Satellite Internet technology and Latency issues

[dakels]

I know latency is a big problem on satellite but how are the newer 2 way dishes? I understand the overall trafficflow of sat technology, but not the exact specifics.

a few major questions about it. How bad is the latency on 2 way? In practical figures, well I guess somewhat theoretical since some networks will be worse then others. How many extra steps is the system taking over a landline? I am assuming an extra 4? That extra relay and routing with the satellite for both directions.

Roughly how much average time per 1,000 miles (in ms) does a microwave signal take to send to the sat, back to the ISP, then ISP back to sat, then back to dish take? (I would assume the distance/speed of light (miles per/sec)/1000 = ms time) That would be a big part of the physical limitations. I am seeing only like 50 ms total for all 4 directions in 1000 miles though. And of course, thats just the actual physical time of the signal. Then you have all your other overheads of global routing and server delays.

So what would be/is a practical response time for 2 way satellite at say 2000 miles (I would assume 2000 is a decent average?). I would assume the distance to the sat/equator makes a difference of about 50ms per 1,000 miles? How much better could it get within a reasonable 3-5 year time frame? Assuming zero latency routing doesn't get invented anytime soon.

Also whats the difference between ms response time, and an average ping? Ms response time shows the time the signal takes to get to the target? and ping is how long the signal takes to get to the target and back? so does 2 x ms response = ping?

I hope I am making sense

 

[Armitage]

Originally posted by: dakels

So what would be/is a practical response time for 2 way satellite at say 2000 miles (I would assume 2000 is a decent average?).

Ok, maybe I'm really misunderstanding you here? Are you suggesting that the range from you antennat to the satellite is 2000 miles? All the satellite internet stuff I've heard about is to geosync at a altitude of about 42000 Km

 

[dakels]

Oh see I was under the assumption that internet satellites would use low orbit sats (200-300 miles). I think you are right though, they use GEO sats which orbit at like 25,000 miles or so to stay perfectly aligned with the earth rotation at 1:1 speed.

dam then you gotta figure your signal has to travel 27,000-30,000 miles, 4 times for you to get one ping. (dish to sat, sat to ISP, then back to sat, then sat to dish) Ok so if the speed of light is 186,000 miles per sec, then the physical time for the signal ALONE to travel that distance (say 30,000 x 4), with no overhead would be over 640ms!??? Dam that sucks!

 

[Armitage]

Sounds about right. There were/are several projects to launch constellations of LEO satellites for low-latency and/or broadband comm. Teledesic, iridium, orbcomm, etc. Many died in the planning phase, Iridium got on orbit, then went bankrupt. I think Teledesic is still alive in some form, but greatly scaled back from their early plans (> 800 satellites I think!?!).

The problem with LEO sats is that you need so many birds to get the coverage. At GEO you can cover the earth with 3 birds. Well, at least to about +- 70 degrees latitude.

 

[d00m]

Yep, about 600 ms for two way. Even more (I've heard about 800 ms) for one way.

 

[Smilin]

I did the math on this once long ago and yeah for a ping to go across the street and back via satellite it's over a half second delay. The bitchin problem is just how stinkin far up geosynchronous satellites are.

 

[CTho9305]

Originally posted by: Smilin
I did the math on this once long ago and yeah for a ping to go across the street and back via satellite it's over a half second delay. The bitchin problem is just how stinkin far up geosynchronous satellites are.

and if you had closer sats, then your dish would have to track them as they moved across the sky.

 

[Armitage]

Originally posted by: CTho9305

Originally posted by: Smilin
I did the math on this once long ago and yeah for a ping to go across the street and back via satellite it's over a half second delay. The bitchin problem is just how stinkin far up geosynchronous satellites are.

and if you had closer sats, then your dish would have to track them as they moved across the sky.

Not neccesarily ... depends on your antennas, how much gain you need etc. For example, you don't need to track GPS sats. Those are in 12 hr orbits, so the relative motion isn't as high as LEO though. Or Iridium for that matter, and that's LEO.

A design I saw, back when I was following teledesic, had an broad-band ground station that was kind of like a truncated pyramid with an antenna on each face. But that was back when teledesic was HUGE and they planned to have coverage such that you would always have at least 2 bird > 40 degrees above the horizon (IIRC)!

 

[dakels]

what do they use as the carrier wave (is that the proper term?) to send the data on? Not microwave is it? Is it radio? I wouldn't think they would allow microwave usage in civilian residential use 2 way sats.

always have at least 2 bird > 40 degrees above the horizon (IIRC)!

I don't know what impresses me more, what you just stated, or the fact that I actually understood it.

 

[Armitage]

Originally posted by: dakels
what do they use as the carrier wave (is that the proper term?) to send the data on? Not microwave is it? Is it radio? I wouldn't think they would allow microwave usage in civilian residential use 2 way sats.

I'm not really a comm guy, but alot of the geo stuff uses C and Ku band. I think the LEO stuff uses (or planned to) L and Ka band.

always have at least 2 bird > 40 degrees above the horizon (IIRC)!

I don't know what impresses me more, what you just stated, or the fact that I actually understood it.

Constellation coverage analysis is fun