Interplanetary Internet
Will Martian astronauts be able to load YouTube?
Mars rovers, as mentioned last week, can’t be controlled directly with a joystick. Even if you have fast download speeds and a working Internet network spreading through space, the signals simply won’t get to Mars fast enough.
But that’s not stopping people from trying to take Internet to other planets. The Internet, after all, has many other uses.
Right now, every space-probe sent out has to have its own mechanism for sending signals. Special software has to be written for every single spacecraft, telling it how to convert information into radio-waves, and how to send those radio-waves back, depending on its own special situation.
How would it be, if there were several “Internet server” satellites instead? All space-probes could use standard Internet software, and a lot of time and effort could be saved.
That’s a good idea, but there’s one problem: the “standard Internet” won’t work across planets. The Internet is a bit like Newton’s theories of mechanic. Everything works fine here on Earth — but, as distances get larger, its laws and systems begin to break down.
The Internet doesn’t work like interplanetary spacecraft, with a direct link from your device to the website at the other end. The delivery is more like early letters, which got passed on from one person to the next until they reached their destination.
You send a letter, saying something like “I want to search for avocado”, and send it to some address, such as “google.com”. That letter will be delivered by “runners”: people who run between two fixed points, carrying messages. When one runner reaches his point, he passes it on to the next runner going in that direction.
In the beginning, those runners’ points use to be marked by tall posts. That’s why the whole system is known as the “postal service”.
Posts aren’t always in a straight line from you to Google. They’re scattered all over, so there are many routes your letter can take. How exactly the letter goes depends on which runners happen to be free at the time.
The Internet works in a similar way. Instead of posts with runners, there are “servers” passing the message on through wires. Also, large messages are broken into many small “packets”: each packet is sent out separately, and they’re all put back together at the other end.
The Internet also has another feature of the postal service: timeouts. If your letter doesn’t reach within a few weeks, you’ll assume it’s gotten lost and send it again. On the Internet, the time you wait before re-sending is more like a few seconds. That’s why slow connections often use more bandwidth: they keep getting timed out and having to re-send their packets.
Signals coming all the way from Mars wouldn’t stand a chance.
At first, this looks like an easy problem to solve: just increase the timeout time.
But that won’t work. Timed-out packets are assumed lost for a reason: they probably are lost. When a server gets a packet, it sends it forth immediately — even if there’s nobody to receive it at the other end. (Okay, it doesn’t actually get sent, but is abandoned and lost forever).
In space, there’s a lot of chance that nobody will be there at the other end. To start with, the Earth rotates.
Imagine you’re at the Mission Control building on Earth, and controlling the Yutu rover on the Moon. Suddenly, the Moon sets. How will you send your signals? You can try pointing your antenna downward, but that won’t work. The ground will come in the way.
To avoid this situation, NASA has set up a Deep Space Network. It’s basically a network of radio antennae. There are three main antennae, as well as smaller ones, each on a different side of the Earth. That way, there will always be one of them facing in the right direction.
But the Earth is not the only thing that’s moving. What if the rover goes to the other side of the Moon, where it can’t see the Earth at all? What if you’re trying to contact Mars, but happen to be on the opposite side of the Sun? What if a sudden solar flare disrupts communications, or if your antennae are simply busy with something else?
When travelling in space, you’re bound to have gaps in communication, times when your connectivity becomes zero. And any Internet servers on the way are bound to have them too.
We can’t have them just throwing away packets every time there’s a break in the signal
Making an Interplanetary Internet is not just a matter of tweaking the timeouts. People have to change the whole way things are done. Luckily, a group of programmers — including Vint Cerf, creator of the Internet — are working on just that.
At the heart of the new plan is the Delay-Tolerant Networking, or DTN, system. It works almost like the ordinary Internet, but with one big difference: un-delivered packets don’t get lost.
When messages are being sent over the DTN, they get divided into “bundles”, which are like packets but a bit bigger. Each server on the network forwards its bundles to the next one. But it also — here’s the important bit — keeps a copy for itself.
So if the bundle doesn’t reach one server, it can be sent again right there, instead of having to start from the beginning all over again.
It’s as if post-offices kept a copy of every letter that went through, and automatically re-sent the letters that didn’t make it to the next office. You wouldn’t have to write the letter all over again, because the post-office would do it for you.
Delay-Tolerant Networking is useful because it doesn’t lose messages. It keeps trying, over and over again, until the message finally gets across. (Or at least, it gives up only after a reasonable number of tries).
DTN isn’t only about space. It can have other uses too.
One vivid example is mentioned in Wikipedia — although the website it originally came from has been taken down.
Imagine an isolated Arctic village, with computers but not Internet connectivity. Someone clicks on a link, sending the request to a DTN server. That server is then carried by dog-sled to the nearest networked location, where the actual web page is loaded. When the dog-sled makes a return trip, to the nearest village, the server is reconnected and the loaded page finally delivered to the computer screen.
It’s just like the ordinary Internet: only a bit slower because one of the connections is not wire but dog-sled.
Vint Cerf doesn’t plan to use dog-sleds. But the connections will still be slow, because the distances are long, and light cannot travel faster than light.
The Interplanetary Internet will probably work like a network of Internets. People on Mars will browse Mars websites, and people of Earth will browse Earth websites. The only time the slowness comes in would be when people from one planet wanted to load stuff from another one.
Of course, at least in the beginning, that might be pretty often.
If an Interplanetary Internet is actually set up, and used for things other than scientific data, it could have an interesting side-effect.
As Internet technology has improved, download speeds have gone higher — but download sizes have done so too. Ten years ago, the Gmail inbox could be loaded on a low-speed 64kbps connection. Now, without a fast 3G network, it simply doesn’t load at all.
As Internet speeds have become faster, web developers have become less concerned about keeping their websites small. Scripts and widgets and stylesheets get loaded “just in case”, even if they aren’t going to be used. And that’s fine, if you have a speedy connection. But what about all the people whose networks are still relatively slow?
That’s where the Interplanetary Internet may help people here on Earth. It could encourage developers to make their websites a little lighter.
Want to know more? ‘Interplanetary Internet’ is the second of a two-part series on sending data through space. Check out the first part, if you haven’t yet.
Have something to say? At Snipette, we encourage questions, comments, corrections and clarifications — even if they are something that can be easily Googled! Or you can simply click on the ‘👏 clap’ button, to tell us how much you liked reading this.