SpaceX, a rocket launch company that specializes in re-usable rockets that provide low-cost delivery of payloads to earth orbit, has a project called Starlink. Starlink is a constellation of low earth orbit satellites designed to provide high-speed, low-latency internet service to rural and other under-served areas.
Traditional satellite internet is provided by a big telecommunications satellite in geostationary orbit. A typical one of these satellites orbits at a height of about 22,000 miles. At that height, the satellite magically sits in the same spot in the sky and so your satellite dish on the ground doesn’t have to move to track it. That makes the dish at your house a simple and affordable piece of hardware that you aim once and forget about.

Geostationary orbit is pretty cool.

But there’s a down side to these satellite orbits that don’t move across the sky. They have to be really high up there. 22,000 miles up is a long distance for a signal to travel. Even at the speed of light, the time it takes the signal to make a round trip from the dish mounted on your roof to the satellite in orbit is theoretically about a quarter of a second and most providers tested spend more than half a second establishing those connections. That’s far too much delay, or latency, for something like voice communication or many video games or other interactive internet applications so these satellites don’t make for a great internet experience. Simply put, they’re “laggy”.

SpaceX’s approach with Starlink is different. They put their communications satellites in orbit at a height of only about 350 miles. At first blush, this sounds really smart. The latency for 350 miles is going to be far less than the latency for 22,000 miles. There’s a problem with this “low earth orbit” approach though. Satellites at this height move across the sky and don’t sit in one position relative to the ground. That means a couple of things. First, you need many satellites, not one like with geostationary orbit. When one low earth orbit satellite passes over the horizon, another satellite needs to come into view to provide constant coverage, and then another, and another. And that also means your antenna needs to to move to track a satellite as it moves across the sky, and then your antenna needs to quickly move back the other direction to connect to the next satellite coming into view. A motorized aiming dish makes for an expensive, complicated, and failure-prone setup for a consumer service.

The clever engineers at SpaceX have a solution to this called a phased array antenna. A phased array is a large collection of tiny antennas that sit next to each other in a panel and are each powered independently. Depending on which of the antennas in the array are getting power, the antenna signals interfere with each other in a way that sort of bends the beam to aim it. That way you don’t need a motor to move the antenna to track the moving satellites. Instead, you point the antenna array panel directly up and it electronically tracks the moving satellite across the sky and easily picks up the next one when the current one moves over the horizon.
Now, phased array antennas have a down side. They’re expensive. But SpaceX thinks it can manufacture them in high enough volume to bring the prices down to something acceptable to home internet users. We don’t know what that price is going to be yet, but I’m optimistic. (And also willing to pay quite a bit to get a better internet connection.)

So, why am I telling you all of this. Well, it’s because I can’t use traditional geostationary satellite internet but I will hopefully be able to use Starlink. The reason for this is because of our tall trees. Remember how I said the traditional geosats sit at one position in the sky? Well, typically that position is at an angle of about 40-45 degrees and we live in a redwood forest with extremely tall trees and there’s simply no place we can mount a satellite dish and aim it at 40 degrees without there being a big tree blocking the view of the sky, and thus the satellite. But Starlink’s antenna doesn’t point at 40 degrees. It points straight up, at 90 degrees, and pointing straight up, say above our house, doesn’t run into any trees so it’s a very real option for us.
SpaceX needs to put literally thousands of these satellites into orbit to provide good coverage. (Over the last year they’ve launched about 800 satellites.) Good coverage means a couple of things. First, there needs to be enough up them up there that there’s always one overhead. If there aren’t enough and one satellite goes over the horizon before the next satellite comes into view, you’d have a gap in service and no one wants an internet connection that’s on for 5 minutes then off for 5 minutes. The second reason you need a lot of them up there is that each satellite only has the bandwidth to serve a limited number of customers and if you want to serve large numbers of people you need more satellites. With the 800 or so they have in orbit now, SpaceX has continual coverage for the northern parts of the US but doesn’t have enough to support very many customers. With the continual coverage they’ve got, SpaceX has started a beta program for people willing to test the service and early results are looking good.

SpaceX reported recently to the FCC that they are able to provide more than 100 Mbit/s connections with latencies below 30 ms. Those specs mean that Starlink can compete with modern terrestrial broadband (about 60 times faster than the internet connection I have today.) I’m very excited and hopeful that in 2021 we’ll be able to sign up for Starlink and finally have real broadband.

If you have any questions about Starlink, let me know. I’ve been following it very closely and I think I know pretty well how it’s going to work.

Starlink's plug and play phased array antenna used for internet service. Looks like "a ufo on a stick",