Networking tech promises speedy set-ups

A new signaling protocol will make it the work of a moment to set up or tear down heavy-duty communications circuits, researchers say.
Written by Marguerite Reardon, Contributor on
A group of researchers says it has developed a signaling protocol that will make it faster and easier to install connections on heavy-duty networks used by telephone carriers.

The "Just in Time," or JIT, protocol greatly reduces the time it takes to set up and tear down communication circuits, Dan Stevenson, vice president of advanced network research at the MCNC Research & Development Institute, told the Federal Communications Commission's (FCC) Technological Advisory Committee on Wednesday.

The protocol has been used in optical networks, but researchers are exploring its use in wireless networks, too, Stevenson told an audience of carriers, vendors and industry consultants at the quarterly committee meeting.

"Some of the benefits we have come up with for optical networks could be applied in the RF (radio frequency) environment," he said. "Signal quality and channel allocation are all functions that could be improved when dealing with spectrum as well as optical networks."

The protocol promises to benefit owners of large-bandwidth networks, such as telephone companies, research institutions and big corporations, that want to add a location. At the moment, it can take months to manually set up a connection to an optical network. JIT automatically puts up links within microseconds, Stevenson said.

Another signaling protocol, Generalized Multiprotocol Label Switching, has also been proposed as a solution to this problem. But with GMPLS, which works using light paths, it can take anywhere between several seconds and several minutes to install or remove an optical connection.

Checking ahead
JIT is faster, supporters claim, because it doesn't wait for acknowledgements from destination devices when it configures the addition to a network. Most protocols--Asynchronous Transfer Mode (ATM) and GMPLS, for example--send a signal ahead to alert devices to incoming traffic. Before data is sent, the sender must receive confirmation that these devices are ready. By contrast, JIT sends out messages that nail up the connections as it goes, then transmits the data without waiting for confirmation.

One drawback to JIT is that data could be dropped if it is sent to an overloaded switch or to a break in the fiber network, Stevenson said. Such drops could disrupt service.

Stevenson said reliability could be improved using techniques such as wavelength conversion, which allows traffic coming into a switch to be routed out to the network on a different wavelength.

In the nonresearch world, physicists are using JIT to send large amounts of data across grid computing networks, Stevenson said. Traditional IP routing methods have proven too slow to maintain interactive connections in situations where terabytes worth of data need to be transferred across the United States, he explained.

JIT is seen as much more effective, because it's faster at establishing network connections, Stevenson said. The U.S. Department of Defense is using JIT in its Advanced Technology Demonstration Network in Washington, D.C. This all-optical network links three locations within the Defense Department.

Stevenson said that researchers at MCNC, a nonprofit research organization based in Research Triangle Park, N.C., are still looking at how JIT could be used to allocate spectrum in wireless networks. However, it is still in its early days and no standards currently exist for the technology. The organization is talking to equipment vendors--such as Cisco Systems, Calient Networks and Glimmerglass Networks--to drum up support, Stevenson said.

Audience members at Wednesday's meeting seemed somewhat skeptical about the technology at first, pointing out that the concept has been tried in the past without much luck. But some in the group said it might be time to take another look.

"As engineers, we have a way of discounting a technology because we say, 'We already tried that 20 years ago, and it didn't work,'" said Robert Lucky, chairman of the Technological Advisory Committee. "But I agree with others here who say that this could be a technology whose time has come."

Editorial standards