If you have been waiting for decent Internet connectivity in your corner of the world, then 2021 is promising to be an exciting year. Over the last few years, SpaceX’s Starlink company has not-so-quietly, and rather visibly, put approximately 1,500 LEO communication satellites into orbit. They have also recently started beta testing with end customers in North America, Australia, New Zealand, and a few European economies while they are planning to get another 10,000 or so satellites in place eventually. Commercial operations are rumoured to be just weeks or months away. Their nearest competitor, OneWeb, is recovering from a financial hiccup but is also well on its way to a workable satellite constellation. A couple of other competitors are biding their time for now, but are backed by organizations with significant financial muscle, such as Amazon, so it’s likely they will soon appear on stage as well.
What makes LEOs interesting?
Why are LEO systems exciting for Internet users? Here are a few reasons:
- Coverage: Conventional geostationary (GEO) satellites sit over the equator on a fixed longitude. GEO satellites are expensive to build, deploy, and maintain and tend to congregate over longitudes where there are lots of customers with deep pockets. That naturally favours the Americas, Europe and Africa, and to a lesser extent Asia. The Pacific tends to be a bit of an afterthought, which means new players there are always welcome. SES’s O3b mPower medium earth orbit (MEO) satellites have been a game changer in the tropics and subtropics, but don’t cover the polar regions. LEOs, by the nature of their orbit, cover every degree of longitude and, depending on their orbital inclination (tilt with respect to the equator), can also cover every latitude, all the way to the poles. This all sounds very interesting for customers in the Pacific.
- System capacity: Ever since the emergence of cellular communication, engineers have learned that putting transmitters and receivers close together in ‘cells’ is smarter than having one mighty central transmitter covering everything across the land. That’s because one can reuse frequencies with low power transmissions in multiple locations, which takes a lot of pressure off the scarce radio spectrum resource. LEOs are the satellite equivalent of cellphone base stations; they only ‘see’ a small circle on the Earth’s surface — typically a couple of thousand kms or so in diameter — while their GEO cousins hear and see almost half the planet from their lofty station. Moreover, being closer to the party you are transmitting to or receiving from means that you can do so with less power and a smaller antenna. Again, it’s something we know from our mobile phones — batteries last longer now (in town), don’t weigh as much as a brick, and the antennas are so small that they hide inside the case. A LEO 500 kms overhead has roughly a 5000:1 advantage over a GEO satellite nearly 36,000 kms overhead, power-wise, and about 250:1 compared to a MEO satellite. Goodbye big dish and big satellite with large solar panels.
- Latency: A signal travelling via a GEO satellite has a minimum round-trip time between the ground stations of about half a second. That’s extremely awkward for telephony, but it’s also a pain for the Internet’s staple transport protocol, TCP, which relies on feedback from the far end in order to ’trust’ the channel with more data. If that feedback is slow in coming — and via a GEO satellite, it sure is — the TCP struggles to get its transmission rate through the satellite link bottleneck right. That’s why those lucky enough to live on an island accessing a GEO satellite know that transferring large amounts of data can sometimes be faster by plane or boat. O3b’s MEOs do a bit better here, but they’re still a bottleneck to TCP. In principle, LEOs can offer low latency and larger bandwidth, so present less of a bottleneck to TCP.
- Cost: GEO capacity has, to date come with price tags in the hundreds of US dollars per megabit per second of bandwidth per month — capacity that came with all the latency caveats above. LEO pricing is still a little unclear, but Starlink’s beta pricing indicates something comparable with a domestic fibre connection (if on the higher side of that scale) and with claimed performance in the same ballpark. Also, the entry kit is currently USD 499 in beta. That’s possibly cheaper than what the commercial version will cost, but still a far cry from the thousands to tens of thousands that GEO satellite link ground hardware for much less capacity can cost.
The knowns, the known unknowns, and perhaps some unknown unknowns
This certainly sounds very promising, and it’s rightfully endorsed as a potential game-changer. But as LEOs get people’s hopes up as to what it might do for them, it’s arguably not that simple either. The current LEO crowd haven’t been overly forthcoming with technical details, and this is where devils tend to hide. What has become public knowledge has come to light almost exclusively as a result of public regulatory filings, and has often subsequently changed as LEO providers’ plans and circumstances changed. Most scholarly publications in the field need to resort to speculation as to configuration and strategy rather than being able to refer to open standards, publications by the providers, and published patents that allow an assessment of the state of the art.
That said, there are things we do know, and back-of-the-envelope calculations that we can perform, to work out roughly where the limits of the current and proposed LEO networks might be. In the upcoming blog posts, I’ll be discussing the issues I’d like LEO providers to reveal more information about, as well as what we know already, and what that tells us about what we can expect.
The next instalment of this blog series will look at LEO constellations, gateway placement issues and antennas.
Part two is now available here.
Dr Ulrich Speidel is a senior lecturer in Computer Science at the University of Auckland with research interests in fundamental and applied problems in data communications, information theory, signal processing and information measurement. The APNIC Foundation supported his research through its ISIF Asia grants.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.