Hacker News Reader

Comment by leoedin | Hacker News Reader

The thing to remember with all geosynchronous satellite communication is latency. It's terrible. By virtue of being 36,000 km away, you fundamentally can't have a communication time of less than 1/4s. That's fine for moving large numbers of bits (video transfer) but terrible for 2 way communication. Without any other delays, it takes over 1/2 a second between sending a request and seeing the result.

Perhaps the protocols they're developing will have use for ground level civilian communications, but it's unlikely that satellite based internet will ever see widespread adoption in the west because of latency.

View on HN

But why should the satellites be on geosynchronous orbit? What if we bring them closer to Earth? For example, when we use mobile internet, we travel from antenna to antenna. With satellites on lower orbits, the antennas will travel around us.

I think it's hard to pinpoint a satelite that is moving around all the time.

This is possible, but the problem is that the closer your satellites get to earth, the larger number of satellites you need in your constellation to maintain coverage. Iridium[1] spent $6 billion putting a very large constellation of satellites into orbit and promptly went bust. Convincing an investor that you won't end up having the same problems that iridium had would be tricky.

It's certainly not impossible, but I would be surprised if the economics of satellite based internet can compete with fibre based networks. The areas where fibre isn't feasible are also probably the areas with few people, and so the potential to make money is limited.

Iridium has shown that (once you disregard capital costs), there is a niche for satellite based telecom services, but it's a fairly niche area.

[1]: http://en.wikipedia.org/wiki/Iridium_Communications_Inc.

Well, if it's DARPA, then probably the main purpose of the network will be military. And I'm pretty sure that they have much more money than Iridium.

Obviously military use is quite different from civilian use. I was really trying to address the part of the original article which suggested that this technology might make its way into civil life.

The speed of light is 300,000km/s. Even with a satellite 36,000 kms away, you could have round-trip latency of around 100ms. That's fine for communications.

The reason why news broadcasts have such horrible latency has more to do with compression windows for older video codecs than broadcast latency.

I'm not really sure how you're getting 100 ms. For someone on the equator, time = 2*d/c, which works out as 72000/300000 or 0.24s. Unless you're trying to access a server actually on a satellite, the data has to go up and back down again.

I was being lazy and pulled my numbers from wikipedia.[1] It seems that the calculation there takes into account latitude of the transmitter and receiver, and so gives a slightly larger number. For non-equatorial ground stations, latency is higher.

In the context of communications, it's important to clarify what "round trip" means. The latency to send data from one ground station to another is, at best, 240 ms, but the time taken to request data and then receive it is going to be twice that. In practice it means that if you were browsing the web over a geostationary satellite link, it would take 1/2s between clicking a link and receiving the first data back from the server.

[1]: http://en.wikipedia.org/wiki/Geostationary_orbit#Communicati...

I didn't read this article as discussing satellite internet. They propose connecting cell phone towers with free-space microwave links: a microwave relay from one cell phone tower to another, direct, no satellites involved.

Reply on news.ycombinator.com