[Starlink] "Interesting set of developments with Starlink. Musk says they will support "international aid orgs" in Gaza, Israel now says they will use "all available means" to stop SpaceX from doing so.

[Ulrich Speidel]

On 11/11/2023 6:09 pm, Alexandre Petrescu via Starlink wrote:
>
> I want to say that I think this hexagon is an imaginative idea of the
> GUI designer.  I think it does not correspond to reality.  I am not sure
> about even the most basic fact such as the dimension of the hexagon, or
> of a more circular 'spot' radius.
Well it's the basis on which SpaceX will sell you a fixed subscription 
or not, so it's a bit more than just a GUI designer's fancy.
>
> It was the case like that with earlier maps of cellular network
> deployments.  The name 'cell' itself that comes from it - it's a
> hexagon, like in a honey pot.  In practice, no cellular network
> deployment I am aware of has cells of that kind of precise shape. The
> base stations themselves are not following such precise patterns.   The
> precise forms of coverage shapes can not be given by operators because
> it is unique, difficult to calculate, and depends on many landscape
> factors and other propagation conditions.  What can be given is the type
> of antennas, their precise placement and orientation.  That is public
> info of cellular systems in some countries.

Yes, but I wouldn't think of the Starlink hexagons as "cells" in the 
cellphone sense. In the cellphone sense, you'd have a base station per 
cell - and some seem to think that Starlink has something like a "spot 
beam per cell". But that's clearly not what it is.

Starlink cells are quite obviously predominantly a tool to control user 
density on the ground. We know well that Dishy orients itself to the 
area of the sky where it can expect to see the largest number of 
satellites without falling foul of GSO protection rules where 
applicable. Dishy then associates with one of them at a time for period 
that are multiples of 15 second intervals (we know that from the 
obstruction maps available from Dishy via grpc). We also know that the 
capacity we get via these associations fluctuates along with the 15 
second intervals.

These are all hallmarks of a burst slot based system where each 
satellite handles a set of time and frequency slots (TDM+FDM), such that 
a combination of a periodic time slot, a frequency channel and perhaps a 
number of other parameters (spreading code, polarisation, spatial 
multiplexing through beamforming at the satellite) defines a channel 
through which Dishy talks to the satellite or receives from the 
satellite. This sort of technology has been around for decades (see GSM 
mobile comms). During your slots, the satellite your Dishy is associated 
with will project a beam towards you.

Fluctuations in the capacity are a result of a Dishy having more or 
fewer of these slots assigned during subsequent 15 second intervals. If 
a satellite picks up more users for the next interval, the number of 
slots it can make available to your Dishy goes down. If it sheds users 
as it moves along but you hang on, then you get a few more slots, until 
either the satellite picks up more user or you change satellite.

For a scheme like this to work, you need to ensure that the number of 
slots that the visible satellites can offer to their users on the ground 
works out to an acceptable average minimum number of slots per user at 
all times. As the number of slots per satellite is likely fixed (at 
least for the same generation satellite), that puts a limit on the 
number of users within view on the ground. This is where the hexagons 
come in - they help ensure that the user density doesn't grow to a level 
anywhere where a Dishy would struggle to get enough slots.

I'd presume that the size of the hexagons was chosen to reflect the 
ability of the beamformers on the satellites to resolve a locality on 
the ground (no need to go for higher resolution then). I'd also presume 
that the number of fixed users allowed per hexagon would depend a bit on 
geographical latitude, visible satellite density and load contributions 
from a location's surroundings (Colorado farmers would likely see more 
of those from their neighbours down the road than folks on Rapa Nui). 
Roaming subscribers aren't guaranteed the same service levels (read: 
their number of slots is allowed to dip further than those of fixed 
subscribers), but as they can't be told where to be, SpaceX uses pricing 
to control user density indirectly.

>
> With starlink antennas there is no authoritative public info (from
> starlink) about the precise orientation and types of antennas of the
> sats.  The reported positions of sats are rather irregular - much more
> irregular than that precise shape of hexagons shown in the photo.
Yes - see above.
>
> The placement of teleports is also unknown, but speculated by end users.
In some cases, their locations are precisely known from their spectrum 
licenses.
>
> The info about satellite tracking, their precise situation: I still need
> to find out where that info comes from more precisely and how is it
> obtain (is it reported by sats or is it  other cameras/radars that
> 'range' each one of them, or is it simply speculated from an initial
> plan of trajectory; and what is the delay between the actual fact and
> what is seen on GUI: seconds, minutes or hours delays).

https://celestrak.org/NORAD/elements/index.php

-- 

****************************************************************
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/
****************************************************************