Wireless to the Nth Degree

July 2006
By Maryann Lawlor
E-mail About the Author

Wireless capabilities already have become an integral part of military operations. U.S. Marines practice foot patrols with a mounted data automated communications terminal, a tactical computer that connects to an enhanced position location reporting system for wireless Internet access. Using the system, Marines can communicate with each other while in convoys in current operations.
Smart networks linking inexpensive devices will redefine network-centric warfare.

Envision a future filled with millions of wireless nodes connected through a smart network that automatically adjusts to optimize communications performance. Achieving this reality would require developing low-cost devices and mitigating current weaknesses in networking technology. However, when this vision is realized, troops will be able to infiltrate areas devoid of communications infrastructure yet stay in touch with each other and platforms in the battlespace.

The project that seeks to attain this goal is called the Wireless Network after Next (WNaN) program and is underway at the Defense Advanced Research Projects Agency (DARPA), Arlington, Virginia. It comprises two elements: the device and the network. Phase one research focuses on the device and is called the Wireless Adaptable Network Node (WANN).

According to Preston Marshall, program manager, DARPA, the ambitious undertaking will develop and demonstrate technologies and system concepts that enable intelligent adaptive wireless networks. Hundreds, thousands, even millions of low-cost wireless devices will be connected through networking technology that adjusts its topology and the nodes’ operational mode to reduce the demands on nodes, particularly on the physical and link layers. Working as a collective, a multitude of devices can be more powerful than traditional networks that comprise a smaller number of large systems, he contends.

Marshall says the program came about because DARPA observed two developing issues. The first was the need to build significant room for adaptation into military networks. The agency observed this in its NeXt Generation, or XG, program (SIGNAL Magazine, March 2002), a project it is working on with the U.S. Air Force Research Laboratory, Rome, New York. The project is developing both the concepts and enabling technologies to utilize spectrum dynamically.

“Given that we have dynamic spectrum, that opened the opportunity to be very dynamic in how we form networks and how we form topologies,” he relates. “The basic theme that is running through a lot of our programs now is not to just hit a single performance point. It is to develop technologies that are very adaptable across performance regions and then to develop the technology to understand how to use and exploit that adaptation and still create networks that are stable, don’t thrash and are operable.”

Marshall admits that it is this last piece that is the most difficult to attain. “It’s going to be very hard to have thousands of nodes, each perceiving a different world, adapting to do the right thing and not have them become as if you had a massive parking lot and told everybody to go to the other side at the same time. That is a real technology challenge,” he states.

DARPA’s second observation about the evolution of the wireless world involves the performance—particularly the fundamental limits—of radio frequency (RF) devices and how the devices’ usage is likely to grow. The researchers wondered whether these limits could provide insight into the form of future radios.

To explore this question, they performed a tuner study that revealed real limits to what can be done using highly linear RF equipment. As a result, they determined that they should be exploring strategies that depend more heavily on tunable filters. “We had to move from broadband filters to very narrowband tunable filters because we’re going to create denser and denser RF environments, and we’re going to be victims of the very environments we create,” Marshall relates.

To illustrate the challenge that massive numbers of RF devices pose in the military sphere, Marshall compares it to the commercial cell phone industry. However, one big differentiator between military and civilian networks is that the commercial world features large, very smart and very expensive infrastructure and relatively low-cost user devices at the edges—for example, cell phones and personal digital assistant devices. “In the military, we know we won’t be able to build that infrastructure in advance of our operating [in a location], so we have to be ‘infrastructureless.’ As a result, we have to put all that intelligence and all that management that is in the infrastructure—in the cell towers—in every device,” he explains. “This is the challenge, and it’s a tough challenge because it’s a mix of totally understanding the realities of realizable devices and at the same time bringing to bear very, very complex self-organizing computer science.”

The goal of the WANN program is to demonstrate that a huge number of affordable devices can be built. If the DARPA team accomplishes this objective, it will begin examining the networking requirements to connect all of them.

Building these devices in mass and at a reasonable price is important to the military for a number of reasons, Marshall allows. First, based on the tuner study, it is evident that raw analog performance is limited because as the number of devices using spectrum grows, the amount of energy in the receivers also significantly increases.

Lance Cpl. Bradley E. Ellis, USMC, multichannel radio operator, fixes an antenna receiver assembly on a troposphere satellite support radio. The radio allows units throughout the air base at Al Asad, Iraq, to maintain wireless communications with the tactical command center.
The current number of RF devices in use already is causing some problems, he observes. “Today, we have both in military and in civilian practice lots of problems with co-site. Frequency managers worry about deconflicting co-channel because they’re using the same frequency. But more and more we see issues with adjacent channel [problems] because it’s energy put in the front-end that causes it to not perform well and desensitizes the radio,” Marshall relates. Adaptation may be the only way to resolve these conflicts, he proposes.

One incidental benefit of smart, adaptable networks would be the elimination of the need for highly complex radios, Marshall maintains. Under unfavorable conditions, the radio would adapt to find conditions in which it could work. Today, without dynamic or spur-free range, radios assigned a frequency simply fail. “In the WNaN framework … we would use adaptation instead of spending a lot of money on very, very high-performance hardware,” he says.

Although a lot of research has occurred in the area of identifying a topology that works well, Marshall admits that there is no real theory about how to direct a device to choose the best topology, and that is what much of DARPA’s work will be about. Among the challenges is determining the speed at which adaptation can optimally take place. “These are profound mathematical problems as much as they are programming ones,” he points out.

Because this approach depends on purchasing a large quantity of devices, cost is an important factor in its success. DARPA has set a price ceiling of $500 per unit, and Marshall does not believe this is an obstacle. The commercial sector regularly builds very low-cost RF equipment; however, most work on a single frequency with a single service. The agency is looking for industry to design a radio that features many modes, works on many frequencies and can be combined as flexibly as possible. “The capability of any one frequency may not be different, but the fact that it’s highly flexible across a broad range makes it unique from what’s been classically done in the commercial world,” he explains.

Marshall points out that while the price tag may not be a hindrance, size and weight are critical in the design of the radios. “If I gave you an iPod and it weighed 50 pounds, you couldn’t use it the same way as you do a 5-ounce iPod. We want to not only learn about the technology but also to get some experience with how a dense network would actually be used. We want to be able to understand how it behaves because our experience has been with relatively sparse networks, and we want to push the network down so that every device, every vehicle, every person has a continuous connection. We need to understand how they’ll use it, and to do that we have to give them something that is reasonably form factor,” Marshall says.

Once a device has been developed, the DARPA team can concentrate on the second component of the WNaN program, the networking part, which Marshall says is where the real “secret sauce” is. “We know we can’t build as good a radio when we constrain it, but we think we’ll end up with a better network,” he contends.

Although the WNaN is a DARPA program in and of itself, Marshall relates that it will bring together many other agency projects that have been developing specific technologies. Many of the technologies work well in specific environments, but the challenge WNaN researchers are wrestling with is how to look across all of them and determine how to optimize the benefits of each in one environment, he notes.

And achieving this synergy of programs calls for a fusion of another kind: cross-discipline collaboration. To facilitate this cooperation, DARPA conducted a workshop that focused on knowledge-based networking. The purpose was to bring the artificial intelligence and wireless networking communities together so they could learn from each other and pool their expertise. “We want people across disciplines that haven’t historically worked [very] collaboratively together because we really see networking and creating thousands to millions of devices as a fundamental problem and a big technology challenge,” Marshall states.

Creating these kinds of teams can be a lengthy process, he acknowledges. Examining different paths and various aspects of computer science and physics is what is needed, and no single specialty will be able to provide a solution for the devices’ design. The areas of networking, biology, cognition and many other fields “where we haven’t even scratched the surface to solve the networking challenge” may hold answers, Marshall speculates. “Ultimately, I hope there is a commercial interest, too. If we can find new ways to do networking—and networking is clearly a big piece of the economy and there are a lot of resources in it—and we can find new techniques, it ought to be of interest to the commercial world as well,” he states.


Web Resources
Wireless Network after Next program: www.darpa.mil/ato/solicit/WNaN/index.htm
Wireless Adaptable Network Node: www.darpa.mil/ato/solicit/WANN/index.htm
DARPA NeXT Generation Program: www.darpa.mil/ato/programs/XG