For more than a decade, wireless academicians and the industry's research community have expressed optimism about the potential benefits of cognitive radios that are “smart” enough to automatically find and utilize fallow spectrum — and relocate to other frequencies at a moment's notice when the incumbent licensee decides to use it.

Such radios promise to offer ground-breaking options to public-safety agencies that have been outspoken in their need for more radio frequencies when emergencies arise. With so many first-responder organizations, virtually all spectrum in the public-safety bands has been allocated, but effective cognitive-radio systems could allow agencies to access the airwaves of neighboring agencies without threatening the functionality of the licensed agency's system.

Fred Frantz, chair of the Software-Defined Radio Forum's public-safety special interest group and director of law-enforcement programs for L-3 Communications, says these spectral arrangements could be an extension of mutual-aid pacts that are commonplace in the public-safety sector.

“Agencies have agreements to share all kinds of things — personnel, equipment and things like that,” Frantz said. “If we could get sufficiently flexible and smart radios, why couldn't we add spectrum to that list?

For all the excitement that cognitive-radio theory has generated at conferences, enthusiasm typically has been dampened by the fact that the enabling technology historically seemed to always be at least five years away from being applicable. But a military-sponsored program has delivered test results that indicate dynamic spectrum-access networking can work, and it has offered a road map for developing affordable radios in the near future that could make the technology economically viable in the public-safety sector.

In December 2006, the Defense Advanced Research Projects Agency (DARPA) conducted tests that demonstrated spectrum-accessing technology from Shared Spectrum enabled radios to access — and relocate to — even temporarily fallow spectrum so quickly that the action did not hamper the incumbent licensee's operations. Of particular note, the tests showed that neXt Generation, or XG, radios vacated spectrum in less than the 500-millisecond threshold set by the military to accommodate traditional walkie-talkie usage.

“We feel that is a good number for an application like push-to-talk,” said Preston Marshall, a program manager in DARPA's strategic technology office. “When you push a transmit button on a radio, most radios don't even modulate for a couple hundred milliseconds, because it takes the oscillators a while to stabilize.”

Marshall said the test results were so positive that DARPA is pursuing use of the dynamic-spectrum-access technology in two ways. First, the organization plans to upload the cognitive-enabling software into existing software-defined radio handsets manufactured by Thales and Harris used by the military — with results expected to be released to the public in March 2008, he said.

But the Thales and Harris radios are expensive, costing about $20,000 per unit. In an effort to lower the price tag, DARPA also has initiated the Wireless Network After Next (WNAN) program, which is designed to lower the cost of cognitive radio and spotlight its functionality for the military.

Among other things, the program plans to drive the development of cognitive radios that cost $500 or less. M/A-COM, with Shared Spectrum as a partner, has been chosen to manufacture the low-cost WNAN handsets — boasting four independent transceivers and capabilities like Internet access, server-like storage and chat functionality — with the first prototypes scheduled for delivery in June 2008, Marshall said.

For industry observers familiar with the high costs typically associated with military software-defined radios, the idea that cognitive capabilities can be integrated in a device costing just $500 seems counterintuitive on the surface. But the cognitive characteristics allow radios to be built less expensively because they are designed to avoid typical spectral environments — where strong signals exist next to weak ones — that necessitate expensive radio front ends, Marshall said.

“In the past, we built very expensive and very power-consuming front ends [to radios],” he said. “WNAN has a much poorer performing front end, but it uses a cognitive radio to find the right place for a network, so it doesn't need a very good front end.

“The reason we'll be able to build an inexpensive radio is that, instead of trying to make it bulletproof in all environments — which costs money and energy — we're making it adaptive to find the environments in which it can work well,” Marshall continued. “And because it brings this whole network with it, all the members of the network can vote to make sure they pick an environment in which all of them can operate well.”

Just as important as the cost savings, the use of cognitive radio enhances energy efficiency. Although DARPA's current goal is to keep battery life comparable to existing military radios, Marshall believes the cognitive devices eventually will be much more energy efficient and have better battery life than today's handsets.

“In a particular situation, 10% of the spectrum might be unusable because of front-end overload,” he said. “That would be unacceptable in a normal radio, so you'd have to spend 100 times more energy to get rid of it. In our radio, we just say, ‘OK, those are the 10% of the bands I can't use. I'll use the other 90% for my networks, and we'll pick them.’ So that reduces the energy demand on the radio by orders of magnitude.”

Using a boxing analogy, Marshall said the radios work smarter instead of using “brute force” to address the challenges of wireless communications. “In the past, we had to build radios that kind of had to take the punch, if you will,” he said. “We're teaching the radio how to duck — and taking a punch takes a lot more strength than ducking.”

In the U.S., virtually all spectrum has been allocated, but military studies indicate that only 6% of it is occupied at any given time, so there are plenty of opportunities for appropriately flexible cognitive radios to find frequencies that can be used, especially if the crowded commercial cellular bands are avoided.

One potential way to use cognitive-radio technology is to utilize fallow spectrum between television channels. Because different channels are used in each city, the so-called “white spaces” that are available in each community differ. However, a recent FCC test of a Microsoft device showed that the unit failed to abandon used spectrum quickly enough to avoid hampering neighboring TV signals. This event has been a “public-relations nightmare” for the commercial side of the equation, said Peter Tenhula, Shared Spectrum's vice president for regulatory affairs and business development.

Some have suggested that geolocation technologies like GPS be used to complement the natural sensing capabilities of cognitive radios, but companies like Philips have argued that such a tactic would add too much cost to devices, Tenhula said. Shared Spectrum believes such geolocation technologies may not be needed.

“We always thought that sensing alone wouldn't work, but we've done some measurements, and we think it can,” Tenhula said. “You can tell where you are, if you have a sensitive enough sensor.”

One potential concern for public safety is that some of its wireless signals — for example, those generated by surveillance equipment — are designed not to be detected, said John Powell, chairman of the software-defined-radio working group for the National Public Safety Telecommunications Council (NPSTC). As a result, a cognitive radio may determine that the spectrum is available for use, and in doing so unwittingly interfere with the cloaked signals.

While DARPA's work is not designed to address domestic public-safety needs, Marshall believes the communications challenges faced by first responders in an emergency and those encountered by the military are similar, both in terms of the environment and the mission-critical nature of the information being transmitted.

Despite the concern, Frantz said he believes cognitive radio offers several potential benefits to first responders in the field, particularly in the ever-changing mobile environments typically associated with the largest incidents.

“When you have fixed infrastructure, you may have dead spots, but everybody knows where they are — you deal with that operationally,” Frantz said. “In the aftermath of Katrina, you didn't know where the dead spots were because they changed every day. So you had people in a coverage area that was fine yesterday, but they moved a tower and now they're out of coverage. This kind of [cognitive] capability could be beneficial in those kind of situations.”

Powell agreed, but noted that he believes public safety likely will be conservative in adopting cognitive radio for mission-critical voice. However, he also said that he believes cognitive radio is a logical alternative to commercial networks for tactical communications — for instance, between a mayor, city manager and a fire chief — when commercial networks are unavailable during an incident, especially if a handset can be produced for less than $500.

“The bottom line is, if they can get something like that working, there will a huge market on the public-safety side in some areas, for administrative support and stuff like that,” Powell said. “If we can come up with some technology in the $500 arena so we're not tied down to the network losses that we suffer with cellular every time there's a big incident, that would get used right now. … Public safety will be all over that in areas where it makes sense.”

Marshall said he is confident the DARPA program can demonstrate that a cognitive radio can be built for $500, expressing more concern about the network-management techniques and technology needed to keep communications running smoothly in a battleground environment.

“I don't have any questions about our abilities to build a radio,” Marshall said. “The hard work over the next three years is to do the network — to do things that we used to do statically and very conservatively and do them dynamically every couple hundred milliseconds. That's going to be the challenge part and the fun thing to watch.

“I don't see any reason why it can't be done,” Marshall added. “And the returns for doing it are incredible, because we'll lower the price of delivering communications to the edge by factors of 10 and 20.”

Building a better radio

XG, or neXt Generation, radios demonstrated reliable networking without harming legacy nodes in dense spectrum environments.

Metric Threshold Results
XG causes no harm
Abandon time 500 msec 100% in 465 msec
Interference limit 3 dB Mean: 0.16 dB; Max: 0.49 dB
XG works
Net formation 30 sec with 6 nodes 90%: 3.6 sec; 100%: 8.68 sec
Net join 5 sec 90%: 2.07 sec; 100%: 4.36 sec
Net re-establish 500 msec 100%: 165 msec
XG adds value
Spectrum occupancy 60% with 6 nodes 85% with 6 nodes
Source: DARPA