Research team plans Internet map

by MIKE MARTIN, UPI Science Correspondent

NOTRE DAME, Ind. July 20 (UPI) -- Mapping the entire Internet is the ambitious and controversial project of computer science researchers from Notre Dame, who believe such a map will become essential as the Internet increasingly outgrows the networks that route its traffic today.

A map "that captures the Internet's large-scale topology is crucial for the development of efficient routing protocols and modeling Internet traffic," physicist Albert-Laszlo Barabasi said in a recent paper, with fellow researchers Soon-Hyung Yook and Hawoong Jeong. "Our ability to design realistic networks is limited by incomplete understanding of the fundamental driving forces that affect the Internet's evolution."

An Internet map is a unique concept, according to Jennifer Rexford, a computer scientist at AT&T Laboratories in Florham Park, NJ.

"The Internet is so decentralized that there is no global registry of its topology," Rexford told UPI. "That is, there is no "map" that identifies the many highways and byways that make up the network."

Barabasi believes a map, with the physical layout of nodes, routers, links, and hubs, will allow designers to predict the growth and course of the Internet -- and plan accordingly.
"By combining the most extensive data on the time evolution, topology and physical layout of the Internet, we identify the universal mechanisms that shape the Internet," he said.

Not an easy task, according to Rexford.

"Observing all of the 'roads' in the network is difficult, if not impossible," she said. "Also, seeing a 'road' appear or disappear from view does not necessarily indicate that the road was created or removed. This makes it a fundamentally difficult problem to accurately characterize the changes in Internet topology over time."

Networks such as the Internet share a common form as one moves from one layer to the next -- a structure Barabasi calls "fractal" and hopes to exploit in constructing his map.

"A fractal pattern is a repetitive pattern -- roughly the same pattern emerges at each level of interest," communications professor Steven Jones told United Press International from the University of Illinois at Chicago. "It makes sense that the Internet has such a structure."

"We find that the physical layout of nodes is determined by population density patterns around the globe," Barabasi explained in his paper. Such a node-hub style structure might resemble a set of airline routes -- or more eerily, the neuronal architecture of the human brain.

Barabasi's conclusions are "in agreement with other work in this area on the connectivity of networks on the Internet," Akamai Technologies scientist Leonidas Kontothanassis told UPI from Cambridge, Mass. "That is, certain networks have a much larger number of connections to other networks and the number of connections decreases exponentially from the most popular to the least popular ones."

Would a map of the Internet look like neurons in a brain? It does on a small scale, according to Steven Jones.

"If you type in a URL at a site called Webstalker, which actually maps all of the links to and from the URL, it looks like a cross between a map of neurons and a spirograph -- the old children's toy," Jones, a senior research fellow with the Pew Internet Project, told UPI. Unlike a living organism, however, Internet growth patterns are much harder to predict than Barabasi and his team represent.

"I think they can predict the geometry of the Internet," Jones said. "I have trouble believing they can predict how the Internet will grow."

For instance, Jones said, much of the growth is wireless.

"The wired web is one thing," Jones said. "How do you map the wireless web?"
Jones is also careful to distinguish between "Internet" and "world wide web."
"The Internet is hardware, the web is software -- a connection of links and pages," Jones
explained. "We are increasingly evolving away from hardware."

Mapping the evolution of software is much more problematic, Jones explained.
"As the web becomes more peer-to-peer and less hub-to-hub, it becomes more diffuse and chaotic," he said. "Again, I'm not sure how all that can be mapped."

Barabasi is convinced, however, that a network is a network -- no matter its makeup.
"When it comes to pattern formation, there are a lot of common phenomena," he said in a recent interview with Notre Dame Magazine. "What you try to understand is, why are they similar? What physical forces are creating this?"