Issue No. 05 - May (2006 vol. 7)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MDSO.2006.35
"Five years ago, I wasn't thinking about working on wireless," says Gee-Kung Chang, a professor at the Georgia Institute of Technology's School of Electrical and Computer Engineering. But after 15 years of work on optical-network technologies, Chang and a team of researchers are now crafting a network design that's a unique hybrid—delivering high-speed wired and wireless services from the same signals on one optical fiber.
"Wireless is pushing for the highest bandwidth possible," Chang says. As providers clamor to offer more bandwidth-hungry content and users clamor for the best possible connection speeds, "the two worlds of fiber and wireless have to converge," he says.
Chang envisions his network design being used in locales such as airports, hotels, and shopping centers, and perhaps later in offices. As an individual connecting either wirelessly or over the wire, you'd get access to high-definition television, data, and voice services at much higher throughput speeds than you do with today's Wi-Fi or WiMax wireless technologies.
Technical and component cost challenges remain, but some industry analysts say Chang's team's approach could produce intriguing commercial products within five to seven years.
Chang's design would hook up to today's widely deployed optical-fiber networks. His approach would convert signals traveling on the optical fiber (at the central-office level) before they reach a destination such as a hotel or mall. His technique transforms the signals from their usual infrared wavelengths to the millimeter-wave spectrum. The tricky conversions use optical techniques such as external modulation, four-wave mixing, and cross-phase modulation. Economic feasibility is also a design requirement. So far, the external-modulator techniques look the most promising, Chang says.
Signals, split into two parts, could travel via a building's passive optical-network infrastructure. A high-speed receiver built into a room's ceiling would recognize the first signal part and amplify it for wireless transmission at the 40- to 60-gigahertz range. The second signal part would use inexpensive receivers and optical filters to offer the same data to users who want to plug into a regular wall outlet.
Wired and wireless users would enjoy downstream data rates up to 2.5 gigabits per second—much better than at today's wireless hotspots. Chang's team is targeting slower upstream rates, of less than 1 Gbps per user.
"Radio-over-fiber technology has been used for transmitting RF wireless signals for the last decade," Chang says. "But this is the first time that both optical baseband signals and millimeter wireless signals are transmitted simultaneously in the local broadband access networks."
Multiple service providers delivering different types of content could use the same optical fiber going into a location. Wavelength division multiplexing (WDM) would allow as many as 32 different channels, all offering the 2.5 Gbps speeds, Chang told attendees at the March 2006 OFC/NFOEC show ( http://www.ofcnfoec.org) (Optical Fiber Communication Conference & Exposition and the National Fiber Optic Engineers Conference), where he recently outlined the network design.
"There have been a lot of attempts to do free-space optics. What's interesting here is they're making much better use of free space optics," says Eve Griliches, Research Manager, Telecom Equipment, at market research firm International Data Corporation (IDC). "People have tried to use similar wavelengths before, but they weren't functional. This application for using those wavelengths is the best that I've seen."
Chang's team faces several technical challenges before the network architecture is ready for prime time. Most important, they must perfect a low-cost, effective technique for the millimeter-wave spectrum conversions. The team must ensure efficient full-duplex implementation of protocols for the uplink path, Chang says. Antenna designs must avoid unpleasant encounters with other wireless services in a given building.
"Any time you start converting wavelengths like this, you're going to have optical-dispersion issues," IDC's Griliches says. "Then you have to ask, how does it scale? Will it support numerous connections on the fiber without having problems? This is hard to test because it takes a lot of equipment. Until the tests are done optically, you really don't know," she says.
The other important challenge will be how the costs play out, which is of course hard to judge when the technology is still five to seven years from commercial deployment. "The components are expensive," Griliches says. "They are using some off-the-shelf components, which is good."
Depending on the local area and how much fiber is already present, using WiMax technology with a bunch of access points could prove more fiscally appealing, according to Griliches. It's too soon to predict exactly how the dollar totals will compare, she says.
"I believe the bottom line is volume," Chang says. "The costs can be driven down with volume."
NEC and BellSouth have expressed interest in the technology and are working with Chang's team on problems including component and integration choices.
"We believe it is very interesting research with very good results," says Milorad Cvijetic, Chief Technology Strategist for NEC America. "We have some research collaboration right now, with respect to component-level integration issues," he adds, noting that NEC collaborates with various universities on advanced research topics and follows research results from multiple people working on optical technologies.
"Down the road, the line between fixed and mobile will be a softer line," Cvijetic says. "The point now is how to integrate the two worlds."
What's notable about Chang's work, according to Cvijetic, is that his group has expertise in both the wireless and optical arenas—so it knows how to effectively optimize both types of components for performance. He also appreciates Chang's focus on costs.
Might the hybrid network design have other commercial or academic applications? It's a bit early to tell, but Chang has a few ideas.
"This optical wireless technology can provide the last hop of HDTV distribution in the home, office, theaters, or airports," Chang says. "It can also be used for reconfigurable interconnections between servers, supercomputers, and routers."
Meanwhile, he's intrigued by the converged networking landscape he's helping to paint with his everyday work. "Ultimately, future-proof access can only be provided by the convergence of optical wired and wireless networks," Chang says.