From making phone calls to watching TV to surfing the Internet, every electronic transmission we use must travel over fiber optic lines, the transportation system for mass media. George Rouskas , Emre Yetginer , and doctorate student Zeyu Liu undertook the task of creating a model to solve a 20-year problem in network design—finding the most efficient path for our electronic musings.

“We compared the solution times using previous techniques with the solution time for our technique,” Rouskas said. “Ours is 10,000 times faster.”

The research team recently published their findings in the July issue of the Journal of Optimal Communications and Networking. The discovery deals with a problem inherent in “ring” networks.

According to William Brockelsby , Lead Network Architect at the Office of Information Technology, to minimize the miles of cable when connecting many customers over long distances, providers must deploy “ring” networks, analogous to the interstate highway system, which facilitate faster and cheaper travel over long distances.

According to Rouskas , the problem his team dealt with occurs when the number of nodes, or “interchanges,” increases in the ring.

“When you know there is a lot of traffic going either east or west on I 40, you have other routes to go around it, which helps distribute the traffic so it does not become very congested,” Rouskas said.

Fiber optic transmissions are in wavelengths of light, and two transmissions cannot use the same wavelength on the same line, Rouskas said. Network designers must not only find the best path but also the best wavelength.

Finding the best connections on today’s typical ring network, according to Rouskas and Liu, can take days of tedious work.

“You have to allocate a lot of resources and spend a lot of time doing one small networking thing,” Liu said. “This is unacceptable.”

The team, according to Rouskas , realized that the process for finding the best path for information was inefficient, slow and hampered solving more general networking problems.

Yetginer said solving this old problem was not an easy process.

“Certainly, the first ideas were not the ones that led us to good results,” Yetginer said. “But each unsuccessful attempt teaches something useful about the problem.”

Lui said the key to this research was persistence.

“When you do research, you keep trying and trying,” Liu said. He said the team spent months together, often going in wrong directions, but continued to work off each other’s ideas and toward a result.

“Progress was certainly not linear,” Rouskas said. “The model we developed is not straightforward, and it took us a few months to straighten out all the wrinkles and prove the model is correct.”

The benefits of the model will allow network designers to find the best connections often in seconds as opposed to days, according to Rouskas .

“It is not something you will notice just by clicking a webpage, but it is something that will make the infrastructure more responsive to user needs,” Rouskas said.

Rouskas said he has received positive feedback from the scientific community.

“I’ve heard from people who have seen the work, and they all say they are very amazed at how fast we can solve these types of problems now,” he said.

Rouskas , who has been at N.C . State since 1994, spoke to the quality of researchers and graduate students who were able to work with the necessary in-depth graphical theory. He also praised the freedom he has been given.

“The [computer science] department and the University in general give us the freedom to work on problems that interest us faculty, and this freedom to pursue one’s interests is what eventually leads to breakthroughs,” Rouskas said. “We have built a world-class research group in optical networks.”