI-2 Hackathon, team OrbitOps from K.S. School of Engineering and Management (KSSEM) explored how RFC 9717—the routing architecture for satellite networks—can be used to shave milliseconds off global communication paths.
By comparing real-world OneWeb TLE (Two-Line Element) data against optimized Walker Constellation models, the team identified the “sweet spots” for orbital inclination and plane counts to connect Bangalore to Rio de Janeiro with minimal delay.
1. The Science of the “Space Mesh”
The traditional internet relies on fixed fiber optics. LEO (Low Earth Orbit) networks, however, use Inter-Satellite Links (ISLs) to create a dynamic, moving mesh in the sky. To model this, the team utilized a graph-theory approach where satellites act as nodes and ISLs act as edges.
The simulation focused on the Walker Delta Constellation, a mathematical model used to ensure uniform global coverage. By varying the number of orbital planes and the inclination of those planes, the team could visualize how “connected” the sky actually is.
2. Technical Implementation: From TLE to Latency Graphs
The team built a custom Python-based simulation stack to bridge the gap between orbital mechanics and network routing:
- Orbit Propagation: Used the sgp4 library to transform TLE data into precise XYZ coordinates.
- Pathfinding: Leveraged NetworkX to implement Dijkstra’s algorithm, finding the absolute shortest path across the satellite mesh between Bangalore and Rio de Janeiro.
- Topology Sweeps: Iteratively adjusted the constellation’s inclination and density to find the point of diminishing returns.
3. Key Findings & Performance Metrics
The results proved that “more satellites” doesn’t always mean “lower latency.” There is a critical threshold where adding more orbital planes actually increases the path complexity without reducing the physical distance traveled.
| Scenario | Metric | Result | Insight |
|---|---|---|---|
| OneWeb (Real Data) | Path Latency | 72.51 ms | Real-world baseline for existing infrastructure. |
| Walker (Optimized) | Path Latency | 66.77 ms | A ~8% improvement via topology tuning. |
| Inclination Peak | Optimal Angle | 89°–95° | Near-polar orbits provide the best cross-hemisphere routing. |
| Plane Count | Density Curve | Non-linear | Latency drops as planes are added, then plateaus. |
4. Implementation Challenges
- TLE Data Gaps: Real-world satellite data often contains “stale” elements. The team had to implement error-handling to ensure the sgp4 propagator didn’t result in “broken” links.
- Dynamic Connectivity: In a Walker constellation, cross-plane ISLs are constantly breaking and reforming as satellites pass over the poles. Maintaining a stable graph under these conditions required strict distance-based thresholding.
5. Lessons and Future Horizons
The project successfully validated the principles of RFC 9717, demonstrating that routing in space is as much a matter of geometry as it is of logic.
“Visualizing different constellation designs and seeing how topology directly impacts latency was a highlight of the project. It turns out the ‘sky-path’ is much more complex than a straight line.” — Thanushree N, Team Member
The team’s future work involves moving from static snapshots to time-varying simulations, where the shortest path is recalculated every second as the satellites move at 7.5 km/s.
Read the full report
