4 ms·

The problem with communication satellites is that the technology isn't as good (throughput, latency, cost) as a ground based system, except for handling coverag

The problem with communication satellites is that the technology isn't as good (throughput, latency, cost) as a ground based system, except for handling coverage gaps. Coverage gaps on the ground are slowly going away, so eventually this business will likely decline.

mikeash 10y ago
Now you're changing from "so far" to attempting to predict the future, which is rather different. As far as the future goes, throughput and latency both improve dramatically with low-orbit systems. Cost is high (see the massive failure of Iridium to see how that can go) but is coming down pretty rapidly right now. And given how many satellite dishes I see on buildings in cities, it seems that there are use cases beyond handling coverage gaps.
- chris_va 10y ago
  Mostly fair, but... Latency to space is always going to be slower, as you are going to add a few hundred miles, and power requirements for high-bandwidth applications require large solar panels which are incompatible with low-altitude sats (too much drag). Then there is also a problem with non-geo sync orbits and directional antenna tracking. This is actually a huge problem. Current small sats point directly at their ground stations as they go over, which means they are only talking to one receiver at a time. There are some meta-material antennas that I have seen that might solve some of these problems, but not fully. Any satellite dishes you see on buildings, right now, are pointed at a geo-sync satellite instead of LEO. Half of those are now pointed at a ground-based microwave receiver instead, and the current trend is away from satellite. Anyway, time will tell, I guess.
  - mikeash 10y ago
    Lower launch costs make drag less of a problem, since you can give the satellites more fuel for reboosts. A few hundred miles of latency is no big deal. 200 miles adds one millisecond of latency. The speed of light is noticeably slower in fiber, too, so that balances out once you get past a certain distance. Talking to one receiver at a time is clearly not an inherent limitation of low-orbit satellites. Iridium satellites can handle multiple simultaneous connections without directional antennas on the ground side, and with a phased array on the satellite side. I'm well aware that dishes are pointed at geosync satellites. The fact that there are so many in areas with good coverage from terrestrial radio and cables means that there are applications beyond filling gaps, even with the latency that geosync implies. I don't recall ever seeing a satellite dish pointed horizontally, and certainly all the little ones I see on houses aren't, so I completely don't believe that "half of those" figure.
    - chris_va 10y ago
      Sorry, didn't mean to imply that your argument was wrong. For Iridium, the total bandwidth per satellite is really really low (like 28.8? Newer ones might be higher), and they had to put them slightly higher up (500 miles, I think?) to avoid drag issues since they needed bigger solar panels. The latency thing is more compared to microwave beaming, which is how most cell backhaul is done now, but you are right that the extra few hundred miles isn't critical (my head was still in geo-synch).
      - mikeash 10y ago
        The current Iridium satellites are ridiculously slow. They were designed in the 90s for voice, and having any data capability at all is sort of an afterthought. One "call" is a whopping 2400bps. Each satellite can support 1100 calls simultaneously, but even if you ganged them all together that's still only 2.6Mbps total. Iridium NEXT is going to start launching in September and will have vastly better capabilities, up to 8Mbps per connection. Still not great compared to a terrestrial wired connection or even modern cellular connections, but a lot better. If launch costs drop significantly, you could presumably do a lot better still.
- pilom 10y ago
  >As far as the future goes, throughput and latency both improve dramatically with low-orbit systems. The problem is that they still pare in comparison to ground based systems. On LTE, with aggregation, you can get something like 100 Mbps out of a cell phone on about >90% of the US's land area with Verizon's network. Irridium Next which doesn't even start launching until September will be limited to 8 Mbps and won't be completely ready for another couple years. Two-way LEO satellite communication will only be worth it to people on the oceans very soon. The satellite dishes on buildings only work with one way communication.

mikeash 10y ago

Now you're changing from "so far" to attempting to predict the future, which is rather different. As far as the future goes, throughput and latency both improve dramatically with low-orbit systems. Cost is high (see the massive failure of Iridium to see how that can go) but is coming down pretty rapidly right now. And given how many satellite dishes I see on buildings in cities, it seems that there are use cases beyond handling coverage gaps.

chris_va 10y ago
Mostly fair, but... Latency to space is always going to be slower, as you are going to add a few hundred miles, and power requirements for high-bandwidth applications require large solar panels which are incompatible with low-altitude sats (too much drag). Then there is also a problem with non-geo sync orbits and directional antenna tracking. This is actually a huge problem. Current small sats point directly at their ground stations as they go over, which means they are only talking to one receiver at a time. There are some meta-material antennas that I have seen that might solve some of these problems, but not fully. Any satellite dishes you see on buildings, right now, are pointed at a geo-sync satellite instead of LEO. Half of those are now pointed at a ground-based microwave receiver instead, and the current trend is away from satellite. Anyway, time will tell, I guess.
- mikeash 10y ago
  Lower launch costs make drag less of a problem, since you can give the satellites more fuel for reboosts. A few hundred miles of latency is no big deal. 200 miles adds one millisecond of latency. The speed of light is noticeably slower in fiber, too, so that balances out once you get past a certain distance. Talking to one receiver at a time is clearly not an inherent limitation of low-orbit satellites. Iridium satellites can handle multiple simultaneous connections without directional antennas on the ground side, and with a phased array on the satellite side. I'm well aware that dishes are pointed at geosync satellites. The fact that there are so many in areas with good coverage from terrestrial radio and cables means that there are applications beyond filling gaps, even with the latency that geosync implies. I don't recall ever seeing a satellite dish pointed horizontally, and certainly all the little ones I see on houses aren't, so I completely don't believe that "half of those" figure.
  - chris_va 10y ago
    Sorry, didn't mean to imply that your argument was wrong. For Iridium, the total bandwidth per satellite is really really low (like 28.8? Newer ones might be higher), and they had to put them slightly higher up (500 miles, I think?) to avoid drag issues since they needed bigger solar panels. The latency thing is more compared to microwave beaming, which is how most cell backhaul is done now, but you are right that the extra few hundred miles isn't critical (my head was still in geo-synch).
    - mikeash 10y ago
      The current Iridium satellites are ridiculously slow. They were designed in the 90s for voice, and having any data capability at all is sort of an afterthought. One "call" is a whopping 2400bps. Each satellite can support 1100 calls simultaneously, but even if you ganged them all together that's still only 2.6Mbps total. Iridium NEXT is going to start launching in September and will have vastly better capabilities, up to 8Mbps per connection. Still not great compared to a terrestrial wired connection or even modern cellular connections, but a lot better. If launch costs drop significantly, you could presumably do a lot better still.
pilom 10y ago
>As far as the future goes, throughput and latency both improve dramatically with low-orbit systems. The problem is that they still pare in comparison to ground based systems. On LTE, with aggregation, you can get something like 100 Mbps out of a cell phone on about >90% of the US's land area with Verizon's network. Irridium Next which doesn't even start launching until September will be limited to 8 Mbps and won't be completely ready for another couple years. Two-way LEO satellite communication will only be worth it to people on the oceans very soon. The satellite dishes on buildings only work with one way communication.