[Starlink] Re: Starlink direct-to-device in NZ

[Ulrich Speidel <u.speidel@auckland.ac.nz>]

On 30/10/2025 4:29 am, Michael Richardson wrote:
> Ulrich Speidel via Starlink <starlink@lists.bufferbloat.net> wrote:
>      > In July 2025, One NZ extended the service to phones on prepay plans,
>      > including the plan I'm on ... as long as I'd have an eligible phone.
> Is your phone eligible because it lacks the right radios, or because it's not
> blessed?
I've no idea, but I suspect that a range of cheaper phones (mine does 
3G/4G/5G) have lower sensitivity radios than the higher end devices 
(read: elevated noise floors due to proximity to heat generating 
components in the phone, smaller antenna size, conductors carrying RF 
that aren't gold-plated, coarser signal processing, interference from 
processing chips in the phone etc.).
> My experience with NZ consists of changing planes in Auckland in 2024.
> (And that's my own time to the southern hemisphere).
That's how I first got here so you're in good company ;-)
> I don't know how much
> of NZ is unserved by 3G/4G.  It sounds to be that the One NZ terrestial
> network coverage is pretty good?

Basically, yes. The three mobile carriers here cover - at least to TXT 
quality - most places where people would normally be. Rural locations 
used to be an issue until the government knocked heads together and 
forced the three carriers to found a common subsidiary, the Rural 
Connectivity Group (RCG), to set up cell sites in rural areas on which 
the three carriers would be virtual tenants. The downside of this of 
course is that when these go down, so do all three networks, as we had 
ample opportunity to witness during Cyclone Gabrielle a couple of years 
ago. There is also pretty good coastal coverage for boaties from the 
terrestrial network, although the Starlink-based service reaches further 
out.

The remaining "uncovered" area still makes up around 40% of the land 
surface according to One NZ, but these are generally places where there 
are extremely few people. Much is bush / forest and where there is 
remote farmland, the locals usually have land mobile / CB radios to 
communicate.

>
> Canada has similiar concentrations of coverage, with most of the smaller
> operators having big-city-only coverage.  Many smaller towns can have
> effective single suppliers (Yes, there is a duopoly.  But some towns have
> very few towers from the "other")
Canada is similar in many respects but has a much larger land area and 
lower population density over much of it.
>
> What I'd want is a $4/day-pass for when I go hiking.  I don't think current
> emergency call support covers emergency txt.  Is there even spec for txt to
> 911, I don't know.  It would, I think be tolerant of much lower bandwidth.
> A day-pass could be "messaging" only, like the airlines "free" wifi level.
... which is pretty much what you'd get here (provided you subscribed to 
the more expensive basic prepay package).
>
>      > They're also working on getting the data service working. Which will support
>      > a limited number of mostly messaging apps only by the looks of it. Different
>      > flavour of TXT I suppose.
>
> I'm not sure why this is difficult; if I were asked to implement I'd just
> block a bunch of well-known streaming end-points on day one.  Yes, blocking
> youtube blocks all sorts of other google services.  I'd fix that day two
> as I got bandwidth based caps/throttling implemented.

This isn't a networking problem predominantly, but an RF engineering 
one. Essentially, your classical cellphone network works on the premise 
that you can serve more users by bringing the base stations closer to 
them, which allows (a) lower power use and (b) conserves their battery. 
The lower power use then allows you to re-use your frequencies further 
down the road.

This alone implies that when your "cell tower" is many hundred of km 
away, this goes against the grain of the cell system design philosophy. 
You now need more power (or larger antennas) and re-using your 
frequencies isn't all that easy anymore. You can't put large antennas on 
people's phones, so the only place to put them is in space.

But wait, it gets worse. Frequency matters.

Normal Starlink downlinks to DIshy users on Ku band frequencies. That's 
between 10,700 MHz and 12,700 MHz (note also that's a 2 GHz bandwidth). 
For the Starlink cell service, One NZ gets them to use 1800 MHz in what 
is now a 15 MHz bandwidth 
(https://www.linkedin.com/posts/richardhaas99_new-new-zealand-mobile-operator-one-has-activity-7358493392294580225-NVsr/). 
That's at least a factor of 6 in terms of frequency.

Consider an RF communication system with fixed physical dimensions 
(antenna sizes, distance between transmitter and receiver) and a fixed 
transmit power.  Assume for a moment that you can make this use any 
frequency you like (i.e., ignore antenna resonances, 
transmitter/receiver tuning etc.). The received power that you will then 
have available at your receiver is proportional to the square of the 
transmit frequency. This gives Ku band a 6*6=36 fold advantage over the 
cell band, and that's before you start looking at the bandwidth.

Moreover, the higher your frequency, the more directional your antennas 
become. That is, Starlink has a much easier time projecting a Ku band 
beam at a location than a cell signal. And it sure looks like they're 
struggling a bit with the former, even with Ku band cells much larger 
than your typical mobile phone cell. And that's with you pointing your 
Dishy at the sky as instructed rather than having it at the bottom of 
your gym bag. So your cell phone signal from space isn't exactly laser 
pointer material, and getting the tiny device in your pocket to hit just 
the satellite you're meant to communicate with is an uphill struggle at 
the best of times.

So, basically, fitting data in next to TXT isn't trivial.

For One NZ and their colleagues at T-Mobile etc. overseas, this means 
that once they earmark a cell phone frequency for satellite use, they 
can't really use it on the ground anymore because a satellite using it 
is now going to be "heard" all over the place and not just where the 
user is. Neither can they re-use that frequency in multiple locations 
all that easily. Read: Commit a frequency for satellite use in the 
northern North Island and you can't - in all probability - use it 
anywhere in Auckland. Engineering aside, they now face the extra problem 
that ... spectrum is expensive. In 2021, that cost NZ$720,000 per MHz 
(https://www.rsm.govt.nz/about/news-and-updates/renewal-of-management-rights-in-the-1800-mhz-and-2100-mhz-bands). 
So that 15 MHz band for D2C would have cost One NZ just upwards of US$6M.

So by running D2C, they're essentially throwing an expensive resource at 
an application with fairly limited earnings potential.

But hey, it's great if all you need is TXT on a hike.

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

School of Computer Science

Room 303S.594 (City Campus)

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