Slashdot Mirror
US Air Force Building Space Router
Saint Aardvark writes "From the ISTS daily news comes a story on the US Air Force seeking to build a space router. From TFA: "Northrop Grumman and Caspian Networks are collaborating to develop an Internet Protocol router that can withstand the constant barrage of solar radiation in orbit. The space-hardened IP router will be part of the Air Force's Transformational Satellite Communications System, which will provide IP-based communications to warfighters." I wonder what the ping times would be like..."
I wonder what the ping times would be like...
(nb: I worked on some satellite internet stuff a few years ago.) If this unit is in geosynchronous orbit (so a fixed dish can always hit it), it's sitting almost 36,000 Km over the equator. Assuming your dish is at the equator a round trip is ~72,000 Km / 300,000 Km/sec (the speed of light) means the signal travels about a quarter second earth->earth not including any processing time at the satellite midway point or either end.
Trolling is a art,
The ping time would probably be pretty good, considering I can ping servers all over the planet with good response and that almost invariably takes the data through at least one satellite somewhere... so it would be the same except it would be DEDICATED traffic instead of shared with a bunch of spam and porn!
=Smidge=
It would be around 270ms for a satellite in geosynchronous orbit. It would be a bit more for a router on the moon. :)
You have enemies? Good. That means you've stood up for something, sometime in your life. --Winston Churchill
http://ipinspace.gsfc.nasa.gov
Actually this is old news. In the rare event anyone here wants detailed information you can find it here: http://ipinspace.gsfc.nasa.gov/documents/NRO.pdf
Ping times can be found on page 82 of the pdf
Yes, but not nearly enough for a neural network complex enough to mimic the human brain when ... [this post terminated for violating the time traveler act of 2143]
"Learning is not compulsory... neither is survival."
--Dr.W.Edwards Deming
We already have TDRS (tracking data relay system). It is a system that can acquire data from satellites in low-Earth orbit with near-global coverage. A set of specialized satellites in geosynchronous orbit are used to track, command, and acquire data.
Mea navis aericumbens anguillis abundat
How does that work? In the worst case the satellite appears on the horizon, which adds one Earth radius to the range. The total distance changes to about 84000 km (from 72000km in the parent post), which still leaves the propagation time at under 1/3 second.
Less is more.
Um, Nasa beat you to the idea on both counts. In fact they have an entire system called the Deep Space Network that uses an adapted form of TCP/IP to relay information around the solar system. (Adapted because TCP/IP normally cacks when delays exceed 2 minutes. Travel time for signals around the solar system can be hours or days.)
"Learning is not compulsory... neither is survival."
--Dr.W.Edwards Deming
You didn't catch that current technology doesn't do the routing IN space, it does it at a single point on the ground. This allows several uplinks to be used more effectively. As an example, if you make use of these vsat IP providers to connect between two remote sites, the communications would be ground->sat->ground (hub)->sat->ground, meaning the packets have to traverse twice as far as they otherwise would if routed in space.
You're pings are not being routed through a satellite.
The Router
Here's an ISS status report that mentions it.
/sig
Unless you're sitting at a very out-of-the-way spot, your traffic to servers over most of the planet does not go through a satellite. Satellites are pretty much out of the the internet business. The latency is simply too long for geostationary satellites, and LEO satellites either need fancy rotating antennas on the ground stations or provide lousy bandwidth. Iridium gives you 2400bps. Yay.
You can easily check for yourself with traceroute. If a hop costs you more than 240ms, it's probably a satellite. I tried testing the link to Greenland, but Tele Greenland seems to block traceroute (both icmp and udp).
Finally! A year of moderation! Ready for 2019?
I'm not sure what the grandparent was talking about, but in the case where the sat appears on the horizon, you might not be able to get a good enough signal through that much atmosphere. Worst case, you might need to relay through a satellite that appears in (guess here) the 90 degrees of sky directly above you. This being the case, you might then take substantially longer than 1/3 of a second to get from point A to point B.
;-)
Where your original distance was a function of 2R, you're now travelling distances that are a function of 2*pi*R+2R... the other inputs are the distance between the earth's surface and the sat (X), as well as the distance between sats in radians (Y). The final formula, if you are directly below the first sat, and your target is directly below the second, 2X+YRsat (keeping in mind here that Rsat is the Rearth plus X), and that grows a bit if your sat is not, or you are relaying through a sat that is not directly above your target, of course.
Ok, that's enough of that
Is that why the Internet works?
IPv4 simply routes data. Its not supposed to be secure, at least until IPSec. Usually for electromagnetic waves the layer 2 protocol provides the encryption, and everything above it works as normal. Thats the simplest and most reliable implementation. Trying to encapsulate routed packets, setting up routing rules to work with it etc gets more complex than defining one layer 2 channel, encrypting it, and letting all layer3 packets route themselves over it.
Thats Why IPSec isnt used much, except in VPNs.
Just encrypt the EM waves, like the military has since WWII. The digital data in the waves can also be encrypted as a part of the layer 2, above of which everything becomes normally routable without much configuration and the device(s) used in such communication can easily be deployed everywhere without fat manuals explaining tunnelling, IP headers, routing rules and the likes.
"Give orange me give eat orange me eat orange give me eat orange give me you." -Nim Chimpsky
Almost totally wrong, the chances of your data going though a satelite on a normal day are pretty low, unless you have satelite internet access.
There are two options to the location of a space-router, one is a geo-stationary orbit, this would take it to 36,000 km, so a round-trip for the signal of 72 milion meters and a 'lag' of 240ms just in getting there and back (slightly longer since you don't want to send/recieve from the same palce).
The other choice is that if the system works they opt for a selection of fast-orbiting satelites which won't stay over the same spot and instead work 'shifts' over the diferent parts of the globe, these satelites could orbit at significantly lower orbits, the lowest being 320km, which would only incur a latency of just over 1ms (to a station directly below).
Most commercial satelites tend to opt for the geostationary, Iridium is the only one I can think of which has enough satelites to cover the world at a relatively low orbit around 750km I believe, 66 satelites that takes, lower would need more due to the curvature of the earth.
Well, if it's in orbit, the ping time will be around 750ms minimum, as it takes about that long for a satellite internet connection to return a ping. About 600ms to go from terrestrial to orbit and back again, + processing time and such.
"Champagne for my real friends - and real pain for my sham friends!" http://ericblade.postalboard.com/
First of all, Northrop and Lockheed are one of the 2 parties competing for the TSAT contract. Boeing & company (ie IBM, etc) are the other party.
Let's see... w/o divulging any proprietary or competition sensitive info:
In regards to space effect... loads of radiation can cause Single Event Upset (SEU) or latch up. Getting semiconductor parts made for space is extremely costly, and afaik, there isn't a single router chip that can reliably operate in space. As far as shielding is concerned, you'll need a ridiculous amount of tin-foil to block enough of the heavy particles, so that's not very fesible. SOI is inherently better when it comes to space effects, but still pretty shitty compared to a rad-hard part. I think the most advanced commercially available space-qualified CPU is Honeywell's chip based on Motorola's (Freescale) 750, and there are very few FPGAs (for DSP etc) with little gate count that are space qualified...
Last but not least, the capacity that the Air Force wants is pretty insane... I think that's all I'm allowed to say...
Uh, many many military folks use the word warfighter. It describes the jobs of the front-line combat service personnel from all branches of the military (not cooks or other non-combat personnel).
"We provide the best XYZ possible to our nation's warfighters".
I've seen it for a number of years in (a) magazines devoted to military equipment (Journal of Electronic Defense (JED)) (b) heard it from the various military customer-type people I come into contact with as a defense contractor (they may have been infected with 'bizspeak', though, for all I know).
JED in particular has a column called "I: First Person Singular" that is usually accounts of retired service personnel describing their experience with various electronics defense systems (radar, ECM, ECCM, other EW gear) in combat situations (World War II, Korea, and Vietnam). The people writing these columns often use warfighter without any indication of discomfort or irony.
It's a real word that is just now percolating into common usage via the enlightened interest in things military spurred on by the conflict in Iraq.
I work for a corporation that uses a satellite for data communications.
As of five minutes ago, I did a ping test from my computer on our WAN, through the AMERICOM-4 satellite, to a location connected to a VSAT dish.
PING RESULTS:
Approximate round trip times in milli-seconds:
Minimum = 1412ms
Maximum = 2013ms
Average = 1682ms
Looking further into it using a tracert, I have the following results (IPs/Hosts removed):
(less than signs should be infront of each 10 ms entry)
1 10 ms 10 ms 10 ms LAN/WAN
2 10 ms 10 ms 10 ms LAN/WAN
3 10 ms 10 ms 10 ms LAN/WAN
4 10 ms 10 ms 10 ms LAN/WAN
5 10 ms 10 ms 10 ms LAN/WAN
6 10 ms 10 ms 10 ms LAN/WAN
7 10 ms 10 ms 10 ms Sat Hub
8 1482 ms 1392 ms 1442 ms VSAT Host
As you can see from the numbers above the ping times would be like 1.5 seconds!!