Spaced Out on the Interplanetary Internet
By David Pescovitz, Fri Apr 09 08:00:00 EEST 2004

Interplanetary telecommunication needs an upgrade. That's the aim of the Interplanetary Internet, an effort to extend the capabilities of cyberspace to outer space. And while a ".mars" address is quite a few years off, the latest InterPlaNet research may have a dramatic impact on Earth-based connectivity as well.


Dropped calls and pricey service plans may be annoying, but that's nothing compared to NASA's wireless woes. A message sent at light speed from Earth to Mars can take up to twenty minutes to arrive. Worse, the geometry of a planet's rotation means that there's a limited communications window each day.

When NASA's Jet Propulsion Laboratory and the MITRE Corporation first, er, launched the Interplanetary Internet project in 1998 with funds from the US Department of Defense, it not-surprisingly grabbed the attention of mainstream media. But the project also raised eyebrows among hardcore network engineers, impressed by who signed on to oversee the research: Vinton Cerf, one of the "fathers of the Internet." In the mid-1970s, Cerf co-invented TCP/IP, the lingua franca of networked computers. A decade later, he developed MCIMail, one of the first commercial email systems. He's now senior vice president of Technology Strategy for the company.

"All of a sudden, my two great loves, space science fiction and the internet, are coming together," Cerf told me shortly after the Interplanetary Internet effort began. "How could I possibly pass up an opportunity to work on that?"

Someday, Cerf says, the InterPlaNet will involve electronic "post offices" that are scattered throughout the solar system routing messages between space explorers, both human and robotic. However, while the Internet's TCP/IP protocols do a great job of moving data packets around over terrestrial electronic networks where high-quality communication between nodes is essentially instantaneous and two-way, it simply won't cut it in the vastness of space.

According to JPL principal engineer and InterPlaNet project leader Adrian Hooke, the best we can hope for is "a network of disconnected networks" talking to each other using standardized protocols. Indeed, the Consultative Committee for Space Data Systems was established to aid space agencies and industry in the creation of these kinds of standards.

At the heart of the InterPlaNet is a delay-tolerant network (DTN) protocol overlay that uses hop-by-hop transmission and storage to shuttle information across the solar system. While working on a project with Intel Research Berkeley, I recently had the opportunity to meet Kevin Fall, a top-notch DTN alchemist and one of InterPlaNet's key architects. I hadn't heard much about InterPlaNet since it was first announced, so I was surprised when Fall told me that the group had just publicly released open source code for a generalized architecture based on the InterPlanetary Internet protocol.

One component of the DTN architecture, Fall explains, is known as "custody transfer," similar to "delegating responsibility for delivering postal mail to a person or service who promises to do so." For example, a "bundle" of data designated for a rover on Mars may travel from Earth to a satellite orbiting the red planet where it is held until a communication link to the rover can be established and the bundle dropped off.

Along with this "store and forward" technique, the Delay Tolerant Networking protocols also provide a form of intelligent routing to deal with the unpredictability of making a connection. The network itself, Fall explains, must keep track of how likely a message is to reach its destination along a particular route. This is especially difficult when the data waiting for transmission is integral to network applications, for example the control of an autonomous vehicle on Mars.

"You may not even know when links will go up or down," Fall says. "A measure of a contact's predictability can help to choose next-hop forwarders for message routing as well as select the next message to be sent."

Beginning with NASA's Deep Impact comet-penetrating mission scheduled for lift off on December 30, various spacecraft file-sharing protocols will be tested over the next five years, Hooke says, "incrementally leading to the eventual InterPlaNet." Then, in 2009, NASA plans to launch the Mars Telecommunications Orbiter, a networking "hub" to route the data streaming in from multiple Mars probes. Unlike the Mars Odyssey, Global Surveyor, and Mars Express satellites that did double-duty forwarding Spirit and Opportunity's stunning images back to Earth, the new Orbiter will be dedicated to interplanetary networking. This Orbiter will be outfitted with a high-bandwidth laser communication system to phone home.

As the InterPlaNet starts to take off, the researchers are also bringing the fruits of their labor back to Earth. DTN protocols, they say, could improve ad-hoc sensor networking for applications like monitoring soil conditions on farms or detecting tiny tremors along fault lines. Intel Research's Fall is also involved in a joint project with UC Berkeley to bring networking capabilities to developing regions where "always-on Internet" is truly an oxymoron. Another likely spin-off, Hooke says, are DTN systems for "communicating in stressed, tactical battlefield environments."

When I called Cerf for an update on the InterPlaNet, he grew most excited about the terrestrial applications of delay-tolerant networking. He's involved in an effort to build a delay-tolerant network for the Saami Reindeer herders in isolated areas of Swedish Lapland. In this store-and-forward model, snowmobiles will act as data mules, periodically relaying the Saami population's outbound communications to the nearest Internet access point and gathering their incoming messages.

When Cerf told me the story of the Saami, he prefaced it with, "this might sound really crazy." In reality though, it's no more crazy than emailing Mars.