Photonics aglow

Researchers at McGill University in Montreal plan to explore novel photonic technologies that could make their way into enterprise networks, according to one industry observer.

McGill in January launched the Agile All-Photonic Network (AAPN), a project that aims to create more efficient communication architectures. Improved efficiency spells “a wider array of services and potentially less-expensive services for users,” said Dr. David Plant, the AAPN’s scientific director.

Plant explained that network technology today tends to squander speed and power, making life less than perfect for users and service providers.

The status quo infrastructure employs optical-electrical-optical (OEO) conversions. Devices on the network change data from the optical format to the electrical format to perform destination checks and amplify signals.

This transformation facilitates control. When data’s in the electrical format, network elements ensure that packets arrive where they should.

But the OEO switch also saps power and delays transmission somewhat. An all-photonic platform – network architecture that foregoes the costly conversion – would bring significant performance benefits over OEO infrastructure, according to Plant.

“If we can replace electronic switching with all-photonic switching…there’s the possibility that we may be able to provide better services, faster and cheaper,” he said.

The AAPN plans to create a new photonic switch that makes quick work of sending data. The group is working with a fast time-division multiplexing (TDM) system that essentially carves smaller time slots into signals and presents fewer transmission delays.

“At the limit of what we do, people refer to this as a ‘burst-switched’ network,” Plant said. “It’s the limit of time division multiplexing…totally asynchronous.”

He clarified that the AAPN’s switch would maintain a modicum of synchronicity. It’s not a burst-switched device, but close.

Beyond the hardware, the AAPN must devise new amplification and transmission schemes to make a reality of its all-photonic vision.

Since electronic amplifiers have no place in the all-photonic network, packets require sufficient power to reach their destinations. As well, with no OEO conversion, data must know its route before setting out.

“This will be a key aspect of our research, to decide if you have an all-photonic switch…what does that mean for deployment and management?” Plant said.

According to Julian Rawle, an industry analyst with Boston-based Pioneer Consulting LLC, the enterprise could be the first to benefit from all-photonic networks.

Rawle figures corporations would employ all-photonic switching in metro-sized links between campuses. “I think it’s probably going to occur in those areas before we see it in the long haul,” he said.

Rawle added, “Carriers are extremely conservative. They’re not going to switch over and completely rebuild their networks when a new technology comes along. They’re going to make sure it’s entirely proven and tested.”

To that end, the AAPN has collected a handful of equipment manufacturers as partners to help keep the research process from going overboard.

“The idea is we play off each other, push back and forth over what’s possible, what’s feasible, and what’s not practical,” Plant said.

Tropic Networks Inc. in Ottawa is one of the AAPN’s corporate partners. According to Dan Oprea, the firm’s vice-president of systems architecture, although Tropic offers an all-optical device for metro-area networks, the firm cannot solve the photonic puzzle alone.

“Tropic is addressing the real needs of the industry today,” he said, referring to the TRX-24000, the company’s DWDM platform. “AAPN is supposed to address technology for the future. I believe it makes sense to look into it.”

The AAPN has on hand $7 million for five years of research; the cash comes courtesy of the Natural Sciences and Engineering Research Council of Canada (NSERC). Plant said the research would yield practical applications well beyond that timeframe.

“We’re looking at networks beyond, say, 2012 or 2015…. We don’t have a customer and a network to deploy in 18 months.”