Starlink : General Discussion - Thread 2

[Barley]

3) The comment thread is dedicated specifically to the feeder link, since I have seen in  Internet the statements of individual commentators that StarLink uses 2 feeder lines at the same time, each of the 2 antennas with its own gateway, which I consider impossible.

The only reason you gave to support this being impossible was the need to do handoffs. Which isn't actually a reason, as the user link also needs to do handoffs, and these can coincide. With multiple satellites always visible to every user, they don't need continuous uptime on every satellite. Giving up 1 second or so of transmit time every minute in order to double the uplink throughput is a huge overall increase.

Stagger the transfers.  Unless it takes as long to repoint an antenna as it takes to pass over a ground station you gain some capacity without any increase in dead time.  That does require managing traffic for quality of service, but Starlink has to do that anyway, and it's easier if you have more capacity.

[SpaceCadet1980]

well, let's analyze it separately
 Each StarLink satellite changes its gateway at least once every  2...4 minutes 10 seconds (max time). In order to change the gateway, the Ka-band antenna on sat must turn back  up to  130 degrees.

First of all, I asked you to reply to the people who had already pointed out the ways you are wrong. By replying to me instead and including arguments that were already shutdown, you are rudely ignoring both me and them which is simply uncivil behavior.

Second, while I have not worked out the math for Starlink, LEO satellites can easily have 15-20 minute downlink windows, so 2-4 minutes should not be typical.

Have you ever seen how a SeaTel or Intellian marine antenna is tuned? How many time they need to turn? (in seconds?) Please "quickly" is not answer , Answer is X seconds..

You have already been given numeric answers by other people. It should easily be less than the 15 second window that people have observed Starlink routing configurations to happen in. The antennas you linked are irrelevant. I know of satellites where the entire satellite could reorient at speeds that approach the needed speed here. For the relatively small dishes on Starlink, claiming they can't possibly move fast enough is a level of concern trolling equivalent to calling the SpaceX engineers idiots.

After prime pointing "in direction" , if it receives the signal from the beacon on GW begin  precise pointing both (!!!) Antennas - on Sat and on GW. After that, as fine pointing is completed, the Antenna  and link Sat-GW 2 is ready for operation with nesessary SNR, server on the gateway can send a message to the NMС that GW2 is ready to serve satellite X and the corresponding cells NN  was served from  Gateway 1, now it must be served by Gateway 2.
At the same time, the NMS should redirect traffic for users from POP 1 to which GW1 is connected to POP2 to which GW2 is connected ..

And only after all these processes are implemented, the operating configuration of the full StarLink Network "Gateway - satellite- cell" for the next 15 seconds is created, service through the gateway 2 can begin. A temporary break for these processes is at least 15 seconds, and maybe more.
(I would personally take an additional 15 seconds as a margin of time so that if it is not possible to сonnect to GW2. will be time to switch to GW3)

As you have already been told by people who have demonstrated more comprehension of this subject than you, that long paragraph stating the interactions can be less than 1 second in practice.

That is why OneWEB Engineers (who created early O3B NGSO Network) have chosen the method described above to ensure the 100% continuity of the service.

Again, One Web is a fundamentally different architecture in many ways including their overall business model. Stop bringing them up it is off topic and irrelevant.

If the Space X Engineer believed that it was permissible to interrupt the service for the user every  4 minutes for 15-30 seconds per, they would have been fired on the same day.

Each satellite losing just half capacity for 15 seconds every 15 minutes would not interrupt any single user's service for even 10 ms. It would be handled just like any other handoff and another satellite could pick up the slack if needed.

And I am happy to read any SpaceX document that can confirm your words each Sat will be served from 2 GW in one moment.

Burden of proof. You made the assertion that the satellites will waste nearly half of their capacity to handle a non-issue. You provide actual evidence for your claim, but first apologize to everyone who already gave you numbers that you decided to rudely ignore.

P.S. I have a question Have you ever worked in Telecom in the field of service quality? or Satellite Networks?
What service availability should StarLink provide in your opinion?

I am not in a position where discussing my background is appropriate, it is also irrelevant, because providing it would only serve the purpose of making the argument from authority fallacy. Seeing as you have yet again failed to acknowledge the facts I provided about phased arrays and the consequences of those facts, which I provided a citation for, you have repeatedly cited non-authoritative and irrelevant sources, and in this post you have ignored numbers given to you by people including Nomadd, who I know from his past posts to have relevant practical experience, the conclusion I come to is that nothing you assert should be taken as valid without independent relevant confirmation. This is the result of your actions, my word too is only as good as the knowledge I have demonstrated. My posts are in agreement with the others here who have demonstrated knowledge of the subject, yours are not.

The service availability of Starlink as designed is quite clearly 100% with at least single fault tolerant once enough satellite coverage is available and either enough gateways or ISLs. I am guessing you meant to ask something different though I am not sure what.

[oldAtlas_Eguy]

The answer comes from the evaluation of the complete system vs that of a single sat. In the environment where there is >2000 sats there will be several sats that have both locations in their Ku and Ka band footprints. You only need one sat that can talk to both at one time for a 16Gb/s throughput comm link. The Gateway would likely have as many as 4 Ka band steerable antennas and more likely 6 or more. From this type of system level look. The Ku and Ka antenna on the sats may not change their "spots" at all. It only requires for frequency reuse for each antenna spot to not overlap with others of the same band/frequency. An interesting Note here is that this system level design view would also work for One Web with only 600 sats. These are not GEO sats in which the number of sats possible to connect to is less than 4 likely only 1. But a situation where the number is >4 if not >>10. Many more sats to pick from than available Ka band antennas in the Gateways. Hence the use of multiple cheaper Ku band UTs (> than the total number of Ka antennas) to connect to many sats simultaneously to get detailed status, orbit throughput capability of the possible sat to do a connect through.

This is the way to look at it. Macro not Micro.

[daavery]

from multiple aerial photos the starlink base stations have 9 domed steerable parabolic dishes in a 3x3 grid

[vsatman]

//Second, while I have not worked out the math for Starlink, LEO satellites can easily have 15-20 minute downlink windows, so 2-4 minutes should not be typical.

So can you calculate first? and then you will write nonsense about 15-20 minutes for StarLink?

//to calling the SpaceX engineers idiots.

This is a good point. So we have 2 options for organizing a feeder channel with a speed of 20 Gbps.

Option A Proven many times, as OneWeb engineers did.  to use one feeder channel at a time from a single 1.5m antenna on the Gateway with a maximum throughput of 32 Gb and a typical 25+ Gbps

Option B, which, according to your words, was implemented by SpaceX engineers - use simultaneously 2 feeder lines and 2 Antennas on different gateways, each with a maximum capacity of 32 Gbps and a typical 25+ Gbps. In total, they transmit either 25 or 50 Gbps to the satellite, although 20 gbps are needed. At the same time, there is some kind of magic box on the satellite for redistributing traffic at the packet level on board (a task that no manufacturer in the world has solved so far for a serial satellite ). Naturally, this unique magic box increases the cost of the satellite, requires additional power, and reduces the reliability of the satellite. At the same time, no one knows about the magic box yet, and there is no mention in any SpaceX document, although its appearance would be a sensation and an achievement much greater than laser channels.

From my point of view, the Engineer who proposed option B instead of A can be called an idiot. But note that you called SpaceX engineers idiots, not me ..

[vsatman]

from multiple aerial photos the starlink base stations have 9 domed steerable parabolic dishes in a 3x3 grid

Always 9, but sometimes in one or 2 rows

[launchwatcher]

At the same time, there is some kind of magic box on the satellite for redistributing traffic at the packet level on board (a task that no manufacturer in the world has solved so far for a serial satellite ).

The hardware necessary can be found in commodity network switch ASIC's - the chip parses enough of each packet header to form a flow identifier then uses that to pick the outbound link for a flow.   Packets within a flow all go on the same outbound link and remain in order.   Read up on link aggregation and equal-cost multipathing. 

[SpaceCadet1980]

//Second, while I have not worked out the math for Starlink, LEO satellites can easily have 15-20 minute downlink windows, so 2-4 minutes should not be typical.

So can you calculate first? and then you will write nonsense about 15-20 minutes for StarLink?

You are the one who made up a number based on nothing. I gave a number based on experience working LEO satellites. You can either choose to learn something about the subject and do the math yourself, or you can stop making baseless assertions.

//to calling the SpaceX engineers idiots.

This is a good point. So we have 2 options for organizing a feeder channel with a speed of 20 Gbps.

And the rest of your analysis is all based on this 20 Gbps number which is an old number for Starlink throughput. As I have already mentioned, there is no reason to believe that this is still true for the latest Starlink satellites, especially since ISLs mean that not all data going to or from a ground station will be tied to user beams on the satellite.

Speaking of user beams on the satellite, you are still replying to me while ignoring that the subject of my posts to you was originally about your baseless claims for the number of possible simultaneous user beams. Since you are refusing to reply, should I take this as an implicit admission that you were wrong and have no clue what you are talking about?

From my point of view, the Engineer who proposed option B instead of A can be called an idiot. But note that you called SpaceX engineers idiots, not me ..

As the above posters have pointed out, you referred to common networking equipment which is the basis for the modern internet as magic in your description of option B. Being able to do onboard packet routing is trivially required for ISLs to be useful (see my reply to you in the Starlink Markets thread) Making such as claim goes beyond ignorance, and to then try to use that nonsense to claim I am calling the SpaceX engineers idiots is just hateful.

[vsatman]

So can you calculate first? and then you will write nonsense about 15-20 minutes for StarLink?

You are the one who made up a number based on nothing. I gave a number based on experience working LEO satellites. You can either choose to learn something about the subject and do the math yourself, or you can stop making baseless assertions.

Oh, let's analyze this nonsense, so the 2 most banal questions:
1) What is the orbital period of the StarLink Satellite around the Earth?
2) What is the diameter of the circle in the sky that the gateway antenna sees?
Anyone with even the slightest interest in StarLink should know these numbers. Next 2 calculations at the elementary school level.
a great way to understand what is behind your words and what is your real experience in satellite communications :-), and is there any point in reading further everything that you write ..

[zubenelgenubi: Fixed quotes. Please proofread your reply posts before posting.]

[SpaceCadet1980]

So can you calculate first? and then you will write nonsense about 15-20 minutes for StarLink?

You are the one who made up a number based on nothing. I gave a number based on experience working LEO satellites. You can either choose to learn something about the subject and do the math yourself, or you can stop making baseless assertions.

Oh, let's analyze this nonsense, so the 2 most banal questions:
1) What is the orbital period of the StarLink Satellite around the Earth?
2) What is the diameter of the circle in the sky that the gateway antenna sees?
Anyone with even the slightest interest in StarLink should know these numbers. Next 2 calculations at the elementary school level.
a great way to understand what is behind your words and what is your real experience in satellite communications :-), and is there any point in reading further everything that you write ..

[zubenelgenubi: Fixed quotes. Please proofread your reply posts before posting.]

Orbital Period for LEO satellites in 90 minutes as a rule of thumb. A precise number for Starlink is not something that someone would know without doing the calculation if they haven't been directly working with that data recently. (formula is in the SMAD textbook sitting on my desk, but I am not going to do the math for you for your claim.)

Your second question is poorly defined, more relevant is "What is the angle relative to the center of the Earth from the ground station to a satellite, when the line from the satellite to the ground station is at the ground station's minimum elevation angle from the ground?" This is not something that anyone would know off the top of their head (again unless they work on Starlink on a technical level), and I'd bet many high school students would struggle with the relevant trigonometry. (Inputs to this are Starlink altitude and the ground elevation angle) Again, you made the claim, you do the math.

I already gave my input with an estimate based on systems I have worked with, Starlink is lower altitude, so if it is maybe half of the range I provided that would be reasonable, as I never claimed that number was correct for Starlink, just that your provided number was significantly too low.

And to be clear here, you have already demonstrated that your knowledge is severely lacking on essentially all relevant subjects, from the number of simultaneous beams a phased array can handle (essentially arbitrary) to the fact that packet routing is common and not magic, etc. Never mind the incoherence and/or irrelevance of your second question, I'll just assume that is just the language barrier again. You can't even seem to admit to these mistakes, so you have already demonstrated that there is no point in reading anything you say, I am doing it only to correct the nonsense you are spouting. Please don't pretend that you are actually reading what I have written. You have repeatedly ignored what I and others have written, only to change the topic every time you are proven wrong on something. You asking questions to judge my understanding simply is not relevant at this point, multiple people here have provided evidence pointing out ways you are wrong.

[Chris Bergin]

New thread. Vsat, be civil or lose your posts.

https://forum.nasaspaceflight.com/index.php?topic=55795.0