
This is the second part of a three part article series. In the first part, we explained the growing need for fast internet, everywhere. In this second part, we talk about Elon Musk’s new Starlink internet service; the third part talks about other and competing exciting new technologies that also promise to deliver the fast affordable internet that is becoming so essential.
As a quick reminder (and please read the entire first part, of course, if you wish) the problem that rural and low-density population areas struggle with is that the most common forms of internet – DSL, cable, and sometimes fiber, are not available. With individual houses so far apart, there are no economies of scale, making it very difficult for commercial internet service providers to come up with an internet service that is both appealing to customers and acceptably priced.
Some estimates suggest there are as many as 50 million households in the US with either no internet availability at all, or internet that is either too slow and/or too expensive. It isn’t just the US – most other countries with patches of low density rural populations have the same problems. There are other types of users who could benefit from new types of anywhere/everywhere internet service too – ships at sea, airplanes flying through the air, and many other sorts of non-fixed internet users.
Clearly, whether considering only rural households in the US, or all potential users, in all countries, there is an enormous marketplace opportunity for internet service that is fast and affordable.
We can only guess what drew Elon Musk’s attention to this (over and above the phrase “enormous marketplace opportunity”!) but it is clearly an intersection between at least two of his other interests – Tesla cars (prodigious users of internet data, both sending and receiving) and SpaceX rockets, as well as light connections to his Gigafactory battery production and his solar panel production. With his impressive ability to sense both an opportunity and a potential solution, invariably applying cutting-edge technology ahead of others, he developed the concept of a “constellation” of an enormously large number of low-earth-orbiting (LEO) satellites to beam internet down to the ground. This concept, named Starlink, was first announced in January 2015, and is now becoming more and more operational, with dozens more satellites being launched every month.
The twin problems of geosynchronous satellites, necessarily hovering over the equator, and the internet service they offer, is there are not and never can be enough of them, so that means insufficient bandwidth to serve large numbers of users; and they’re too far away, meaning long latency and slow perceived connection speeds. Both these problems are addressed with LEO satellites. Instead of being 22,200 miles above the equator and usually 25,000 – 30,000 miles from a user, the first tranche of Starlink LEO satellites are between 335 – 354 miles above the ground, and flying on paths at times almost directly overhead the user. The satellites come and go quickly – they typically take 90 – 95 minutes to do a complete orbit of the earth, traveling at a speed just over 17,000 mph.
The large number of Starlink satellites mean lots more bandwidth, and their low orbit and proximity to the users on the ground mean low latency, acceptably comparable to normal ground type connections (all the more so because data travels 50% faster through space than through fiber cables).
Where the concept becomes grandiosely Muskian in nature is when you start to look at the details. The “large number” of satellites is anticipated as growing to over 40,000 (42,000 has been an often mentioned number – a lot depends on the regulatory approvals he secures). To put that into context, the constellation of satellites that provide service to the Iridium satellite phone service comprises a mere 66 satellites (plus nine orbiting spares, and another six on the ground). In total, since the very first satellite launch (Sputnik on 4 October, 1957), there might have been as many as 11,139 satellites launched, with about 7,389 still in orbit. Musk’s plan to deploy tens of thousands is extraordinarily ambitious by any and all measures, and would not have been possible without the latest developments in miniaturization, and his low-cost rocket launching via the reusable SpaceX Falcon 9 rockets.
The first launch of 60 operational Starlink satellites was on 23 May 2019, and as of the same date in 2021, there were 1,737 Starlink satellites in orbit, with 60 more due to be launched any day (as of the end of June). A “beta test” version of the service is already being used by perhaps 100,000 people around the world, with plans to grow that number to over 500,000 later this year (at least that many people have already signed up and joined the waitlist).
At present, the version 1.0 satellites are described as being about the size of a table, and weigh 573 lbs each. They get power from a solar panel (about 3 kW). Various internet estimates refer to projected operational lives per satellite ranging from a low of 3 years to a high of 7 years. Probably no-one exactly knows yet, with the life expectancy being primarily a function of how quickly each satellite uses up its store of maneuvering fuel for maintaining its orbit.
An interesting thing about this is that if we assume an average life of 5 years, then with a 42,000 satellite constellation, Starlink will need to be replacing 23 satellites a day. If they continue to launch them on Falcon 9 rockets, 60 at a time, that’s a launch rate of just over one launch every three days.
And as for the dead satellites? They’ll be “de-orbited” – a polite way of saying either deliberately steered down to a lower altitude so they then return to earth and burn up on the descent. If the satellite has become unresponsive or has insufficient fuel for this, the satellite will naturally do the same thing over some time, due to the friction of the air, which, while almost non-existent at 340 miles above the earth, is still measurable over time and the distance/speed the satellites are traveling.
The End User Experience
With so many satellites in the sky, on a series of different orbit paths, the concept is there will always be a nearby satellite for users to connect with. Actually, as the numbers of satellites grow, there will be multiple satellites within range, allowing for better management of connection quality.
With multiple satellites available, it will also be possible to reduce the impacts of bad weather – a notorious problem with current geosynchronous satellite service. The system could simply connect each user to the satellite with the least interference, and with the most available spare bandwidth.
Starlink has made various different claims as to the amount of bandwidth it expects all users to have available, but it seems at present (while in Beta test mode) they are expecting about 100 Mb/sec and as the satellite constellation increases, they expect 150 Mb/sec, maybe more, and with latency ranging between 20 – 40 msec at present, and becoming shorter as the constellation matures.
Installing and setting up Starlink has been designed to be simple and easy and not require any special equipment or technicians. You simply mount the dish somewhere/anywhere that has a good view of the sky, and turn it on. Musk jokes “you can even do these two steps in either order”. The dish can be on your roof, but does not need to be, it could be on a short pole just a few inches above the ground, too. The dish units even have built in heaters to melt snow off them in the winter.
A cable runs from the dish to an indoors router/Wi-Fi hub – the starter kit includes the dish, the hub, and the necessary cabling.
Where Starlink Won’t Work
Most of the satellites go up to about 55°N and down to about 55°S, and due to their coverage fanning out below them like the beam from a flashlight, they can provide coverage for some distance further north and south of their orbits. A few are planned to go higher/lower, over the poles.
That means almost the entire inhabited planet, other than for northern Canada (and Alaska), Greenland, Iceland, Scotland, Scandinavia and Russia will be well covered, and noting how Starlink has three groundstations either already in service or planned in Alaska, clearly they planning for the far north and south.
This is an amazing map that shows, in real time, where all the Starlink satellites are. You can zoom in, and move the globe around, and get details of each satellite and its orbit. There’s something almost hypnotic at watching the never-ending march of satellites around the world.
Note the rings around each satellite are not the same as the prime coverage of each satellite, but rather are the approximate limits. The six red rings are from satellites the program suggests could provide coverage at my present location, at the moment the image was taken. At times I’ve seen as few as two available satellites (as was the case five minutes after this screenshot), and it is possible the number could drop down to even one or zero, especially after considering practical issues such as obstructions, weather, and satellite loadings.
Keep in mind also that this image was taken when there were 1757 satellites in orbit. Now imagine what it would be like with seven times as many (ie 12,000, the total number the FCC has approved) and what it would be like with a full 42,000 (the number asked for from the ITU).
So, geographically, for most of us, there is likely to be a Starlink satellite overhead pretty much wherever we are. Starlink say the next launch or two will get enough satellites up to ensure continuous coverage all over the globe.
But just because there is a satellite physically “in sight” doesn’t mean it will be available. It is possible the satellite currently servicing a specific area might be fully used by other users.
Much of the more technical details of the satellites and their capabilities has not yet been fully and formally disclosed, so there’s a fair amount of guesswork involved in trying to understand what their limits currently are. There are also plans to greatly extend the capabilities of the satellites during an evolution from type 0.9 satellites to type 1.0 satellites (the current generation) and then on to 1.5 and 2.0 versions. More bandwidth, managing more simultaneous connections, and better satellite to satellite linking all seem to be factors being improved.
Each Starlink satellite is described as currently having about 20 Gb/sec of bandwidth (or possibly three times more, it depends how you decode some of Starlink’s and Musk’s cryptic comments). It is also not clear if that means bandwidth available to users, or total bandwidth, both for receiving and sending signals. Because the satellites are simply relays, every byte sent to a satellite is then repeated and passed on to either another satellite, an end user, or a ground station, so each byte to or from an end user actually takes two bytes at the satellite – one in and one out.
If we assume the 20 GB is available for end users, that suggests one Starlink satellite works well with 200 users, if all are using 100 Mb/sec. Of course it would almost never happen that all users were simultaneously maxing out their bandwidth, so beyond that, a satellite could “oversell” its bandwidth and handle more users.
Maybe a single satellite can handle 2,000 users (ie, overselling the bandwidth ten-fold), maybe more. It has been suggested that some cable companies oversell their bandwidth by 50 and even by 100 times, so overselling by “only” ten times seems extremely conservative.
We also don’t know if there is a limit on how many different connections a satellite can be “talking to” simultaneously. This might be a limit that could be more restraining than the bandwidth limit, and would explain why Starlink is talking about offering higher bandwidth to users in the future. On the other hand, improving electronics in future satellite generations might allow for increased numbers of concurrent connections.
We understand that Starlink has mapped the earth into a number of coverage cells for the purpose of controlling demand and density of coverage. These are believed to be anything from as small as 15 miles across to as large as, well, something very much larger. Currently Starlink does not allow a satellite receiver to work outside of its assigned “home” cell so as to manage user densities. We expect this will change in the future for two reasons – firstly, with more satellites, the Starlink network becomes more resilient and better able to handle shifts in demand, and secondly, we think part of the restriction at present is due to the form of licensing granted to them by the FCC, perhaps only for “non-mobile” use.
However, one thing is likely to remain impossible – Starlink service in larger urban and suburban areas. In such places, a single 15 mile sized cell might hold many hundreds of thousands, even millions, of potential users – definitely at least ten times more users than Starlink could ever hope to handle, and probably a hundred or more times too many. Not only could Starlink never handle high densities of users, but existing services such as fiber and cable and do good jobs at much lower cost. Starlink accept that and say their service is designed for only 5% or so of all internet users.
There are other limitations, too. A Starlink dish needs to have a broad view of as much of the sky as possible – ideally unobstructed most of the way to the horizon in every direction, and certainly, more than 45°. No trees, no buildings, and of course, no hills or mountains. The radio waves used by Starlink can not really penetrate anything. The dish should be within 100 ft of your indoors Wi-Fi router/hub (although this can be exceeded with some workarounds).
The more blockage there is, the more regularly you’ll lose signal – perhaps only for a ten seconds or less at a time, perhaps for 5 – 10 minutes. We hope the major video streaming services will increase the size of streaming buffers they have, so that if you’re watching something on eg Netflix, when the internet goes down, the video will keep playing from your streaming buffer, and by the time the buffer has emptied, you’ll have internet service restored and never realize it went down.
YouTube seems to store 60 – 90 seconds of feed (at least, it does so on our Amazon Fire TV 4k stick); that is enough for brief glitches, and clearly there’s no reason (other than having sufficient memory in the streaming device) why that couldn’t grow to a larger amount.
The Cost and Availability of Starlink
Currently, it costs $499 to buy a Starlink starter kit, and then $99 a month to receive Starlink service. Unlike many regular internet services that then proceed to add all sorts of fanciful additional fees on top of the quoted monthly cost, the only extra cost that might appear is a state sales tax, if one is imposed by your state on internet service.
Starlink says (in June) that the dish kits are costing them about $1000 each. But they’ve also said they want to get the retail cost of the kits down to $300 or less, so hopefully the pricing direction on that will be down rather than up.
There are no data caps, and there are no alternate plans with faster or slower data rates. You don’t need to sign up for a fixed period of service, you could join today and leave a month later if you wished (but of course, you’ve sunk $500 plus shipping into the starter kit).
So, if you want to sign up for Starlink, can you? Currently, it is in a Beta test phase, with a limited number of users in a limited number of regions, and seems to have filled the available spots, with perhaps half a million extra people on a waiting list, hoping to be added to the service in the future.
You can try signing up for the service on their website, but the chances are you’ll get a message the same as one of the two shown here. The top message is a polite way of saying “Don’t be ridiculous, you live in a city with lots of other options”, the bottom message is more encouraging.
Starlink has been keeping reasonably to schedules over the last year or so, which gives one hope that their objective of adding more users in some areas in mid to late 2021 is realistic.
At present it is available in about a dozen different countries, but earlier this week Musk said it will become available, world-wide, in August. However, we would caution that “world-wide” is not the same as “every country in the world” – many countries will have regulatory requirements which Starlink might not qualify under.
Is Starlink the Answer?
If you have no other internet alternatives – well, that’s a big “if”. Just about everywhere in the US (other than in some valleys) can get some type of geosynchronous satellite based service with either Viasat or Hughesnet. But when Starlink becomes available, it will clearly be vastly superior to these two alternatives. Even if both other companies reduce their prices, they can’t do anything about their long latency and limited amount of bandwidth.
If you are somewhere that is already giving you reliable internet, but slower and/or more expensive than Starlink, maybe Starlink will be a good option. It might not be an ideal option if you need to do a lot of interactive voice and video calls such as even short 5 – 10 second losses of signal are problematic, but if you mainly use your service for email, web browsing, and video streaming, Starlink should be excellent.
And to answer the question we posed in the headline – Will Elon Musk’s Starlink service revolutionize internet prices, speed, and availability? The answer to that is a limited and qualified yes. It won’t make any difference in places that are already well-served by wired or wireless type internet services, but in lower-density areas that currently lack fast and affordable internet, it will indeed be a game changer.
But wait. There’s more to consider.
Other Alternatives
There are a number of other internet services in development. Some are similar to the Starlink concept, and have big name backers – as big as Jeff Bezos for one such service, and the British government for another. Others are extensions of cell phone services, and still others are in early stages of development but with great promise.
For details on these, please see the third part of this series.
This article you’ve now finished reading is the second part of the three part series. In case you missed it, please also check out the first part of the series about the growing need for faster internet here.