AST × Starlink: two ways to reach the same phone

Two ways to hear the same weak phone

Put the same phone under two satellites. It has the same small antenna and the same battery. AST and Starlink must make the space end of the connection do the extra work. The first visible difference is how much antenna they put there; the next is where they put the cellular base station.⁴¹³

Both use phased arrays. AST’s first five BlueBirds unfold a much larger radio surface than the early Starlink Direct-to-Cell array SpaceX documented in January 2024. In the network designs examined here, AST’s gateway keeps the baseband units on Earth, while Starlink puts the LTE base station in orbit. Those choices change how each system collects the signal, corrects it and hands it to the next satellite.^1613311

Follow one connection through those decisions: hear the phone, keep the beam useful, interpret its signal and maintain the call.

How different are the antennas?

Start with the radio surface facing Earth. SpaceX specified 2.7 × 2.3 metres for its early V2mini Direct-to-Cell phased array: a rectangular footprint of about 6.2 m². AST gives 693 square feet, or about 64.4 m², for BlueBird 1–5. That is roughly ten times the stated physical array area. These are company disclosures for two 2024-era designs; this calculation compares their geometry.¹³¹⁶

A phased array combines signals from many elements, with controlled relative phase so they reinforce in a chosen direction. At the same wavelength, a larger effective aperture can provide more gain and a narrower beam. For the same received power density, a larger effective receiving area collects more signal. That gives AST’s larger surface a concrete potential benefit: more margin for hearing the weak phone and more spatial selectivity.¹⁷

The physical outline alone does not give us the effective aperture, receiver noise, operating band or beam pattern. A tenfold area ratio therefore cannot be carried over as a tenfold speed result. SpaceX’s disclosed response to the weak-phone problem combines its smaller array with custom radio silicon, sensitive receivers and high-powered transmitters.¹³¹⁷

The size comparison must also keep moving with the hardware. AST describes nearly 2,400 sq ft arrays for its next generation; SpaceX now describes a proposed Starlink Mobile V2 with thousands of spatial beams and much higher throughput. The January 2024 dimensions are a dated reference point, not a specification for that newer Starlink generation.⁷¹⁸

What does AST pay for the larger surface?

A much larger surface has to fit inside a launcher, unfold and preserve the relationships between its radiating elements. AST’s deployment patent shows antenna assemblies packed into a stack, with connectors releasing stored energy to open them. Its oscillation patent follows the next problem: a light, extended structure can flex, changing the signal path between its elements and the phone.²¹

This is where the aperture advantage becomes a control problem. AST’s illustrated radiation patterns compare the intended surface, the deformed surface and a compensated case. The issued claim requires a sequence: apply mechanical force, measure the displacement left over, then correct phase. The marked amendment matters because it makes physical correction part of the claimed answer; the electronics are not asked to compensate for an arbitrarily distorted array.¹⁵⁶

SpaceX’s smaller early array was designed around the V2mini platform. AST’s selected patents expose the extra deployment and shape-control work involved in its large modular aperture. That is the supported contrast. We have no equivalent Starlink structural-compensation record in this comparison, so we cannot conclude that its array is perfectly rigid or needs no calibration.¹³²¹

The operational question is how much useful gain and beam shape each array retains while it scans and flexes. AST’s record makes that constraint unusually visible.¹

Same radio job, different location

Follow the phone’s signal past the antenna. In the AST gateway drawing, baseband units on the ground process the cell signals, with satellite links relaying them between gateway and handset. In Starlink’s early network drawing, the LTE base station—the eNodeB—is aboard the satellite, and onward traffic travels through the Starlink network to the carrier. Both deliver cellular service; the base-station processing sits on opposite sides of the feeder link.³⁴¹⁴¹¹

The practical difference is what must remain coordinated. AST’s disclosed gateway exchanges radio-signal data and control information associated with particular cells and satellite paths. Its shared buffers keep those associations together as the route changes. Starlink’s depicted satellite performs the cellular base-station work before carrying the traffic onward. Ground-based coordination is central to the AST drawing; working base-station hardware in orbit is central to the Starlink drawing.³¹⁴¹¹

That suggests different maintenance and transport tradeoffs. Ground baseband equipment can be physically serviced on Earth, but the radio path to it includes the moving spacecraft and feeder link. Putting the base station in orbit puts more of that processing at the receiving satellite, but also puts that hardware beyond routine hands-on repair. These are consequences of the depicted placement, not measured cost or reliability results.³¹¹

AST’s specification also permits some operations at a control satellite. The distinction is the location of the baseband units in these particular architectures; neither network puts every operation in just one place.³

Two phones in one beam still need different corrections

A satellite can correct its signal for the center of a ground cell. The trouble is that your phone may be near the edge. Its path to the moving spacecraft differs from the reference path, leaving an error after that shared correction. Both AST and Starlink address this same geometry.⁴⁹

AST’s handover disclosure has the gateway normalize delay and Doppler to the cell center. It places the center phone near the midpoint of the timing-advance range and describes keeping the edge phone’s remaining timing and frequency error within the range its radio can handle. Starlink’s Doppler patent also begins with a cell reference location, but then explicitly estimates coarse and fine portions of the remaining Doppler shift for an individual phone and removes them at the satellite receiver.⁴⁹

The useful contrast is what happens after the common reference correction. AST’s cited passage describes engineering the cell and compensation so the residual stays tolerable. Starlink’s cited claim specifies an additional per-phone residual estimator. This is more precise than saying one system “solves Doppler” and the other does not. The AST passage is not a complete inventory of every receiver correction it may use.⁴⁹

The first request to join the network exposes a related difference in the records. AST describes adapting the base-station procedure to the longer path, including access-response timing. Starlink’s access patent goes inside the detector: it uses a coarse arrival-time estimate to choose among candidate timing/frequency pairs from the phone’s preamble. Procedure adaptation and signal detection are two layers of the same access problem; neither description proves that the other network lacks that layer.⁴⁸

How do you change satellites without moving the subscriber?

A stationary subscriber creates two different jobs for a moving constellation: keep an active call on a usable radio path, and retain a sensible address for finding the phone later. Both records try to preserve continuity on the ground while changing the machinery overhead.⁴¹⁰

AST’s handover example gives the overlapping cell two satellite paths with different physical cell identities. The gateway arranges the transfer from the setting satellite to the rising one, while the core retains the tracking-area or location-area code. The radio identity changes, but the subscriber does not have to acquire a new geographic identity merely because the satellite changed.⁴¹⁵

Starlink’s mapping patent makes that separation explicit through a different mechanism: a beam identifier plus a time maps to a ground-area identifier. Its tracking-area notes explain how a stationary phone can avoid repeated location updates as the beams pass. AST’s example preserves the tracking code across a coordinated radio-path transfer; Starlink’s mapping supplies a time-dependent translation between moving beams and ground locations.^1012194

That comparison also marks a boundary. AST’s described two-path procedure tells us how active users are transferred. Starlink’s geographic map tells us how location identifiers are maintained; it is not, by itself, the complete active-call handover procedure. The common test is whether a stationary user can keep a call and remain reachable without unnecessary signaling as satellites pass.⁴¹⁰

Does the bigger antenna mean the faster service?

More effective aperture can improve the link under matched conditions. Service capacity also depends on spectrum, power, interference, receiver performance, active beams and the users sharing them.¹⁷⁴⁹

AST’s next-generation page describes a nearly 2,400 sq ft array, an AST5000 processor and a target above 150 Mbps per coverage cell. SpaceX’s newer Mobile V2 page describes thousands of beams and around twenty times the throughput capability of its first generation. These are company targets with different denominators. Neither statement is a matched measurement of an individual subscriber’s rate.⁷¹⁸

For a revealing comparison, put the same handset under comparable spectrum, elevation and load, then measure both its uplink and downlink. Repeat while it moves toward a beam edge and while the serving satellite changes. That would test the antenna margin, the correction strategy and the handover together. Peak rate alone would miss the very places where the two architectures differ.⁴⁹

The current evidence supports a comparison of those engineering choices. A performance winner still requires the same task under the same conditions.

The tradeoff follows the signal

The clearest physical distinction is aperture: AST’s first BlueBirds expose far more antenna area than the early Direct-to-Cell array described by SpaceX. The clearest network distinction in the selected drawings is baseband placement: AST coordinates it from the gateway, while Starlink carries the base station in orbit. One comparison concerns collecting the signal; the other concerns where that signal becomes cellular traffic.^1613311

Those choices lead to different questions. Can AST preserve useful gain and beam shape across its large, flexible surface while coordinating the changing feeder paths? Can Starlink’s receiver and orbital base station make its disclosed aperture useful at the difficult edges of coverage and through transfers? Both must still handle shape, timing, frequency, interference and continuity; the records give us different levels of detail for each.¹⁴⁹¹⁰

That is the comparison worth following as both generations advance: how effectively each turns its antenna, processing and moving radio paths into a dependable connection for the same ordinary phone.