Interplanetary Internet passes test
Posted: Wednesday, November 19, 2008 12:02 AM by Alan Boyle
After more than a decade of tinkering, NASA has successfully conducted the first deep-space test of a communication protocol that could serve as the foundation of an interplanetary Internet.
To mark the occasion, NASA team leader Adrian Hooke provided an e-mail reply to a message I wrote him back in 1999, asking when the interplanetary Internet would be deployed. He wrote, "I think that we just made it .... ;-)"
The fact that Hooke saved my nine-year-old e-mail message hints at how doggedly he and his colleagues have pursued the goal of creating a networking system suitable for deep-space missions.
Today, NASA's information superhighway to outer space flows through one major gateway - the Deep Space Network - to a host of space probes, scattered all the way out from Earth orbit to the edge of the solar system. As those probes proliferate, the Deep Space Network has to keep up with an increasingly complex communications schedule.
Hooke's team has been developing new networking tools to cope with the increasing load and the usual glitches and time delays that space missions have to weather. Those tools include a communication protocol known as DTN (which stands for Delay-Tolerant Networking or, more recently, Disruption-Tolerant Networking).
An Internet tough enough for space
The four-decade-old protocols that rule the Internet, known as the TCP/IP communications suite, are designed to work over a continuous end-to-end connection between the various parts of the network. That isn't well-suited for Earth-to-Mars communications, where the delay between sending a message and having it received can run as long as 20 minutes. And that's assuming that the antennas on both sides of the signal are working.
DTN is designed to accommodate a store-and-forward system, with built-in smarts. If one link in a communication chain is broken, a robot on Mars could decide for itself the next-best way to get its data back to Earth.
"By making the best use of the contacts you've got, you can smooth out the load on the network, and avoid having the network just loitering on one spacecraft," said Hooke, who is manager of space-networking architecture, technology and standards at NASA Headquarters.
For years, engineers on Hooke's team have been working with other network experts to wring the bugs out of DTN, through a series of earthly pilot projects. Hooke said the protocol has been used by Laplanders herding reindeer on snowmobiles, as well as cell-phone users on the bleeding edges of their coverage areas. It's even being deployed by the Pentagon for battlefield communications.
"There's quite a community now, the happy band of delay-tolerant networkers," Hooke told me.
Last summer, the UK-DMC satellite used the protocol to send sensor data down from Earth orbit to a British ground station and onward to NASA's Glenn Research Center in Ohio. That set the stage for October's monthlong deep-space test, involving NASA's Epoxi spacecraft.
"We have finally got the resources and the momentum up to take our own technology back and start putting it into space," Hooke said.
Simulated Mars missions
Epoxi used to be known as the Deep Impact spacecraft. After it flew past Comet Tempel 1 in 2005, the craft was recommissioned for a new mission, including a fly-by of Comet Hartley 2 in 2010. Right now, it's 20 million miles from Earth.
Using the DTN protocol, NASA bounced image data between Earth and Epoxi several times via the Deep Space Network. The network knit together 10 nodes, including Epoxi and several computer servers at NASA's Jet Propulsion Laboratory in California that masqueraded as Red Planet probes.
"We have one computer which pretends to be a camera on Phobos [one of Mars' moons], and we have another one pretending to be on Mars," Hooke said.
Epoxi was chosen for the test because the spacecraft's architecture was well-suited for uploading the new communication protocols, Hooke said. "We got the first images on the 20th, and since then we've been running about two passes a week," he said.
Some of the people involved in developing the deep-space Internet also played a role in building the very first Internet - and the first image transmitted as part of the DTN experiment paid tribute to those pioneers: It was a reproduction of a hand-drawn diagram of the original four-node Arpanet, sketched in 1969 by Steve Crocker, who is one of the Internet's founding fathers as well as a participant in the DTN effort.
The second image was a photograph of another networking pioneer, the late J.C.R. Licklider. Licklider's concept of a "Galactic Network" was an inspiration for the present-day Internet as well as the interplanetary Internet, Hooke said.
Among those who have been working on the new protocol is Vinton Cerf, another Internet founding father who is currently a vice president at Google. Cerf continues to be "very hands-on" in his involvement, Hooke said: "I've been in almost daily contact with him for the whole decade."
Next steps
Hooke said he was surprised to see that the protocol worked as well as it did for automatically routing data back and forth. "In the space community, that goes a long way toward persuading mission managers to put it on their spacecraft," he said.
The next step would be to install the software on the international space station, creating a permanent DTN node in Earth orbit. "These flight demonstrations are really important, because they show the thing working in a real mission environment," Hooke said.
It's hard to predict exactly when DTN will be needed for deep-space communications. Over the past decade, the "mission density" hasn't been heavy enough to require networks built to tolerate significant disruptions, Hooke said. But that time will come someday. Hooke just hopes that DTN will be fully tested and standardized by the time NASA starts building up networks of landers, orbiters and sensors, all talking amongst themselves.
"With Mars, we've already seen point-to-point-to-point archtecture," he noted, referring to the use of NASA's Mars Odyssey orbiter and Mars Reconnaissance Orbiter as relay satellites for Phoenix Mars Lander. "As you put stuff on the surface that could be networked itself, if you do everything by stitching together links, you end up with a horrible operations problem."
By the time astronauts make humanity's next giant leap, they may well be getting their e-mail via a dot-space address.
"The moon and Mars are probably the primary targets," Hooke said.
Those of us stuck here on Earth, meanwhile, can look forward to a brave new world (video from space!) ... or a brand-new nightmare (spam from space!). Will the interplanetary Internet be any better or worse than today's international Internet? Feel free to weigh in with your comments below.
