Small, flat, rooftop satellite antennas could replace big, bulky, costly dishes

I asked ChatGPT to explain following to me:

From https://newatlas.com/telecommunications/flat-rooftop-satellite-antennas/

“Ground Control to Major Tom,” the evergreen lyrics to David Bowie’s 70s hit, Space Oddity. What’s odd is that, despite huge advances in satellites themselves, much of the physical infrastructure connecting those spacecraft to Earth still relies on large mechanically steered dishes, a model increasingly strained by the rise of massive low-Earth-orbit constellations.

Engineers at the University of California, San Diego may have developed a different way to connect satellites to Earth, replacing large mechanical dishes with networks of smaller, flat antennas distributed across rooftops, telecom towers, and other buildings. Their system, called ArrayLink, could dramatically increase satellite data capacity while making ground stations cheaper, easier to deploy, and far more scalable.

Satellite communication has quietly become one of the most critical infrastructures of modern civilization. Far more critical than you probably think. Beyond satellite internet, these systems underpin GPS navigation, financial transactions, weather forecasting, military communications, emergency response, aviation, shipping, remote healthcare, and Earth observation.

Over the past decade alone, the number of active satellites in orbit has exploded from a few thousand to many thousands more, with tens of thousands expected in the coming years. Modern satellites are also vastly more capable than their predecessors.

While communications satellites from the 1970s often weighed several tons and handled comparatively tiny amounts of data, many modern LEO (Low Earth Orbit) satellites are compact, software-defined systems capable of delivering high-speed broadband, direct-to-phone connectivity, and real-time imaging.

Although the satellite industry has already embraced cloud-based ground-station networks, software-defined radios, and electronically steerable systems, high-gain feeder links still heavily depend on large parabolic dishes. Electronically steered phased arrays can, in principle, replace them, but matching dish-class performance remains prohibitively expensive to deploy at scale. This is the specific problem ArrayLink is designed to solve.

“The fundamental bottleneck in scaling satellite connectivity today is not in space; it is on the ground,” says Dinesh Bharadia, senior author of a paper on the research, which was presented at IEEE INFOCOM 2026.

Every bit of data transmitted by a satellite must eventually pass through a ground station before reaching the wider internet. Today, most of these stations still rely on large parabolic dishes, some over 1.8 m (6 ft) wide. These dishes are powerful but also inflexible. Each dish can track only one satellite at a time and must physically rotate to follow fast-moving LEO satellites streaking across the sky at roughly 17,000 mph (28,000 km/h).

This gap creates a serious bottleneck.

The researchers note that some existing satellite dishes rotate at just 2 to 5 degrees per second, meaning transitions between satellites can take several seconds or even close to a minute. During those transitions, the ground station is temporarily unavailable.

To solve the problem, the team turned to phased arrays, flat electronic antennas that steer radio beams without moving parts. Phased arrays already exist in technologies like Starlink user terminals, military radar systems, and advanced 5G infrastructure. However, building one large enough to match the gain of a massive satellite dish would require tens of thousands of antenna elements, making it prohibitively expensive and complex.

• Instead of building one giant phased array, the researchers used many smaller, commercially available phased-array panels and coordinated them as a distributed system.
• “This work enables the industry to scale ground stations rapidly and cost-effectively, even through crowdsourced deployment,” says Bharadia. “Any rooftop owner or enterprise can install our solution and carry satellite data back to the internet.”
• Their ArrayLink architecture combines up to 16 laptop-sized phased-array panels spread across a kilometer-scale area. Each panel individually lacks the power required for robust satellite backhaul links. Together, however, they behave like a giant coordinated antenna capable of approaching dish-class performance.
• But the breakthrough goes beyond replacing dishes.
• By spacing the antenna panels far apart, the team discovered they could exploit a phenomenon called near-field line-of-sight MIMO, enabling multiple simultaneous data streams between satellite and ground station.
• Normally, line-of-sight satellite links are highly limited because every antenna essentially receives the same signal. However, once the panels are spread far enough apart, each one begins to perceive the incoming radio waves slightly differently. These differences allow the system to separate multiple independent data streams from the same satellite simultaneously.
• The concept is somewhat similar to the MIMO technology used in modern Wi-Fi routers and cellular networks, but applied at a satellite scale.
• In simulations, ArrayLink supported up to four simultaneous spatial streams at distances of hundreds of kilometers, while maintaining two streams beyond 2,000 km (1,243 miles). According to the researchers, the setup could achieve up to three times the throughput of traditional single-stream dish systems.
• The system also introduces another unusual capability: focusing energy not just in direction, but also in distance. Conventional antennas typically beam signals in a specific direction. ArrayLink, however, can localize energy both angularly and radially, potentially reducing interference with other satellite systems.
• Importantly, the system is not just theoretical.
• The team conducted real-world hardware experiments at 27 GHz using phased arrays and software-defined radios in outdoor line-of-sight testing. Their measurements closely matched theoretical predictions and simulations, validating the core physics behind the approach.

CHATGPT INTERPRETATION OF ABOVE:

Yes. The key idea is that they are not replacing a dish with one flat antenna. They are replacing it with many small phased-array antennas working together as one giant virtual antenna.

Why dishes are still used

A satellite ground station needs:

1. High gain (lots of signal concentration)
2. Accurate tracking of moving satellites
3. Low noise reception

A large parabolic dish provides all three.

The problem is that a dish:

• is expensive
• has moving motors
• can usually track only one satellite at a time
• takes up space
• requires maintenance

With thousands of LEO satellites now overhead, ground stations are becoming the bottleneck.

What is a phased array?

Instead of one reflector focusing radio waves, a phased array contains many small antenna elements.

By adjusting the phase of each element electronically, the antenna beam can be pointed without moving anything mechanically.

Think of it as:

• Dish = move the whole antenna
• Phased array = keep antenna still, steer the beam electronically

This is already used in:

• Starlink user terminals
• military radar
• some 5G systems

The problem with huge phased arrays

To match a large dish at 27 GHz you need a very large effective aperture.

A giant phased array might need:

• thousands or tens of thousands of elements
• expensive electronics
• lots of power

So replacing a dish with one giant panel is often uneconomic.

ArrayLink’s trick

Instead of one giant phased array:

Roof A       Roof B       Roof C

 [Panel]     [Panel]      [Panel]

      \        |        /
       \       |       /
         Satellite

Each panel is:

• relatively cheap
• laptop-sized
• electronically steerable

The signals from all panels are combined digitally.

To the satellite, they behave somewhat like one much larger antenna.

Why spreading them apart helps

This is the really interesting part.

Normally a satellite is so far away that every antenna sees almost exactly the same wave.

Satellite
    |
    |
    |
Antenna 1
Antenna 2
Antenna 3

There isn’t much information difference between antennas.

However, when antennas are separated by hundreds of metres or even kilometres, tiny differences appear in the incoming wavefront.

Those differences can be exploited using MIMO techniques.

What is MIMO?

MIMO stands for Multiple Input Multiple Output.

It’s used in Wi-Fi and cellular systems.

Instead of one data stream:

Satellite --> Ground

you can send:

Stream 1
Stream 2
Stream 3
Stream 4

simultaneously on the same frequency.

The receiver separates them mathematically.

The UC San Diego team claims their geometry allows multiple spatial streams from the satellite to the distributed array, increasing capacity.

Why this is surprising

Traditionally satellite links are considered almost pure line-of-sight.

Line-of-sight channels usually don’t provide much spatial diversity.

The researchers are exploiting what is called near-field LOS MIMO.

In simple terms:

• Satellite is far away.
• But at 27 GHz the wavelength is only about 11 mm.
• If antennas are spread over large enough distances, measurable phase differences occur.
• Those differences can be used to separate streams.

This is an active research area and is one of the more novel aspects of the work.

What does “focus in distance” mean?

A conventional antenna can focus energy in a direction:

      ^
      |
      |
   Beam

ArrayLink may be able to focus both:

• direction
• range (distance)

This is similar to beam-focusing or near-field beamforming.

Instead of illuminating a whole line through space, the energy can be concentrated near a particular location.

That could reduce interference with other satellites.

As a radio astronomer, what is the closest analogy?

It’s very similar to an interferometer.

For example, if you have several antennas separated by significant distances:

Antenna ----- Antenna ----- Antenna

and combine them coherently, you create an effective aperture much larger than any individual antenna.

ArrayLink is doing something related, but for communications rather than imaging.

The principle of combining signals from spatially separated antennas will probably feel familiar from radio interferometry.

The important caveat

The article’s “up to three times the throughput” is largely based on the ability to support multiple spatial streams and on simulations.

The physics is sound, and they report successful 27 GHz hardware tests, but this is not yet a proven replacement for every large satellite dish.

The biggest engineering challenges are likely to be:

• precise synchronization between panels
• distributing timing references
• fibre/network links between panels
• calibration
• atmospheric phase variations
• keeping costs low enough to beat conventional dishes

So the advance is real and interesting, but it’s better viewed as a distributed phased-array ground station architecture than as “a flat panel replacing a dish.” The gain still comes from collecting signals over a large effective area; they’re just achieving that area using many coordinated panels rather than one large reflector.