[Starlink] Starlink Internet Speeds Could Skyrocket to 2 Gigabits Per Second, SpaceX President Says

Sat Nov 23 08:30:32 EST 2024

On 23/11/2024 7:05 pm, David Lang wrote:
> Ulrich Speidel wrote:
>
>> Right now, Starlink have reached capacity in quite a number of places.
>
> True, but they are still launching under their existing 
> authorizations, with several thousand more satellites authorized 
> (IIRC, they can at least double their existing number, plus they have 
> authorization to replace older satellites with newer ones.

Yep. But the "sold out" is new. Previously, when they ran short, it was 
"coming soon" or something along those lines.

> they will add capacity, add users, be 'at capacity', launch more...

Not so sure.

They only have 2 GHz for downlink in the Ku band, and the power flux 
density limits they're subject to pretty much limit downlink to between 
12 and 20 Gb/s max depending on which generation sat contributes to the 
downlinks into a cell. More isn't possible for a single cell, but even 
to get to this limit, you need more than one satellite because the 
maximum EIRP of the beam transmitters on each bird is only just enough 
for a part of that band. So you always need multiple beams - between 3 
and about 12 - to fill that band completely, and this needs beams from 
multiple sats usually. That gets further de-rated if neighbouring cells 
also want / need service: In this (overwhelmingly common) case, one cell 
can't monopolise all beam frequencies and polarisations across the 
entire Ku spectrum - it needs to cede at least some freqs / 
polarisations to its neighbours.

More sats help if there simply aren't enough spare beams available up in 
the sky to service a part of the Ku band that isn't already in use in 
your cell or its neighbourhood. That's a "I simply haven't got enough 
beam transmitters available" problem that can be solved with more beam 
transmitters, which is what you get with more sats.

More satellites don't really help if all frequency + polarisation 
combinations are taken in your area. Any combination that is taken can't 
simply be supplied by another satellite, even if that satellite has a 
spare beam available. That's a "I have more users that I can possibly 
serve" problem.

Think of it like a popular restaurant where the metric is how many 
diners you can serve in an hour. If the restaurant is understaffed, you 
can hire more staff and have them cook and serve faster. If all tables 
are taken and you can't fit any more in, you can employ as many extra 
chefs and waiters as you like, but that doesn't really help you much as 
people still take as long to eat as they always did.

The areas that have now gone "sold out" appear to be without exception 
urban fringe areas that are underserved with fibre. Expect a few more to 
move into that category in the next little while (watch Australia's big 
cities, for example).

There is perhaps some wiggle room in terms of power flux density limits 
(these have been challenged), in terms of making beams narrower and 
cells smaller over the longer term, and in terms of improving frequency 
re-use by serving nearby cells on the same frequency / polarisation from 
different directions. If POTUS is your best friend and you can get him 
and your mate that's running the FCC to armwrestle the ITU into giving 
you another gig of Ku band, then we might also see this open up a little 
again. But none of these techniques appears to be a game changer that 
offers the growth potential that you need to keep pace with Internet 
traffic growth.

If there is some semblance of a local distribution network (or one can 
be built short term) then another option are SpaceX's Ka-band community 
gateways. But in Ku, there's not much room left in the inn anymore by 
the looks of it. At least in some areas.

>
> David Lang
>
>> The availability map on Starlink's home page shows that Starlink is 
>> "sold out" in many places, including London, Manila, Rio de Janeiro, 
>> Seattle, Portland, Sacramento (California), Edmonton, San Diego, 
>> Austin (Texas), Mexico City, Guadalajara, Brisbane, Accra, Lagos, 
>> Nairobi, Lusaka, Harare, and many more:
>>
>> https://www.starlink.com/us/map
>>
>> This isn't surprising given the fact that Dishys to date only use 
>> Ku-band, there's only 2 GHz of it for user downlink, and you can't 
>> use the same beam frequency in adjacent cells.
>>
>> SpaceX have a modification application before the FCC that, if 
>> successful, would allow them to:
>>
>> * Up power flux density on the ground. This'd allow satellites to
>>   transmit with higher power. Note that none of the current beam
>>   transmitters on the satellites have sufficient EIRP to hit the
>>   current PFD limits across the entire Ku-band. But the Gen. 2 ones
>>   are supposedly only by a factor of about 2.7 off, so with Starship
>>   able to carry heavier sats, there might be room for a bit of growth.
>> * Use satellites down to 20 deg above the horizon instead of the
>>   current 25 deg (this mightn't look like much, but if my calculations
>>   aren't wrong, means that they'd see about 43% more of the orbital
>>   sphere with that increase alone).
>>
>> SpaceX have tried for a long time to get into lower orbital height 
>> shells. This makes sense from their perspective: Each satellite's 
>> beam footprint becomes smaller, which makes frequency re-use easier. 
>> Path loss decreases, and a ground station sees a smaller fraction of 
>> satellites in that shell, so they can argue that since the ground now 
>> sees transmissions from fewer satellites, EPFD limits are less 
>> critical, which allows them to up power. Makes for a couple of bits 
>> more per symbol perhaps. Latency goes down a little, too, and they 
>> now have the numbers in terms of satellites, so it doesn't matter so 
>> much that these shells need a larger number of sats to work.
>>
>> Now there are drawbacks also: The lower the orbits go, the more 
>> residual atmospheric drag there will be, and this expresses itself in 
>> either shorter sat lifespan or the need to carry more fuel, which 
>> either means they'll need to launch at a faster rate or with fewer 
>> sats per launch. It's also a bit more crowded in lower space, as this 
>> is where a lot of earth observation spacecraft sit (if you want to 
>> take detailed pics of the Earth's surface, you want it to be as close 
>> to your camera lens as you can have it), and some of those aren't 
>> there for open source public good science.
>>
>> On 23/11/2024 11:33 am, Dave Taht via Starlink wrote:
>>> To me, the additional speeds don't matter all that much.
>>>
>>> I am presently in gale force winds, my boat rocking, and my latency
>>> stable, and only about 50mbit down:
>>> https://www.waveform.com/tools/bufferbloat?test-id=a14b4467-16d7-4b6e-8736-1593813d6eda 
>>>
>>>
>>> Maybe a little less packet loss would help, as my last (hour long)
>>> videoconference broke up twice, and bbr is seriously outperforming
>>> cubic. In addition for aiming for higher speeds, improving density and
>>> reliability would be nice, but otherwise I am a pretty happy camper
>>> with the service I have, compared to 5g.
>>>
>>> On Fri, Nov 22, 2024 at 2:16 PM Hesham ElBakoury via Starlink
>>> <starlink at lists.bufferbloat.net> wrote:
>>>> https://cordcuttersnews.com/starlink-internet-speeds-could-skyrocket-to-2-gigabits-per-second-spacex-president-says/ 
>>>>
>>>> _______________________________________________
>>>> Starlink mailing list
>>>> Starlink at lists.bufferbloat.net
>>>> https://lists.bufferbloat.net/listinfo/starlink
>>>
>>>
>>
>
> _______________________________________________
> Starlink mailing list
> Starlink at lists.bufferbloat.net
> 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/
****************************************************************