Is there any document/information on how SpaceX manages its fleet of
thousands of Starlink satellites ?

Since the TLEs are crucial to the service (allowing retail antennas to
point in right direction and for those newer satellites with the space
lasers, know where to aim the laser to destroy ^H^H^H^H^H^H communicate
with other satellites,

how does SpaceX keep all the TLEs up to date? I assume radar tracking
from ground is not realistic, or does that work for such small guys in
the sky?

Or is each satellite equipped with enough instruments to calculate its
own TLE and transmit back to planet Earth? Can precise TLEs be done
with GPS alone? would it need some star tracker on each satellite?

How precise must the time of crossing ascending node be? aja, if ytou
get GPS samples each 2 seconds would being 1 second off detecing when
you crossed latitude 0 be a big no-no requiring calculation of
estaimated time of crossing it based on the 2 nearest readings?

Would it be correct to state that the only way for SpaceX to detect a
lost satellite is when it doesn't say "hello" when it passes over a
ground station? Typiocally, would lack of "hello" when passing over 1
ground station trigger the alert, or would they wait for no "hellos"
from a satelite over multiple stations?

I take it when SpaceX needs to send stuff to a satellite, it calculates
which ground station will next have view on it and get that ground
station to send the "execute order 666" command to the satellite as soon
as it says "hello" ?

Or would they send the command to all ground stations and the first one
that sees the satellite sends the command, and report back to HQ that
the command was sent succesfully and all other ground stations told to
remove the command from their queue?

On 2022-10-05 8:06 AM, JF Mezei wrote:
>
> Is there any document/information on how SpaceX manages its fleet of
> thousands of Starlink satellites ?

A very preliminary search didn't turn up anything. I did a short query
of my old professional society databases which would most likely have an
academic whitepaper on the topic, if any exist. Try the IEEE. This is
the kind of topic that would come up in a Spectrum article. But detailed
info would have to wait for a whitepaper. Eventually somebody will no
doubt write a book about it. The value (to me) of doing this search was
to come up once again with the proper acronym. In this case its TT&C
which stands for Telemetry, Tracking and Control.

So before I get into the weeds on some of the questions below, here is a
quick summary of two items of interest:

The folks at Carleton University (from your neck of the woods IIRC) have
been busy working some of the details of issues of importance in terms
of optimal spacing between Starlink satellites (satellite aka 'a bird'
in the vernacular). Here's a paper that might be of interest to you:

https://www.computer.org/csdl/proceedings-article/5gwf/2021/430800a493/1yEYKs7Dies

It will cost you some $ to get a copy but shouldn't be too bad.

Then there is this organization:

https://www.space-track.org/

They might be able to provide you with additional insight.
Below are my prejudices.

>
> Since the TLEs are crucial to the service (allowing retail antennas to
> point in right direction and for those newer satellites with the space
> lasers, know where to aim the laser to destroy ^H^H^H^H^H^H communicate
> with other satellites,
>
> how does SpaceX keep all the TLEs up to date? I assume radar tracking
> from ground is not realistic, or does that work for such small guys in
> the sky?
>
> Or is each satellite equipped with enough instruments to calculate its
> own TLE and transmit back to planet Earth? Can precise TLEs be done
> with GPS alone? would it need some star tracker on each satellite?
>
> How precise must the time of crossing ascending node be? aja, if ytou
> get GPS samples each 2 seconds would being 1 second off detecing when
> you crossed latitude 0 be a big no-no requiring calculation of
> estaimated time of crossing it based on the 2 nearest readings?
>
>
> Would it be correct to state that the only way for SpaceX to detect a
> lost satellite is when it doesn't say "hello" when it passes over a
> ground station? Typiocally, would lack of "hello" when passing over 1
> ground station trigger the alert, or would they wait for no "hellos"
> from a satelite over multiple stations?
I think this is a fair guesstimate. However, I'm also assuming there is
enough diagnostic processing going on with each Starlink bird to
indicate to ground control when it is having issues serious enough to
warrant intervention. Complete loss of a bird might prompt ground
control to send up a de-orbit signal 'in the blind' in the hopes of
taking a non-functional satellite out of orbit if it can still process
ground commands. There is even a passive re-orientation process that can
place the satellite in a configuration to increase its drag and hasten
its return. I presume the inverse of the 'duck' maneuver vis-a-vis the
atmosphere and it's orbital path.

https://www.spacex.com/updates/#sustainability

>
> I take it when SpaceX needs to send stuff to a satellite, it calculates
> which ground station will next have view on it and get that ground
> station to send the "execute order 666" command to the satellite as soon
> as it says "hello" ?

It could but why bother? Why not just transmit it to all ground stations
for up relay. (see below)

>
> Or would they send the command to all ground stations and the first one
> that sees the satellite sends the command, and report back to HQ that
> the command was sent succesfully and all other ground stations told to
> remove the command from their queue?
It could but why bother? It would depend upon how frequently such
commands are being sent. I know there are a lot of satellites in the
Starlink system, but commands can be sent fast. There are just a few
thousand today, maybe low ten thousand someday. No big deal with today's
communication speeds. The effect on trajectory may already be a function
of the data 'packet' that gets sent to ground control for each bird
already. GC can then pick apart the data for any particular bird if it
needs to. You want to minimize the back-and-forth on your TT&C network.
So a lot of this probably operates in half-duplex mode. So a command is
sent 'up' by broadcasting it to all ground stations to transmit up to a
particular bird. I think you are dead on about that message being in a
queue with in addition having a unique identifier. Stations would
transmit what they have in their queue for that bird. If the bird sees
that it has already processed that command it just ignores it and the
ground station removes it from its queue automatically. No extraneous
response needed. If GC needs to confirm they can always send a query
command. In my prejudice this is the way I'd do it. But it's just a
guess on my part.

I've dealt with non-satellite telemetry and control systems on a
small-scale and I can tell you such an approach works, scales well and
being a hybrid of half and full duplex in nature helps when you want to
'broadcast' control messages.

Hope this helps.

Dave

On 2022-10-10 16:58, David Spain wrote:

> It could but why bother? Why not just transmit it to all ground stations
> for up relay. (see below)

Consider when they send a software upgrade to a satellite. Each station
only has a limited window to send the software update package until the
satellite disappears from view. So it becomes interesting on how they do
this (likely split the software update into sequential bundles sent by
different groudn stations over time. Or send a quick command to
satellite "please fetch XXXX-1 through XXXXX-53 at your convenience" and
satellite makes request for each bundle as it passes over a ground station.

Separate question: SpaceX and T-Mobile recently announced use of its,s
licensed 1900 spectrum (PCS) to provide normal cellular service from
satellites worlwide.

Have they provided any information how spectrum licenced to T-Mobile
only in the USA would be used worldwide and conflict wioth licence d
users of those frequencies elsewhere (*cellular providers in
Canada/Mexico, military in Europe etc) ?

If those radios will be turned on only while over continental USA, I am
curious on how close to border the cone of coverage will be allowed to
get before radio is shut. With each satellite covering a circle of
about 2000km diametre on ground, it would mean that very little of USA
could be covered if they need to shut radio of each satellite the second
part of its footprint is outside of USA.

I really hate it when they make PR annoucements to create buzz when such
annoucnment makes no sense.

This morning, about 06:00, saw what I assume was a falling satellite.
very bright and large, with a trail and not that fast and the brightness
ended abruptly (likely combination of falling into darkness)
Definitely not a meteorite.

So this brings about a question:

When you manage a fleet of say 10,000 low orbit satellites, how do you
manage loss of a few per orbital plane? So you just order the remaining
SVs in that orbital plane to space themselves a bit more to provide for
even gaps between each? Or leave the gap left by the de-orbited satellite?

And how would SpaceX end up replacing these satsllites? Since it usually
launches over 50 satellites at a time, it isn't as if it could launch
just one satellite to be inserted in that plane at the right spot to
replace the fallen one. So what strategy will it use?

Are there statistics on how many per week de-orbit?

BTW, SpaceX annoucned it will now allow subscriptions in
Nunavut/Yukon/NWT, which likely means it has enough po;ar orbit
satellites with the space lasers function working so signals can reach a
base station in the south. Not long ago, it had announced it had done
so in Nunavik (northern Québec) with Government of QUébec agreeing to
subsidize the initial hardware purchase.

Are the space lasers "fixed" in that only satellites in the same orbinal
plane can talk to each other (which reduces need for mechanical movement
of laser and receiver since lasers can point forward and back and reach
the next/previous satellite in that orbital plane). or are they truly
moving and tracking satellites in other orbital planes? (and how would
such a satellite know to move its optical sensor to pointr at some other
satellite that is moving when there might be many that want to talk to
it at same time?

On 2022-11-03 7:03 PM, JF Mezei wrote:
> So this brings about a question:
>
> When you manage a fleet of say 10,000 low orbit satellites, how do you
> manage loss of a few per orbital plane? So you just order the remaining
> SVs in that orbital plane to space themselves a bit more to provide for
> even gaps between each? Or leave the gap left by the de-orbited satellite?
>
I suspect the answer to that question is yes, until you can get
replacement satellites returning into that plane to replace the gaps.
There is one website I know of that can show you what Starlink
satellites are available over any given spot on Earth. What you notice
are that not all the satellites are at the same altitude. And their
coverage footprint varies with altitude. At 500km or so you get a pretty
wide footprint but have to wait a bit to gain coverage. At 300km or so a
narrower footprint but its orbits into and out of view quicker. At this
time there are enough satellites on-orbit that often the coverage
circles overlap. I suppose your dish software make a determination which
one to use, maybe based on signal strength or other factors.

At some point in time I'm sure someone will write a book or publish a
whitepaper on how this works.

> And how would SpaceX end up replacing these satsllites? Since it usually
> launches over 50 satellites at a time, it isn't as if it could launch
> just one satellite to be inserted in that plane at the right spot to
> replace the fallen one. So what strategy will it use?
>

I'd guess the same one they are using now. Which is to continue
launching sat "trains" that place satellites is specific planes at
launch and the uses the on-board propulsion to do the spacing. Takes
less delta-V and thus reduces the amount of propellant needed, insuring
a longer on-orbit life.

> Are there statistics on how many per week de-orbit?
>
I haven't seen any. Not from SpaceX anyway.

>
> BTW, SpaceX annoucned it will now allow subscriptions in
> Nunavut/Yukon/NWT, which likely means it has enough po;ar orbit
> satellites with the space lasers function working so signals can reach a
> base station in the south. Not long ago, it had announced it had done
> so in Nunavik (northern Québec) with Government of QUébec agreeing to
> subsidize the initial hardware purchase.
>
> Are the space lasers "fixed" in that only satellites in the same orbinal
> plane can talk to each other (which reduces need for mechanical movement
> of laser and receiver since lasers can point forward and back and reach
> the next/previous satellite in that orbital plane). or are they truly
> moving and tracking satellites in other orbital planes? (and how would
> such a satellite know to move its optical sensor to pointr at some other
> satellite that is moving when there might be many that want to talk to
> it at same time?

I used to think that (in plane only). But it might not be necessary. In
plane is obviously the easier to imagine. If the lasers operate in both
forward (pro-grade) and reverse (retro-grade) in-plane, they should be
able to maintain nearly constant contact with their nearest neighbor.
Cross plane could be used depending upon the strength and dispersion of
the laser beam to maintain contact for short periods of time with
crossing satellites. Enough time perhaps to send many packets worth of
data. It would come and go but as more satellites are orbited the gaps
between would diminish somewhat. I'm not convinced the beam dispersion
is so tight that it requires active steering. Seems too complicated to
me. But I'd love to see a whitepaper on the topic.

Intermittent communication is a built-in feature of Starlink so having
it between satellites as well as between ground stations is probably no
big deal for the Starlink software.

David

On 2022-11-05 5:41 PM, David Spain wrote:
> I'd guess the same one they are using now. Which is to continue
> launching sat "trains" that place satellites is specific planes at
> launch and the uses the on-board propulsion to do the spacing. Takes
> less delta-V and thus reduces the amount of propellant needed, insuring
> a longer on-orbit life.

The reason I mention 'trains' is if consult the Starlink tracker website
I previously mentioned, you'll often see "new" satellites in a 300km or
so 'parking' orbit and very tightly spaced. No coverage circles are
usually associated with these satellites so I presume they are not yet
in-service. They are clearly co-orbiting in the same plane.

David

On 2022-11-05 17:41, David Spain wrote:

> Intermittent communication is a built-in feature of Starlink so having
> it between satellites as well as between ground stations is probably no
> big deal for the Starlink software.

Northern Canada was just opened to sales. There are no base stations
there so service requires that whatever satellite is above you is able
to talk to another satellite and so on until there is one over a base
station. So intermittent inter-satellite links would't allow such service.

On 2022-11-07 7:28 PM, JF Mezei wrote:
> On 2022-11-05 17:41, David Spain wrote:
>
>> Intermittent communication is a built-in feature of Starlink so having
>> it between satellites as well as between ground stations is probably no
>> big deal for the Starlink software.
>
>
> Northern Canada was just opened to sales. There are no base stations
> there so service requires that whatever satellite is above you is able
> to talk to another satellite and so on until there is one over a base
> station. So intermittent inter-satellite links wouldn't allow such service.

As I said in my previous post, intermittency would only be between plane
crossing satellites, not co-orbital nearest neighbors. So yes, co-planar
satellites could forward to nearest neighbors until one is in range of a
base station.

Also as you get closer to the poles, the density of plane crossing
satellites will be much higher. So perhaps giving even more options of
which base stations get chosen or perhaps even shared.

Even if strictly co-planar, which plane is employed will also vary
depending upon which satellite is overhead and servicing your dish at
the time.

Since Northern Canada seems of some interest to you, are you considering
switching to Starlink?

Dave

On 2022-11-07 7:28 PM, JF Mezei wrote:
> On 2022-11-05 17:41, David Spain wrote:
>
>> Intermittent communication is a built-in feature of Starlink so having
>> it between satellites as well as between ground stations is probably no
>> big deal for the Starlink software.
>
>
> Northern Canada was just opened to sales. There are no base stations
> there so service requires that whatever satellite is above you is able
> to talk to another satellite and so on until there is one over a base
> station. So intermittent inter-satellite links would't allow such service.
As I said in my previous post, intermittency would only be between plane
crossing satellites, not co-orbital nearest neighbors. So yes, co-planar
satellites could forward to nearest neighbors until one is in range of a
base station.

Also as you get closer to the poles, the density of plane crossing
satellites will be much higher. So perhaps giving even more options of
which base stations get chosen or perhaps even shared.

Even if strictly co-planar, which plane is employed will also vary
depending upon which satellite is overhead and servicing your dish at
the time.

Since Northern Canada seems of some interest to you, are you considering
switching to Starlink?

Dave