Vinton Cerf needed a new challenge so he's creating the mother of all Internets - to link us to other planets. By Steve Homer
Funding has finally been agreed for a man who cannot hear very well to listen to faint messages from space. It is an odd thought, but Vinton G Cerf, the man credited by many with bringing the disparate parts of the Internet together, probably would never have become so famous had his hearing not let him down. Now he has bigger plans.

Cerf has been wearing hearing- aids since he was 13. "My hearing loss coloured my interest in networking. It was easier to communicate via e- mail," he explains. "Telephones were pretty noisy back then. So I drifted into computer networking and that led me into helping to create the Internet."

Push him on his role and he is rather reticent. "I was lucky to be at UCLA when the Arpanet [precursor to the Internet] started," he says. But Cerf's modesty hides genuine achievement. He and Robert Kahn co-invented the Internet Protocol, which allows data to be exchanged between networked computers and the Internet. But it was Cerf's flair for bridge-building that helped bring the Internet to life.

In 1995 he was honoured by the International Telecommunication Union and in 1998 he received the Internet Society's Marconi Award "for technical achievements and ambassadorial leadership". But Cerf is not resting on his laurels. He combines the hard-nosed business world with continuing involvement with the Internet. He is senior vice-president for Internet architecture and technology at Telecom's giant MCI Worldcom. He is also responsible for the company's advanced Internet engineering group, the metropolitan area exchange engineering group, the packet networking group as well as an advanced architecture group. His latest passion will see his influence reach even further.

In late 1997 Cerf looked around for a new challenge. "The Internet was essentially invented in 1973 and did not really take off until 20 years later," he says. "So I wondered what we would need in 20 years' time. I thought we ought to try and extend the Internet out into space to communicate with robots and people.

"I was talking about this with Greg Miller, who told me that [Nasa's] Jet Propulsion Laboratory was already working on this and I was introduced to Adrian Hooke. Talking about it we got increasingly excited about the idea."

British born Hooke had founded the International Consultative Committee for Space Data Systems almost 20 years before. Through the committee, standards were already in place allowing space communication systems to talk together. The work mirrored the early days of the Internet, when Cerf and his colleagues were developing ground-based networks to communicate with each other.

Now the two network communities - Earth's Internet and the deep-space communicators - are joining forces to build the "network of networks" in space. Or as Cerf prefers to put it: "We created the Internet on Earth as a network of networks. The interplanetary network will be a network of Internets."

"We need to extend Earth's Internet so that little fragments of it can be deployed on other planets to allow today's smart machines and in the future, possibly humans, to communicate with each other. Those remote Internets then need to be connected back to Earth via long-haul communications networks," says Hooke. The two have no limits. At their most fanciful, they are looking into space and wondering what sort of strategy will be needed to communicate between stars.

An early part of the Space Internet could see its first outing as part of the Mars mission which will be launched in 2001. As more international missions are sent to Mars Mars rovers communicating with orbiting Mars satellites and sending their messages back to Earth will provide demonstrations of the new technology. But building the space Internet will not be easy. The biggest problems are latency and the interruption of signals. The Internet breaks messages up into tiny packets of data that whiz around the networks and are reassembled at the delivery site. Some packets can get lost or damaged. When the packet arrives at the receiving site, it is checked. If it looks OK the receiving site pings off a message saying "Packet 12345 has arrived safely."

Meanwhile, the sending computer waits to hear from the other computer. If there's no response in the maximum time, it sends the original packet again. This round-trip time, known as latency, is a major headache for the Space Internet.

"Latency is not so much a problem with delays measured in milliseconds," says Cerf. "More difficult is when it will take 20-80 minutes for a message to go from the Earth to Mars and back again and even worse to the outer planets, where the round trip can take hours, or to our nearest star where it will take eight years." Retransmitting missing pieces would take a very long time." These two factors mean that messages must be coded in a way that even if a large piece of data is lost, special codes will allow the computers at the far end to re-create it. To make matters more difficult, the deep-space network is oversubscribed and the ground stations lose sight of the spacecraft as Earth rotates, so there is only a limited amount of time to communicate over those incredibly long distances.

There is one other big difference. "The links between Earth and space are highly asymmetric - because we tend to need to send far more information back to Earth than we send up to control the spacecraft," says Hooke. "So in deep space we have an environment of very long delays, noisy data channels, different data rates in each direction, and links that keep coming and going. The Earth's Internet protocols were designed for exactly the inverse."

Cerf hopes that the Space Internet will eventually be able to offer a data rate of 1 megabit per second from space to Earth and perhaps a tenth of that rate in the other direction. But then again symmetrical communications between a manned base on Mars and another on Venus may be needed one day.

Developing a fully specified Space Internet will take years. There are bizarre questions to be faced. Cerf and the others are struggling with a familiar terrestrial problem - domain naming. "We are trying to work out the best strategy," he says. " We are wondering about planet-by- planet domain naming, but the whole structure gets very complicated especially with the long delays in delivering messages."

But the project is just at its beginning and Cerf and his colleagues are all enthusiasm and excitement. "We are still at such an early stage of exploration of what this architecture can be that we are looking at a variety of sources for ideas. We'll take good ideas from wherever we can find them," says Cerf. Those sources include everyone from Russian, French and other space authorities to outside industry bodies.

Surprisingly, the rather limp British space sector has managed a major contribution. The main Internet protocols (FTP, TCP and IP) have been adapted and they were loaded on to a British Space Technology Research Vehicle in Earth's orbit. "Two years ago we sent the first IP-based signals into space and back as a test of the new technologies," says Hooke. "Later, when the British project ran out of money, those standardised communications interfaces were used to pass control of the satellite over to a US university and extend its scientific data return for another year."

Cerf believes the exciting nature of the project will bring out the best in people. With much of the signalling technology aimed at solving problems in the mobile phone and military sectors, there should be a healthy cross- pollination of ideas.

At 56, space seems to have given Cerf a new boost. He is certainly going where no Internet guru has gone before. A lifelong sci-fi fan, he recently made his screen debut in Earth: Final Conflict, a television series written by Gene Rodenberry, the creator of Star Trek. Space is now in his blood and with luck, he will use his technical and diplomatic skills to help give the Space Internet a rapid and relatively painless birth.