Ground Station Handoff

Why LEO needs constant handoffs

A LEO satellite at 550 km altitude orbits Earth in about 95 minutes — traveling ~7.5 km/s relative to the ground. From a user's perspective, a single satellite rises above the horizon, transits overhead, and sets — typically 5-10 minutes from rise to set depending on pass elevation. To stay connected, the user terminal must:

1. Identify the next satellite that will come into view.
2. Predict the overlap window when both old and new satellites are simultaneously available.
3. Coordinate with the network to plan the handoff.
4. Switch the active link to the new satellite within the overlap window.

For a Starlink user, this happens every few minutes, 24/7.

The overlap window

The handoff happens during a brief period when both the old and new satellites are above the horizon. Constellation operators design orbital geometry to ensure overlap exists at every transition — sufficient density that as one satellite is setting, another is already up and high enough to be useful.

For Starlink-density constellations (thousands of satellites in multiple shells), at any moment there are typically 4-10 satellites visible from any populated location, even though only one is the active link.

The handoff sequence

1. Prediction. The terminal uses orbital ephemeris (provided by the constellation operator) to compute when the current satellite will become unsuitable and when the next-best candidate will be ready.
2. Coordination. The terminal and the network exchange messages: "I plan to hand off to satellite X at time T."
3. Parallel acquisition. Before the cutover, the terminal acquires the new satellite (laser-link or radio acquisition; phased array can do this without losing the old one).
4. State transfer. Session state, queued packets, sequence numbers — whatever the network's stateful machinery needs — moves from the old satellite's path to the new.
5. Cutover. Traffic shifts to the new satellite. Old satellite's link is released.
6. Verification. The terminal confirms the new satellite is working before the old becomes inaccessible.

Phased arrays enable seamless handoff

A mechanical dish has to physically rotate to track each satellite, taking seconds. By the time it's pointed at the new one, traffic to the old one has been dropping packets. A phased array switches beams electronically in microseconds — multiple beams can be active simultaneously during the handoff window.

This is why mass-market LEO terminals use phased arrays. See phased array antennas.

Network-side state

The user's IP, TCP connections, application state — all of these need to survive the handoff. Approaches:

• Anchored IP. The user has an IP anchored at a ground gateway; satellite changes don't affect the IP. Traffic always tunnels back to the gateway.
• Mobile IP-style. The user's IP can move with them; routing updates propagate to ensure packets reach the current satellite.
• Application-layer survival. TCP connections survive because the connection identifier (IP + port) doesn't change; the path beneath changes, briefly causing packet drops handled by TCP retransmit.

Different constellations use different combinations.

What users perceive

When handoffs work, users see nothing — continuous bandwidth, stable connections. When something goes wrong:

• Brief stalls of a few hundred milliseconds.
• Periodic latency spikes during transitions.
• Rare connection drops that recover quickly.

Compared to traditional GEO satellite (where pointing is fixed and there's no handoff), LEO services have more frequent transitions but each one is smaller. The average user experience is a stable connection that occasionally has microsecond-to-millisecond hiccups during handoffs.

Coverage gaps and handoff failures

In areas with sparse satellite coverage (early in a constellation's deployment, or in polar regions for non-polar-coverage constellations), the next-best satellite may not be available when the current one sets. The user terminal experiences a coverage gap until the next pass.

Operators publish coverage maps that account for this. Marketing materials often show "99.x% availability" — the small fraction is handoff-related stalls and coverage gaps.

Inter-satellite link interaction

For constellations with inter-satellite laser links (see ISLs), the handoff can be even cleaner:

• The new satellite's link to the broader network is already established via the constellation.
• The user terminal's only switch is from one nearby satellite to another.
• The constellation handles the back-end routing changes transparently.

Without ISLs, every handoff also involves swapping which ground station is in the path — more state movement.

Doppler shift compensation

LEO satellites move so fast that the Doppler shift on their radio signals is significant — a satellite approaching shifts the carrier frequency up by several kHz; a satellite receding shifts it down. The terminal and satellite both compensate by adjusting their reference frequencies based on the calculated relative velocity.

During handoff, the Doppler shift to the new satellite is different from the old. The terminal pre-computes and applies the correction so the new link can acquire quickly.

Why GEO doesn't have this problem

A geostationary satellite stays fixed over one spot on Earth (in geosynchronous orbit at the equator). The user's dish points at one direction and never re-aims. No handoffs.

The tradeoff is the round-trip distance (~72,000 km for the path up and back), which yields ~600 ms of inherent latency — fine for TV broadcast, painful for interactive use. LEO trades handoff complexity for dramatically lower latency. See satellite vs fiber latency.

Frequently Asked Questions

Why do LEO satellites need handoffs?

Because LEO satellites move across the sky in minutes. A single satellite is in view of any given user for typically 5-10 minutes per pass. To maintain continuous service, the user terminal must seamlessly transition from one satellite to the next as the first sets and a new one rises. The handoff is the choreographed process that swaps active satellites without interrupting the user's session.

How is handoff different from cellular handoff?

In cellular, the user is mostly stationary and the tower is fixed; handoffs happen when the user moves. In LEO satellite, the user is stationary but the satellite moves — handoffs happen constantly even for users who never move. The handoff frequency is much higher (every few minutes vs occasional cellular handoffs), and the satellites' fast motion means precise timing is essential.

What happens during a handoff?

The terminal predicts the next satellite based on orbital ephemeris, coordinates with the network on the handoff timing, briefly establishes parallel connections to both satellites (overlap), transfers session state, then drops the old satellite. Users see no interruption if everything works; in failure modes they might see a brief stall.

What is satellite diversity?

Using multiple satellites simultaneously — typically the strongest and second-strongest available. The user terminal can split traffic across both for higher throughput, or use one as a hot standby for handoff redundancy. Multi-beam phased arrays enable this without additional hardware. Some constellations route different types of traffic over different satellites to optimize latency or throughput.

What causes handoff failures?

Misalignment with predicted ephemeris (satellite slightly off its orbit), atmospheric conditions interfering with the new satellite's signal, terminal pointing issues, network-side coordination delays, or simply running out of next-satellite candidates in coverage gaps. Modern constellations are designed with overlap margins to handle small failures; large failures show up as brief drops in service.