This Week in NW
Bigger role seen for defense R&D
Defense Department's DARPA funds network research that others find too risky.
ARLINGTON, VA. - A lack of venture funding for start-ups combined with a heightened fear of cyberterrorism may bring greater prominence to a long-time, behind-the-scenes investor in high-risk network research: the Defense Advanced Research Projects Agency.
DARPA, the research and development arm of the U.S. Defense Department, has a track record of funding key technologies used in military and commercial communications systems. DARPA is best known for creating the Internet in the early 1970s, but in recent years the agency has driven the development of wavelength division multiplexing and Gigabit Ethernet.
"DARPA doesn't get anywhere near the credit it deserves for the contributions it makes to research in communications," says Mike Schmidt, technical director for the leveraged technology group at BAE Information and Electronic Warfare Systems. "A lot of the patents that come out have roots in the [Defense Department]. . . A lot of the hot-shot engineers you find at start-ups came out of the [defense] business."
After last month's terrorist attacks, DARPA's role in funding research in such areas as high-speed optical networks and cybersecurity may become even more important as the commercial world looks to the military for better ways to protect information assets.
Last month's terrorist attacks will "enhance" DARPA's role in communications-related research, predicts Bill Collatos, managing general partner of Spectrum Equity Investors, a venture capital firm that invested in a start-up that evolved out of a DARPA grant.
"There are two implications of cyberterrorism," Collatos says. "One is how to protect against it, and the second is how to anticipate it and deal with it. Both mean processing a lot more information at higher speeds."
Housed in an office building in Arlington, Va., DARPA's 140 technologists award matching grants to university and corporate researchers to prototype promising technologies. The agency's $2 billion annual budget includes $590 million for research related to advanced networking and high-performance computing, according to Federal Sources, a market research firm.
DARPA traditionally works with defense contractors, but in recent years the agency has sought out network equipment suppliers, such as Hewlett-Packard, Lucent and IBM, to participate in research projects related to microelectronics, photonics and wireless communications.
DARPA's goal is to develop network components that can be used in commercial and military systems.
"When [technologies] go into the commercial world, that improves the yield, uniformity and reliability of parts," explains David Honey, deputy director of DARPA's Microsystems Technology Office. "And the costs go down."
One commercial network technology that DARPA helped advance in the late 1990s is the vertical-cavity surface-emitting laser (VCSEL) used in Gigabit Ethernet systems. DARPA worked with several companies, including HP (now HP spinoff Agilent Technologies) and start-up PicoLight, on VCSEL research.
"DARPA played a very big role in funding VCSEL-related research over the last 10 years," says PicoLight founder and CTO Jack Jewell. "VCSEL technology is the workhorse for Gigabit Ethernet fiber-optical links and . . . for parallel optical switches for the core."
Last fall, Agilent began shipping a parallel optical module using VCSEL technology that was developed with $18 million in DARPA funding. The module delivers 30G-bit/sec capacity in 1.5 inches of board space. Customers include switch, router and server manufacturers.
The DARPA funding "helped to accelerate the [parallel optical] program, which was quite sophisticated and parts-intensive," says David Dolfi, a department manager at Agilent's Communications and Optics Research Labs. "What that money enabled us to do is try and explore several different avenues - two or three different types of integrated circuit design, laser design and thermal packaging."
Sometimes DARPA research projects spawn start-ups. BAE Information and Electronic Warfare Systems created a separate company called TeraConnect to commercialize a high-speed opto-electronic module developed under a DARPA grant. BAE brought in venture capitalists to finance TeraConnect's product development.
"One of the things that DARPA helps provide is funding to get a technology to a certain maturity level," BAE's Schmidt says. "That makes DARPA more important in the rolling out of technologies from the defense industry because the market now demands that maturity."
Founded in the fall of 1999, TeraConnect raised $40 million from Goldman Sachs, Spectrum Equity and Kodiak Venture Partners. TeraConnect has shipped prototypes of its module - which is four times faster than today's technology and has a smaller footprint - to router, switch and server manufacturers.
"One of many factors in attracting venture capital was that DARPA had seen enough value in the technology to fund it," says Glenn Thoren, vice president of business development at TeraConnect. "Everyone we spoke to was willing to invest."
Spectrum Equity invested in TeraConnect because it could demonstrate a compelling technology, thanks to its DARPA-funded beginnings.
"If you forget that DARPA is a government agency, it's like one more source of venture funding that doesn't have a form of equity ownership," Collatos says. "It's the best of both worlds because it funds cutting-edge technology without dilution to other private investors."
Moving beyond the TeraConnect work, DARPA is now funding the development of optical components at the board and chip level that can be used for commercial routers and servers as well as weapons systems.
"We are migrating optical data networking deeper and deeper into the hardware itself," DARPA's Honey says. "Optical components inside systems provide higher data speeds and no cross talk. . . . The box gets faster and smaller."
Honey says DARPA is spending $45 million to $50 million per year on optical data network research.
"The on-chip optical interconnect work that we are funding today should be available to corporate users in about eight years," Honey says. "It will take a little longer to get it into defense systems."
DARPA also is developing a new class of antennas for mobile, wireless communications. Instead of broadcasting communications out in all directions, the new antennas can send communications in one direction at a time.
The antennas are designed to be low-cost and support voice, data and video.
"With directional antennas, communications can't be overheard or jammed unless you're between us," says Jim Freebersyser, a program manager in DARPA's Advanced Technology Office. These antennas will provide "faster data rates, a lower probability of detection and antijamming."
DARPA is spending $15 million to create a directional antenna network. The first demonstration of the technology is scheduled for February.
Commercial applications of directional antennas include communications to vehicles including taxis, trucks, trains and airplanes. These antennas would provide higher data rates and less interference than today's cellular networks, and they would allow cellular carriers to use their spectrum more efficiently, Freebersyser says.
DARPA officials say the economic downturn is encouraging more companies to work with them on network research projects such as these. Traditional defense contractors are finding it easier to attract and retain top scientists for DARPA projects, and commercial labs are more eager to compete for DARPA research dollars.
"This is the first time in 10 years where [network research] is better on the government side" than on the commercial side, Freebersyser says.
Observers predict DARPA will play an important role in keeping network innovation alive during the lean years.
"DARPA is going to be instrumental in making higher-risk investments that many companies won't make in an economic downturn," says TeraConnect's Thoren. "DARPA is going to play a very large role in what the next-generation network technology looks like."