[Starlink] apnic piece on starlink

[Ulrich Speidel]

Not so fast! :-)

It's one thing to extend LEO coverage to new regions, especially rural 
and remote ones, but quite another to do so economically and at scale. 
For an urban user on fibre, Starlink's performance is anything but a 
game changer: lower data rates, more latency (even without jitter - and 
there's plenty thereof on Starlink, especially as Dishy switches between 
differently loaded satellites) and significantly higher cost. Plus: If 
you're on fibre, the equipment at the end of it determines how fast it 
can go, and we're now seeing 2Gb/s and 4Gb/s services to consumers being 
offered in a lot of areas under the "hyperfibre" tag. It's foreseeable 
that as demand grows with IoT, high definition video etc., this will 
scale up further as we go. There's really no serious hurdle to stop it.

With anything that needs to use the radio spectrum, the limited capacity 
thereof is the main hurdle. Sure, you can put more satellites in orbit, 
but without narrower beams this will not allow more frequency re-use. 
You can give them larger phased arrays to have narrower beams, but these 
won't give you the order of magnitude of improvement you need in order 
to serve the unconnected. You can give the satellites more solar power 
and hence more lasers, on-board processing capacity and downlink power, 
but that works against frequency re-usability to a good extent.

Remember how cellphone networks evolve: You start with a few towers in 
high spots using high power to get wide area coverage while you have few 
users. At this point (which corresponds largely to where Starlink is at 
now), spectrum isn't much of an issue (and even that is only partially 
true for Starlink - see Mike's excellent article on this: 
https://mikepuchol.com/modeling-starlink-capacity-843b2387f501). As your 
user base grows, you move off the hills into the valleys and lower your 
power so your cells become smaller and shielded from each others, 
because now, frequency re-use is the name of the game. You use 
beamforming off phased arrays in order to further separate users.

So what we are seeing now is Starlink as the new kid on the block 
turning up with what are in analogy effectively cell towers high in the 
sky. Their current user base is maybe at 1/1000th (ballpark) of 
potential demand before growth. Population growth on this planet alone 
adds a lot more potential users a day than Starlink does. So what 
options does Starlink have to scale? Unlike a terrestrial network 
operator, Starlink can't really come down all that far from their "space 
hills" without burning their satellites up in the atmosphere more 
quickly. "Space hills" also consist of vacuum only, which unlike earthly 
hills can't separate base stations by blocking signal. The distance 
from/to space also requires vastly larger phased array antennas for the 
same spot beam coverage area contour on the ground. It also places 
limits on transmit EIRP both ways. Larger antennas and solar arrays 
constrain the number of satellites that can be launched at a time, 
making constellation building and replacement harder.

It's a joy to see people in some remote places finally getting Internet 
connectivity via Starlink that somewhat resembles what most of the rest 
of us get. But even now, we see Starlink quite obviously dealing with 
capacity issues: See my recent post on the "strange" differential 
hardware pricing in NZ (you pay over 250% extra for Dishy now if you 
live in a big city here), or their exorbitant pricing for maritime 
service at over 10x the rate for a fixed location connection. Or the 
(not so) strange lack of availability in the eastern US, along with many 
spots scattered around the planet in medium-density rural areas. Is this 
what a network looks like that will be able to grow service capacity by 
a factor of a 1000 or more in the next decade?

The good news: Growth by a factor of 5 or 10 might get most smaller 
Pacific islands covered, which is dear to my heart, but affordability 
will still leave us with a digital divide there. But it's not going to 
get Internet to the two billion un- and underconnected in Asia, and 
their peers elsewhere.

One of the three big telcos here (One NZ, until recently known as 
Vodafone) announced last week that they were partnering with Starlink to 
increase mobile coverage from some time next year, from the existing 98% 
to 100% of the country. I promptly got called for media comment again, 
and of course what people read into the announcement was that we'd all 
be able to throw out our existing wired Internet connections and Dishys 
and could use our mobile phones instead. So I tried hard to point out 
the fine print - that it's going to be text and fairly low quality voice 
only, and perhaps some low bandwidth data services such as e-mail or 
messenger apps. That it won't work indoors or from inside vehicles. That 
it was only for the 2% of locations that got no signal at all so far. 
That it wasn't going to bring Netflix to the wap-waps. But that it was 
going to be great for trampers (hikers in local lingo) who needed to let 
their pick-up service know if things changed.

The question I got back was whether it was going to put community-based 
wireless ISPs out of business? Um, no.

6/04/2023 8:09 am, Dave Taht via Starlink wrote:

> I stumbled across this because he cited me, but it is very thoughtful
> and interesting, otherwise.
>
> https://blog.apnic.net/2023/03/31/everything-everywhere-all-the-time/ 
> <https://blog.apnic.net/2023/03/31/everything-everywhere-all-the-time>
>
> I really do think we are on the verge of being able to cover the rest
> of the world with internet, through LEO technologies.
>
> -- 
> AMA March 31: 
> https://www.broadband.io/c/broadband-grant-events/dave-taht 
> <https://www.broadband.io/c/broadband-grant-events/dave-taht>
> Dave Täht CEO, TekLibre, LLC
> _______________________________________________
> Starlink mailing list
> Starlink at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink 
> <https://lists.bufferbloat.net/listinfo/starlink>

-- 
****************************************************************
Dr. Ulrich Speidel

School of Computer Science

Room 303S.594 (City Campus)

The University of Auckland
u.speidel at auckland.ac.nz  
http://www.cs.auckland.ac.nz/~ulrich/
****************************************************************


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/starlink/attachments/20230407/f5085961/attachment.html>