8 ms·

20 Gbps per satellite, according to the FCC filings. Which isn't that great.

tim58 7y ago
So theoretically the more remote a location the faster the starlink will be because it is shared by fewer people. Which is the exact inverse for wired infrastructure.
- jandrese 7y ago
  Except that the more remote your location the further you likely are from the nearest ground station, which means more hops through the satellite network before you get back down to the ground, higher latency, and reduced total throughput of the system. The system Starlink is trying to beat is BGAN and regional Ku/Ka providers. If you're expecting them to beat the pants off of your terrestrial wired service you're probably going to be disappointed, unless you're on the edge of the range of some old neglected AT&T DSL line or still trapped on dialup. Even then it's almost certainly going to be much more expensive but at least it will have a sizeable speed advantage.
  - paulsutter 7y ago
    You want to downlink near the server, not near the client. Packets will move 40% faster through vacuum than through fiber.
    - wmf 7y ago
      Except if the terrestrial fiber backbone is cheaper and less congested you want to downlink near the client.
    - _flgt 7y ago
      Each RF hop is going to have more latency than fiber for RF physical layer processing. Forward error correction intentionally uses large frames to spread out burst errors.
      - Diederich 7y ago
        That's interesting, can you expand on that? Thanks.
      - paulsutter 7y ago
        Sounds speculative. Given link bandwidths and likely error rates, it seems better to allow the higher layers to retransmit than to optimize for correcting burst errors. Especially given Starlink's goal to control latency.
        ianai 7y ago
        Have they given a range on latency goals? Ie seconds, tenths of seconds, milliseconds, etc? I’m seeing figures of 1 million$ per satellite. Over 10 years that’s roughly 100k/month in mortgage or around 1k customers at 100$/mo.
        _flgt 7y ago
        I haven't seen a link budget. Do you have a link for one about the "likely error rates"? My guess is Starlink is going to need FEC for Doppler effects and multipath in urban environments. Maybe they will be able to overcome multipath with modulation. But Doppler shift will be significant with LEO. The latest 802.11n standards have been increasing physical layer frame sizes, not decreasing them. I don't know of any modern wireless standard that doesn't use FEC. Coding gain is significant in most link budgets I have seen. The last satellite RF downlink I worked with was for JPSS-1 in LEO nearly 6 years ago. There was a (255, 223) Reed-Solomon code with an interleave depth of 4 frames. So you effectively had to receive 1024 bytes, decode, and you got 4x 223 bytes worth of information. While latency wasn't a major concern, this scheme seemed pretty standard to me. If you don't believe me, the spec is public information: https://www.star.nesdis.noaa.gov/jpss/documents/CDFCB/GSFC_474-00001-07-01_CDFCB_External_Vol.7-1_JPSS_Downlink_Data_Formats__D34862-07-01_.pdf https://www.star.nesdis.noaa.gov/jpss/documents/CDFCB/GSFC_4... I'm sure SpaceX is trying really hard to control latency - so it doesn't get out of hand! One set of RF processing here, another over there, each with their own filters and decoding algorithms and then you have an extra 100 ms of delay in the system.
    - jandrese 7y ago
      The more traffic routes through the satellites vs. the ground the more you'll be competing for that limited sat-to-sat bandwidth and the more queuing delays you will suffer. You want to get the bits off of the birds and onto the ground ASAP if you want to maintain consistent latency figures. This was a big problem with Iridium. With only one ground station available your latency figures were all over the place thanks to the complicated routing that every packet had to go through.
  - tbabb 7y ago
    > the more remote your location the further you likely are from the nearest ground station I thought downlinks were going to be pizza box-sized antennae. Why would it be far away from you? That could live on your roof.
Paul-ish 7y ago
Netflix claims an UHD stream is 25 Mbps. This is probably the most demanding use case for your average consumer, but it may not be a rare use case in the near future. Each satellite can support 800 UHD streams. That doesn't seem sufficient if each satellite covers a large geographic area. This suggests to me a low data cap and/or rate, perhaps comparable to DSL.
- mastax 7y ago
  12,000 satellites is the current FCC approved launch figures, I believe. 20Gbps per satellite but I don't know if that's full duplex or how much is divided between terminal/gateway, so lets divide that in half. For simplicity let's say they only sell consumer-grade 100Mbps connections, which are oversubscribed 100x as is typical. Divide the possible number of customers by 3 to account for uneven geographic distribution and other factors. 12000 * 20Gbps / 2 * 100 / 100Mbps / 3 = 40,000,000 customers A service like that could sell for a lot in rural North America, but maybe not to 40m customers. Let's assume most of those customers are from lower income countries or from places with more competitive ISP environments. Maybe $20/mo average revenue per customer. That's $800m a month. That already sounds compelling to me, before considering the potential for more or higher bandwidth satellites, or for higher margin commercial or government contracts.
  - ianai 7y ago
    Another calculation is revenue per satellite per customer served. Would 100k per month per satellite finance these things long term?
    - Paul-ish 7y ago
      This seems like the important metric.
  - sneak 7y ago
    I think your simple calculation assumes relatively even geographical subscriber distribution, which absolutely is not the case. Just rough estimating, assuming every other figure is approximately correct, you’ll probably have to cut that by 50-75% based on people’s natural tendency to live near other people. A few HFT customers will probably finance the whole thing, tbh.

So theoretically the more remote a location the faster the starlink will be because it is shared by fewer people. Which is the exact inverse for wired infrastructure.

jandrese 7y ago
Except that the more remote your location the further you likely are from the nearest ground station, which means more hops through the satellite network before you get back down to the ground, higher latency, and reduced total throughput of the system. The system Starlink is trying to beat is BGAN and regional Ku/Ka providers. If you're expecting them to beat the pants off of your terrestrial wired service you're probably going to be disappointed, unless you're on the edge of the range of some old neglected AT&T DSL line or still trapped on dialup. Even then it's almost certainly going to be much more expensive but at least it will have a sizeable speed advantage.
- paulsutter 7y ago
  You want to downlink near the server, not near the client. Packets will move 40% faster through vacuum than through fiber.
  - wmf 7y ago
    Except if the terrestrial fiber backbone is cheaper and less congested you want to downlink near the client.
  - _flgt 7y ago
    Each RF hop is going to have more latency than fiber for RF physical layer processing. Forward error correction intentionally uses large frames to spread out burst errors.
    - Diederich 7y ago
      That's interesting, can you expand on that? Thanks.
    - paulsutter 7y ago
      Sounds speculative. Given link bandwidths and likely error rates, it seems better to allow the higher layers to retransmit than to optimize for correcting burst errors. Especially given Starlink's goal to control latency.
      - ianai 7y ago
        Have they given a range on latency goals? Ie seconds, tenths of seconds, milliseconds, etc? I’m seeing figures of 1 million$ per satellite. Over 10 years that’s roughly 100k/month in mortgage or around 1k customers at 100$/mo.
      - _flgt 7y ago
        I haven't seen a link budget. Do you have a link for one about the "likely error rates"? My guess is Starlink is going to need FEC for Doppler effects and multipath in urban environments. Maybe they will be able to overcome multipath with modulation. But Doppler shift will be significant with LEO. The latest 802.11n standards have been increasing physical layer frame sizes, not decreasing them. I don't know of any modern wireless standard that doesn't use FEC. Coding gain is significant in most link budgets I have seen. The last satellite RF downlink I worked with was for JPSS-1 in LEO nearly 6 years ago. There was a (255, 223) Reed-Solomon code with an interleave depth of 4 frames. So you effectively had to receive 1024 bytes, decode, and you got 4x 223 bytes worth of information. While latency wasn't a major concern, this scheme seemed pretty standard to me. If you don't believe me, the spec is public information: https://www.star.nesdis.noaa.gov/jpss/documents/CDFCB/GSFC_474-00001-07-01_CDFCB_External_Vol.7-1_JPSS_Downlink_Data_Formats__D34862-07-01_.pdf https://www.star.nesdis.noaa.gov/jpss/documents/CDFCB/GSFC_4... I'm sure SpaceX is trying really hard to control latency - so it doesn't get out of hand! One set of RF processing here, another over there, each with their own filters and decoding algorithms and then you have an extra 100 ms of delay in the system.
  - jandrese 7y ago
    The more traffic routes through the satellites vs. the ground the more you'll be competing for that limited sat-to-sat bandwidth and the more queuing delays you will suffer. You want to get the bits off of the birds and onto the ground ASAP if you want to maintain consistent latency figures. This was a big problem with Iridium. With only one ground station available your latency figures were all over the place thanks to the complicated routing that every packet had to go through.
- tbabb 7y ago
  > the more remote your location the further you likely are from the nearest ground station I thought downlinks were going to be pizza box-sized antennae. Why would it be far away from you? That could live on your roof.

Paul-ish 7y ago

Netflix claims an UHD stream is 25 Mbps. This is probably the most demanding use case for your average consumer, but it may not be a rare use case in the near future. Each satellite can support 800 UHD streams. That doesn't seem sufficient if each satellite covers a large geographic area. This suggests to me a low data cap and/or rate, perhaps comparable to DSL.

mastax 7y ago
12,000 satellites is the current FCC approved launch figures, I believe. 20Gbps per satellite but I don't know if that's full duplex or how much is divided between terminal/gateway, so lets divide that in half. For simplicity let's say they only sell consumer-grade 100Mbps connections, which are oversubscribed 100x as is typical. Divide the possible number of customers by 3 to account for uneven geographic distribution and other factors. 12000 * 20Gbps / 2 * 100 / 100Mbps / 3 = 40,000,000 customers A service like that could sell for a lot in rural North America, but maybe not to 40m customers. Let's assume most of those customers are from lower income countries or from places with more competitive ISP environments. Maybe $20/mo average revenue per customer. That's $800m a month. That already sounds compelling to me, before considering the potential for more or higher bandwidth satellites, or for higher margin commercial or government contracts.
- ianai 7y ago
  Another calculation is revenue per satellite per customer served. Would 100k per month per satellite finance these things long term?
  - Paul-ish 7y ago
    This seems like the important metric.
- sneak 7y ago
  I think your simple calculation assumes relatively even geographical subscriber distribution, which absolutely is not the case. Just rough estimating, assuming every other figure is approximately correct, you’ll probably have to cut that by 50-75% based on people’s natural tendency to live near other people. A few HFT customers will probably finance the whole thing, tbh.