Cognitive Radio Prepares for Action

April 2008
By Henry S. Kenyon

 
The Defense Advanced Research Projects Agency’s (DARPA’s) Wireless Network after Next (WNaN) program is developing wireless nodes and adaptive network technologies to permit the creation of self-forming, ad hoc networks. WNaN tech-nology would allow U.S. military forces to deploy and communicate with each person and device across every echelon.
Smart networking system navigates frequency selection without human intervention.

An experimental radio technology could provide U.S. warfighters with assured access to voice, data and video communications. The prototype systems use an advanced wireless networking capability to link troops with larger networks such as the Global Information Grid. The radios also are capable of sensing the electromagnetic environment and selecting frequencies that are not in use automatically.

Maintaining connectivity on modern battlefields has become challenging with the introduction of new communications and jamming technologies. These systems often work within close proximity to each other, putting pressure on frequency allocations in an already crowded spectrum band. A new communications system offers a way to provide individual soldiers with vital information while navigating through a thicket of signals and data transmissions.

The goal of the Defense Advanced Research Projects Agency’s (DARPA’s) Wireless Network After Next (WNaN) program is to create a flexible architecture for military communications (SIGNAL Magazine, July 2006). A key aspect of this effort is to develop and test an inexpensive handheld radio capable of selecting its own frequencies and forming small networks within a larger battlefield network.

The WNaN program is an extension of DARPA’s successful Dynamic Spectrum program that focused on solving many of the spectrum management and access issues affecting the U.S. Defense Department. Preston F. Marshall, WNaN’s program manager, Arlington, Virginia, shares that the effort is taking a different approach to dynamic spectrum allocation. At the beginning of the program, researchers questioned industry firms about the factors that make military radios expensive. Marshall says that this discussion allowed scientists to understand some of the intractable performance and cost problems of building high-performance communications systems.

Citing lessons learned in DARPA’s Next Generation (XG) program (SIGNAL Magazine, December 2003), Marshall maintains that the goal is not to develop a radio that can work well in any arbitrary environment, but to develop one that can adapt and select the environments in which it works best and bring other similar radios into its network. “We use XG, not just to pick frequencies, but to pick environments in which the radio is not overly stressed or challenged. Then we can start to attack some of the driving economics of high-performance military communication,” he says.

In developing WNaN, scientists explored two aspects. The first was determining whether affordable nodes could be built, and the second was to find out whether it was possible to create a network capable of adapting and scaling into hundreds or thousands of nodes. When the program was launched, Marshall notes that it was understood that if the network was built, there must be hardware to exploit it.

DARPA issued a broad agency announcement to determine the kind of radios competing firms would build. Marshall explains that the only requirement the agency insisted on was that the WNaN radios be able to find available areas in the spectrum and that they have four transceivers/channels in each handset. The multiple transceivers overcome network issues such as interference or overload by switching among the three other networks the node is connected to. “This is a core principle of the Internet—never be dependent on any one communication path,” he says.

The WNaN program’s network development phase was launched in mid-2007. Marshall notes that this effort is an extension of DARPA’s research into cognitive radio. Although the path is evolutionary, he hopes the technology’s effect is transformational. Besides solving spectrum issues, XG technology is a critical component for cognitive radio because it allows designers to compromise on some aspects of hardware performance.

DARPA researchers determined which technical aspects could be solved by hardware and which were limited by physics. Marshall admits that physical laws constrain certain aspects of radio development, such as front-end linearity and amplification of some wave forms. This limitation means that a cognitive radio will not use certain waveforms unless it is absolutely necessary. He adds that these considerations provide power and cost savings and permit the use of commercial parts in the system.

Marshall shares that several manufacturers are developing radio frequency integrated circuits for commercial worldwide interoperability for microwave access, or WiMAX, systems. Each of these inexpensive microchips is a super heterodyne receiver. But these receivers do not have the performance typically associated with a military radio. “With WNaN, we think we can build a military network that can have the same reliability but run across these relatively less robust chip sets,” he says.

The WNaN radio also is known as the $500 radio. The term refers to the targeted cost of each individual unit when it is in full production. The $500 radio is designed as a handheld unit that can be worn on a soldier’s hip or carried in a rucksack. By the end of the program, the goal is to have reduced the device’s size to that of a large cellular telephone or personal digital assistant. Marshall says that after evaluating the devices, the services will provide feedback about the need for additional features such as data displays. The program has two cycles in which to include feedback from the field to modify the devices. “Our interest in the technology was how we could apply networking adaptation to allow us to use lower cost, lower performance components and still achieve very high quality military communications,” he says.

 
 
A key objective of the WNaN program is to develop inexpensive handheld radios. Designed to be manufactured in large numbers, the radios would form the nodes in the WNaN network. Each radio would have four channels, operate in the 900-megahertz to 6-gigahertz range and feature commercial components such as this microchip designed for use in cellular telephones. The program’s goal is to operate networks with tens of thousands of nodes.
Although DARPA scientists built the radio as a platform on which the WNaN network could operate, Marshall notes that the platform has itself become very attractive. He is confident that by July, the effort will begin delivering operational prototype radios that will be developed over the next two years for large-scale production. “When we say $500 radio, it’s not just what we think of today as a radio. Each of these nodes actually has four completely independent transceivers inside it. Each transceiver tunes 900 megahertz to 6 gigahertz, so it’s really a bundle of four individual radios, all in one handset,” Marshall says.

The radios also can operate DARPA’s disruption-tolerant networking capability. This function allows the devices to store, cache, index and observe content. Marshall speculates that the radios will change the concept of edge communications. “Instead of the edge just being a pipe to the core, it [the radio] will show that we can actually build networks that operate at the edge and in a completely self-sufficient manner,” he shares.

But the components for the radio are still being developed. Marshall describes them as “scattered across many laboratory workshop benches” with some devices ready for delivery, while others are still being made or modified. The first step will be to import networking technology already developed by previous DARPA programs, such as XG, and self-forming networks. These networks will be imported into the radios. Some 40 radios are scheduled to be ready by September. Marshall notes that this is a large number of devices for use at the launch of a DARPA program. The $500 radios will then be offered to the services for testing.

The heart of the new radio is its front end, which consists of four very sensitive tunable radio frequency filters. Marshall notes that the filters are necessary because the wireless communications environment that U.S. forces operate in is different from that of even a decade ago. More systems are being added to the spectrum, and warfighters are fighting in the same frequencies that are actively jammed by friendly forces. “All of that is just pumping energy into the RF [radio frequency]. The ability for systems to survive in that [environment] is a lot tougher than it was a decade ago,” he says.

Marshall describes the difference between the WNaN radio and a typical military radio front end as that between a boxer who is good at ducking and one who takes the blows. “We want to build the software that makes it duck. We don’t want it being slammed by all this energy. We want to find a place where we’re not being slammed,” he says.

The $500 radios are the first purpose-built cognitive radios, Marshall declares. Instead of putting cognitive software into an existing radio, he shares that DARPA engineers designed the radio around challenges that can be solved with cognitive techniques. “If we can provide on every soldier’s hip the kind of connectivity that he or she would get at home on a cable modem—the same kind of bandwidth, the same kind of capability, the rapid access to content and the ability to initiate content—it could change everything,” he says.

Traditional military radios must be robust to operate in any environment in which a frequency is assigned. By comparison, the front end of a WNaN radio may have only one-hundredth of the energy handling ability of a typical military radio front end. Marshall says that overload conditions occur only in a very small part of the spectrum, with the other 95 percent of the spectrum usually being available. A WNaN radio can operate in 95 percent of the spectrum and bypass the congested five percent. “Had you done that same front end in a traditionally operated radio, that would mean that five percent of the time it didn’t work, and that would be unacceptable,” he contends.

In a WNaN-based system, this high-spectrum availability means that 99 percent of the time the radio will find a frequency to operate on. This efficiency allows designers to decrease their front-end requirements. Similarly, radio engineers design traditional amplifiers to operate in worst-case multipath conditions. But Marshall notes that these conditions do not occur most of the time. The program is seeking to develop waveforms that can adapt to locating clear channels and to using inefficient channels only in rare situations.

The program is shifting to solve other problems. Marshall says that some of these challenges require either very expensive solutions or large amounts of battery power. He adds that military radios use much more battery power than commercial cellular telephones to power all of the linear systems. Instead, DARPA engineers are shunting these communications issues to the network. Cutting linearity in a radio also reduces battery power requirements significantly. “If the receiver eats most of the energy in a modern radio, the receiver energy consumption essentially scales with the linearity. If we can reduce the linearity, we can reduce receiver energy by a factor of 10,” he says.

Marshall explains that the radio is basically an Internet protocol (IP) device. The only feature that is not IP based is voice communications, which functions in a manner similar to asynchronous transfer mode, allowing the radio to tunnel voice through the network without much latency. He says that this function permits all of the IP-based capabilities being developed for the Global Information Grid to port via Ethernet to the radio.

The WNaN network architecture is still under development. Marshall notes that the final boundaries of the network will be determined by the services. One goal for the final phase of the program will focus on the wireless network. If the radios are deployed in the thousands, how will the WNaN network interact with the core? “If this were Iraq, you could imagine wireless networks all surrounding the Green Zone and how they interact with that infrastructure,” he says.

In the program’s final phase, DARPA will use 500 to 1,000 radios to demonstrate that they can collectively conduct the position reporting and data support capabilities of a single network device. When a network of thousands has been established, engineers will study how it relates to the Defense Department’s core infrastructure, including the Future Combat Systems. A key challenge will be to scale the network from thousands to millions of radios while maintaining stability. By putting four receivers in every radio, developers create four networks that can operate independently to prevent ripples through the network. But proving stability is key.

However, if the program is successful, Marshall hopes that the $500 radio will be obsolete in several years. By developing the right networking technology, the WNaN system will permit new radios to be introduced every few years. The system will work much like the cellular telephone industry with the communications devices riding on the network as commercial wireless devices currently do. “If we’ve done our job right, we’re going to have a network that is so flexible that the Department of Defense can go out and tell industry to provide its best ideas every couple of years. We buy them, and when we’re done, we throw them away. It won’t be that we build this radio and people have to live with it for the next 20 years,” Marshall says.

Web Resource
Defense Advanced Research Projects Agency: www.darpa.mil