E-mail from Mars
Plans for an interplanetary Internet are taking shape
If you plan to spend your next vacation on Mars, you currently face two problems: Getting there and sending e-mails with attached snapshots to your relatives and friends. At least the second problem will soon be solved.
In 1998 WorldCom's chief Internet scientist Vinton G. Cerf, who co-designed the Internet's transmission protocol TCP/IP, came up with his vision of an Interplanetary Internet (IPN). He shared it with Adrian Hooke, a scientist at NASA's Jet Propulsion Laboratory (JPL), who already had been exploring more efficient communications protocols for space missions, and they set up a project to make the vision real. In May 2001 the IPN Research Group, led by Vint Cerf and NASA's JPL scientists, presented an architectural definition for the Interplanetary Internet.
Obstacles in outer space
Internet in outer space is confronted with a lot more obstacles than on Earth. Radio signal delays, shortage of bandwidth, limited power supply, interrupted connections and increased security problems require new techniques and more patience. On of the most obvious problems is the sheer distance. It can take 20 minutes or more for a radio signal from a base station on Earth to reach an orbiter on Mars. Continuous and instantaneous connectivity does not work under these conditions.
First deployment in 2004
Scott Burleigh, chairman of the IPN Special Interest Group, and his colleagues at NASA's Jet Propulsion Laboratory are working hard to circumvent this limitation. As a first practical step towards the implementation of an interplantary Internet, they will use a new protocol for file transfer across interplanetary distances, called CFDP CCSDS File Delivery Protocol, which was designed by ten international space agencies, who co-operate in the Consultative Committee for Space Data Systems (CCSDS). The first critical test for CFDP will be NASA's Deep Impact comet mission, which is scheduled for 2004. In addition, NASA is planning to deploy a 'Mars network' of multiple orbiting relay satellites, intended to be launched in the years 2005 following.
However, building up a stable backbone of satellites in outer space for the interplanetary Internet will take some time. Even an optimist like Vint Cerf doesn't expect this to happen sooner than 2040. IPNSIG chairman Scott Burleigh sees at least a good chance that the Interplanetary Internet will be a fact of life in deep space robot operations by 2020 (see the interview). Both expect that IPN will sooner or later be used for commercial services. But what the business opportunities could be is still open.
Fully functional interplanetary Internet by 2020
When was the idea for an Interplanetary Internet born?
The idea has probably been at the back of Vint's (Vinton G. Cerf the editor) thoughts for decades, but we only started to get serious about it at the beginning of 1998. That was when Vint's interest in extending the capabilities of the Internet into the Solar System began to overlap with JPL's interest in improving communication with robots in deep space. We discovered we were looking at the same problem from opposite ends, so it was only natural for us to start studying it together.
Why do we need the Internet for interplanetary communication?
In the very near term, we probably don't; as long as interplanetary communication scenarios remain extremely simple the Earth talking to a spacecraft, and the spacecraft talking back the techniques we've been using for the past decade will work fine.
But NASA is starting to plan for missions whose communication needs will be quite a lot more complex. For example, there may be multiple rovers and/or base stations on the surface of Mars perhaps sensor webs as well that aren't powerful enough to transmit directly to Earth but do have enough power to communicate among themselves, and in some cases enough to transmit to relay satellites orbiting overhead which can in turn communicate with mission operations on Earth. These devices will be too far away for efficient micro-management by people, so their communications will have to be conducted by automated protocols. Those protocols will need to be standardized so that newly arrived devices can readily interact with equipment that is already in place, and the most successful model we've got for standardized, automated communication among a large population of entities, whether human or robotic, is the Internet protocol suite.
What are the main differences between Internet on earth and in outer space?
Actually, communication among 'landed' devices on the surface of Mars has to overcome many of the same problems as mobile computing on Earth; basically we need to use radio (rather than wire) as an effective data communication medium, without consuming a lot of electrical power. The constraints are more severe: electrical power is in 'really' short supply, in part because the arriving solar energy per square meter is so much less at Mars orbit than at Earth, and of course it's hard to replace or repair anything that breaks. All the same, many of the same techniques are applicable and in some cases existing protocols can be used with little or no modification. Even communication between landers and orbiters at Mars isn't radically different from communication with spacecraft in orbit around Earth.
The major differences show up when you look at the 'long haul' communication between Earth and, say, a Mars orbiter. The endpoints of communication are so far away that it takes a radio signal 4 to 20 minutes to travel from one to the other. What's worse, much of the time the orbiter is on the other side of Mars; a signal we send to Mars might arrive just as the orbiter is going out of sight, in which case the response can't even start on its way back until the orbiter emerges from the shadow of the planet, maybe half an hour later. So any protocol that relies on frequent, rapid exchanges of information between endpoints is pretty much useless for deep space communications; that unfortunately lets out most of the current Internet protocols.
Which in turn means that communication between a rover operator on Earth and a rover on Mars, via a relay orbiter, can't use standard Internet protocols end-to-end: even though they might be fine for the orbiter-to-rover leg of the trip, they'd break down over the long-haul link. That problem is at the root of a lot of the design work we're doing for the IPN.
How can you achieve security over the Interplanetary Net?
Security measures are being designed into the IPN protocols right from the start. We plan to rely on mechanisms that are already used for secure communication over radio on Earth, particularly symmetric and asymmetric key encryption. There are problems, of course: timely key distribution is difficult, for example, and we always have to bear in mind that some of the communicating entities will have very limited computing power and/or very low data transmission rates. We're looking at several approaches, including variations on Kerberos (authentication system developped by MIT the editor) and an adaptation of the S/MIME model.
What are the next steps towards IPN?
We've begun developing an initial prototype of the "bundle" protocol suite, in the expectation that building it will help us resolve some of the last remaining design issues; we hope it will also give us something to demonstrate eighteen months or so from now. At the same time, a new protocol for file transfer across interplanetary distances, called CFDP (CCSDS File Delivery Protocol the editor), is just now being adopted as an international standard. While CFDP itself won't function as an interplanetary Internet, it will give us some experience in dealing with the same kinds of operational issues and will serve as a proof of the concept we're working toward. The initial flight deployment of CFDP will be on the Deep Impact comet mission, which is scheduled to launch in 2004.
How do you estimate the chances for commercial services via IPN?
I think commercial IPN service is likely to become available as soon as there's a market for it, which will probably be at about the same time that we start to see private business operations in deep space. The pessimists among us would say that's a century away, but I can imagine some clever entrepreneurs over the next few decades finding a way to make money out of something that you have to go beyond Earth orbit to get. That will probably not happen until the cost of getting into and operating in deep space comes down, and maybe the IPN will help make that happen.
When do you expect the Interplanetary Internet to be fully functional?
By "fully functional" I think you mean more than the successful demonstration of the protocols in a lab; you mean real, practical use in deep space, yes? In that sense, I would expect the Interplanetary Internet to be fully functional as soon as there are enough functioning robots in deep space to make it indispensable. Probably not by the end of this decade, but I think there's a good chance that the IPN will be a fact of life in deep space robot operations by, say, 2020.
The interview was conducted by Milon Gupta.
Please send us your comments on this article.