A conventional home internet connection has the luxury of standing still. The cable entering the building stays connected to the same infrastructure and never moves. For Starlink, there’s an entirely different challenge, as your internet infrastructure races across the sky.
Low-Earth orbit satellites travel at roughly orbital velocity (27,600 km/h or 17,150 mph) and remain visible from any one location for only a short period. A Starlink terminal therefore cannot establish one satellite connection and keep it indefinitely. Its traffic must repeatedly move to another satellite, another beam, or another gateway without interrupting a call, stream, or game.
SpaceX Patent US 11,949,496 B1, titled “Low Latency Schedule-Driven Handovers,” describes how the network can prepare those transitions before the old link disappears.
The Moving-Network Problem
A Starlink connection is more than a radio link between the dish and a spacecraft. Traffic can pass from the user terminal to a satellite, from that satellite to a ground gateway or another satellite, and then through a point of presence to the wider internet.
Every part of that route can change. A satellite moves beyond the terminal’s view. A beam stops covering the user’s cell. A gateway becomes unavailable. Network congestion making another route preferable. A roaming Starlink terminal adds another moving component to an already dynamic system.
If the system waited for a connection to fail before finding a replacement, users would experience packet loss and noticeable pauses. SpaceX’s patent instead treats the motion as something the network can predict.
Scheduling the Handoff
At the centre of the patent is a route distribution service. It builds and distributes communication schedules using satellite ephemeris data. Ephemeris data is the predicted positions and velocities of spacecraft, including terminal and gateway locations, asset availability, network conditions, traffic load, and quality-of-service requirements.
The schedule can identify which terminal should communicate with which satellite, which satellite should use which gateway, and the precise radio frames or time slots in which those links should operate. It can also identify an upcoming handover period in advance.
That turns a handover from an emergency reaction into a planned network event.
The patent covers several kinds of transition: one satellite to another, one beam to another on the same satellite, one ground gateway to another, and combined handovers where multiple portions of the route change together.
Sending Packets Two Ways
Planning alone does not guarantee that every packet survives the switch. SpaceX also describes temporarily “bicasting” packets over both the existing and replacement links before the handover is complete.
For a brief overlap, the old route continues carrying traffic while the new route is brought online. Once the transition succeeds, the network stops using the original link and continues through the replacement. This resembles changing trains by placing both alongside the platform before asking passengers to step across, rather than waiting for the first train to leave.
The network may also buffer, reorder or deduplicate packets around the transition. That matters because two paths through a satellite network may not deliver identical packets at precisely the same time.
The Magic of Starlink
Starlink’s low altitude is a major reason it can offer lower latency than traditional geostationary satellite internet. The tradeoff is that its satellites constantly move relative to users and ground infrastructure. SpaceX’s handover system is not just a minor optimization. It is one of the mechanisms that makes a low-orbit consumer network feel like an ordinary, persistent internet connection.
It also reveals that Starlink is centrally choreographed in ways users never see. The terminal does not simply point toward the strongest object overhead and hope for the best like with traditional satellite internet. The network can look ahead, account for spacecraft motion and demand, distribute a schedule, and prepare more than one part of the route for a coordinated change.
That choreography will become even more important as Starlink serves aircraft, ships, vehicles and direct-to-cell devices. In all of those cases, the user may move while the satellites move too. The cleverest part of Starlink may not be maintaining a radio link with space. It may be making thousands of planned breaks in that link effectively invisible.
