4 ms·

well 3 degrees wide means at best ~28 square degrees on the sky (pi * 3 * 3). But the problem is the other way. It looks like Starlink has 7 GHz total downlink

by cozzyd 4y ago

well 3 degrees wide means at best ~28 square degrees on the sky (pi * 3 * 3). But the problem is the other way. It looks like Starlink has 7 GHz total downlink bandwidth from satellite to customers. A cell may be ~300 sq km, but no way you can pipe all that bandwidth in adjacent cells due to interference (or... if you did, you'd probably lose in spectral efficiency). If we assume something like 10 Gbps/cell total downlink bandwidth (based on two polarizations, 1/4 the total bandwidth and 3 bits/Hz efficiency), then you can maybe serve... 500 people / cell reasonably? How many people live in places with fewer than a few people/km^2? Only Alaska averages below that in the US.

mlyle 4y ago
> well 3 degrees wide means at best ~28 square degrees on the sky (pi * 3 * 3). If you use a diameter for a circle area, you need to divide by 4, to get like ~7. I was just going to do 3*3 = 9. > It looks like Starlink has 7 GHz total downlink bandwidth from satellite to customers. A cell may be ~300 sq km, but no way you can pipe all that bandwidth in adjacent cells due to interference Both the downlink and the uplink are directional. Yes, there's some power lost in side lobes, but.. So, if you have multiple satellites overhead, you can have a mosaic pattern where satellite A illuminates a hex, and then other satellites illuminate its neighbors. And even though some energy spills over, it's strongly rejected by the receivers because it's coming from the completely wrong direction. > How many people live in places with fewer than a few people/km^2? In addition to the incorrect pessimism about downlink bandwidth (also, one should expect ~5bits/Hz)... You're not taking into account any reasonable capacity factor, and you're assuming each service is one person. You're also assuming that you sell through to 100% of the populace and there's no one in the cell served by other technologies. If 95% of a suburban area can get FTTH, cable, or reasonable LTE, that still leaves 5% as possible customers for Starlink to fill in. > Only Alaska averages below that in the US. OK, but people are concentrated into cities, leaving a whole lot of land area with concentrations below this even in states with much higher average population density.
- shaklee3 4y ago
  5bps/Hz is way overestimating their spectral efficiency.
  - mlyle 4y ago
    64 QAM with FEC is specified as the original modulation in regulatory filings. This predicts 4.5-5.5 bits/Hz.
    - shaklee3 4y ago
      Yes, and dvbs-2 also supports very high modulation and code rates. but they're never used in practice since PSD limits don't allow you to transmit that much due to interference on other systems. starlink is not first in line for most of their spectrum, so they're abiding by direct spectral mask limits. I would guess they're at 16APSK at the highest.
      - mlyle 4y ago
        > so they're abiding by direct spectral mask limits Yes, obtained by channelization and OFDM, not by limiting modulation. Reverse engineering of the 10 to 12.7GHz band has shown a mixture of 16-QAM and 64-QAM frames. There's a bit of periodic 4-QAM too, which seems to be for acquisition and synchronization. Even 16-QAM can climb basically to 5 bits/Hz.
        shaklee3 4y ago
        Please show a link where this was reverse-engineered. I'd be curious to see it.
jacquesm 4y ago
> It looks like Starlink has 7 GHz total downlink bandwidth from satellite to customers. I think you meant 7 Gbps; that only works if the satellite itself has that much inbound, which may not be the case if the satellites around (for sat-to-sat links) it or the base station that it sees are maxed out.
- mlyle 4y ago
  7GHz is correct; 2GHz of X band and 5GHz of Ka/V band for downlink to subscribers. Multiply by spectral efficiency-- perhaps 5 bits/Hz. > which may not be the case if the satellites around (for sat-to-sat links) No real sat-to-sat links yet. The exact specifications aren't disclosed, but they'll likely be 100gbps symmetric links. > the base station that it sees are maxed out. You can expect spectral efficiency for the uplinks to be better because of big antennas, and for them to be provisioned with no problem illuminating all the satellites overhead.
  - jacquesm 4y ago
    Sure, but spectrum isn't infinite, at some point it is maxed out, especially the sat-to-sat links, which effectively halve the available bandwidth (assuming just one hop).
    - mlyle 4y ago
      Directionality makes it less inclined to be "maxed out". If you have infrastructure ground stations, and they're built for 5 satellites overhead, you can get 5 times the spectral density total throughput if there's 5 satellites overhead. > which effectively halve the available bandwidth I disagree (and I'm not sure where this idea is coming from-- do you think they use the same spectrum or something?) I think they effectively "add" bandwidth for the system by allowing satellites that do not have saturated uplinks to "share" capacity. E.g. a satellite which is mostly over the ocean but can see a ground station can now be more useful.
      - jacquesm 4y ago
londons_explore 4y ago
Remember that if they are starting to get near capacity limits, they always have the option to make the antennas on the satellite larger, making the beams narrower, and reduce those 300 sq km cells down to say 30 sq km. In the current era of phased array antenna prices falling sharply, and assuming they're willing to go for a folding antenna, I think they could probably easily get a 10x antenna area boost without much extra cost.
- mlyle 4y ago
  If you have 10x the antenna area, you need 10x the antenna elements with appropriate phasing... else you suffer from the thinned array curse and lose lots of power in side lobes. https://en.wikipedia.org/wiki/Thinned-array_curse https://en.wikipedia.org/wiki/Thinned-array_curse

by cozzyd 4y ago

mlyle 4y ago

> well 3 degrees wide means at best ~28 square degrees on the sky (pi * 3 * 3). If you use a diameter for a circle area, you need to divide by 4, to get like ~7. I was just going to do 3*3 = 9. > It looks like Starlink has 7 GHz total downlink bandwidth from satellite to customers. A cell may be ~300 sq km, but no way you can pipe all that bandwidth in adjacent cells due to interference Both the downlink and the uplink are directional. Yes, there's some power lost in side lobes, but.. So, if you have multiple satellites overhead, you can have a mosaic pattern where satellite A illuminates a hex, and then other satellites illuminate its neighbors. And even though some energy spills over, it's strongly rejected by the receivers because it's coming from the completely wrong direction. > How many people live in places with fewer than a few people/km^2? In addition to the incorrect pessimism about downlink bandwidth (also, one should expect ~5bits/Hz)... You're not taking into account any reasonable capacity factor, and you're assuming each service is one person. You're also assuming that you sell through to 100% of the populace and there's no one in the cell served by other technologies. If 95% of a suburban area can get FTTH, cable, or reasonable LTE, that still leaves 5% as possible customers for Starlink to fill in. > Only Alaska averages below that in the US. OK, but people are concentrated into cities, leaving a whole lot of land area with concentrations below this even in states with much higher average population density.

shaklee3 4y ago
5bps/Hz is way overestimating their spectral efficiency.
- mlyle 4y ago
  64 QAM with FEC is specified as the original modulation in regulatory filings. This predicts 4.5-5.5 bits/Hz.
  - shaklee3 4y ago
    Yes, and dvbs-2 also supports very high modulation and code rates. but they're never used in practice since PSD limits don't allow you to transmit that much due to interference on other systems. starlink is not first in line for most of their spectrum, so they're abiding by direct spectral mask limits. I would guess they're at 16APSK at the highest.
    - mlyle 4y ago
      > so they're abiding by direct spectral mask limits Yes, obtained by channelization and OFDM, not by limiting modulation. Reverse engineering of the 10 to 12.7GHz band has shown a mixture of 16-QAM and 64-QAM frames. There's a bit of periodic 4-QAM too, which seems to be for acquisition and synchronization. Even 16-QAM can climb basically to 5 bits/Hz.
      - shaklee3 4y ago
        Please show a link where this was reverse-engineered. I'd be curious to see it.

jacquesm 4y ago

> It looks like Starlink has 7 GHz total downlink bandwidth from satellite to customers. I think you meant 7 Gbps; that only works if the satellite itself has that much inbound, which may not be the case if the satellites around (for sat-to-sat links) it or the base station that it sees are maxed out.

mlyle 4y ago
7GHz is correct; 2GHz of X band and 5GHz of Ka/V band for downlink to subscribers. Multiply by spectral efficiency-- perhaps 5 bits/Hz. > which may not be the case if the satellites around (for sat-to-sat links) No real sat-to-sat links yet. The exact specifications aren't disclosed, but they'll likely be 100gbps symmetric links. > the base station that it sees are maxed out. You can expect spectral efficiency for the uplinks to be better because of big antennas, and for them to be provisioned with no problem illuminating all the satellites overhead.
- jacquesm 4y ago
  Sure, but spectrum isn't infinite, at some point it is maxed out, especially the sat-to-sat links, which effectively halve the available bandwidth (assuming just one hop).
  - mlyle 4y ago
    Directionality makes it less inclined to be "maxed out". If you have infrastructure ground stations, and they're built for 5 satellites overhead, you can get 5 times the spectral density total throughput if there's 5 satellites overhead. > which effectively halve the available bandwidth I disagree (and I'm not sure where this idea is coming from-- do you think they use the same spectrum or something?) I think they effectively "add" bandwidth for the system by allowing satellites that do not have saturated uplinks to "share" capacity. E.g. a satellite which is mostly over the ocean but can see a ground station can now be more useful.
    - jacquesm 4y ago

londons_explore 4y ago

Remember that if they are starting to get near capacity limits, they always have the option to make the antennas on the satellite larger, making the beams narrower, and reduce those 300 sq km cells down to say 30 sq km. In the current era of phased array antenna prices falling sharply, and assuming they're willing to go for a folding antenna, I think they could probably easily get a 10x antenna area boost without much extra cost.

mlyle 4y ago
If you have 10x the antenna area, you need 10x the antenna elements with appropriate phasing... else you suffer from the thinned array curse and lose lots of power in side lobes. https://en.wikipedia.org/wiki/Thinned-array_curse https://en.wikipedia.org/wiki/Thinned-array_curse