Cognitive Radio Adds Intelligence to Wireless Technology
Radio communications are becoming increasingly complex as more devices — including laptops, cellular phones, and even sensors — compete for limited bandwidth in various frequency ranges.
In addition, devices must conform in a growing number of ways to user needs, corporate policies, and government regulations. Also, users often want to utilize their devices with more than one wireless technology, such as for both cellular and Wi-Fi communications.
Proponents say cognitive radios — currently the subject of considerable research and early implementation — appear to answer these challenges.
CR is an emerging radio approach in which transceivers are combined with sensors, intelligence, and adaptability. Proponents say these capabilities could enable radios that monitor transmissions and the network environment and change parameters such as frequency, energy-usage levels, and protocols to adapt as necessary.
Researchers say improvements in and better integration of CR systems' core elements will lead to fundamental advances in radio communications.
"It is the only new idea in RF communications," said Preston Marshall, director of the University of Southern California's Information Sciences Institute (ISI) and formerly director of DARPA's CR program.
"Most of what we do is evolutionary and only advances the technology by a few steps," he explained. However, he said, CR could make huge improvements.
There are no full CR systems yet, but technologies such as Wi-Fi are starting to use basic, limited CR capabilities, said Kingston University senior lecturer Christos Politis, chair of the Wireless World Research Forum’s (WWRF’s) Working Group 8, which addresses spectrum-related topics.
Proponents envision CR advancing important technologies such as femtocells, which are small base stations that improve network coverage and capacity, as well as military and emergency-services radio interoperability.
Nonetheless, the technology must address many challenges before it can achieve mainstream status.
Why Cognitive Radio?
Mobile devices are dealing with an increasingly complicated wireless environment. Many devices — including powerful laptops, less powerful cellular phones, and even low-power sensors — are trying to make the most of available battery power and limited bandwidth in various radio-frequency ranges.
In some heavily used frequencies, signal interference is a problem.
Some wireless-service providers set up policies, such as the maximum power level that devices could use, that customers must follow.
Government regulations also limit the frequency bands and communications technologies that radios can use, and other aspects of wireless transmissions.
Using one device — rather than multiple ones — to meet all these needs is more economical.
Moreover, it is highly efficient to use a single device that can adapt in the middle of a communications session or work with radios that use different technologies when necessary. This would help military and emergency-services agencies, which don’t all use the same wireless technology.
Proponents say CR is a way to address these challenges.
The term "cognitive radio" was coined in a 1999 technical paper by Joseph Mitola, now vice president for the research enterprise at the Stevens Institute of Technology. At the time, he was researching communications for the US Department of Defense.
His pioneering research was designed to help the military enable communications among its 297 families of radios and 42 air interfaces.
CR is an evolution of the software-defined radio (SDR), a term that Mitola also coined. SDRs use software in place of individual components traditionally implemented in hardware, such as amplifiers, modulators, and demodulators. That way, instead of having to use separate radios for different standards, spectrum ranges, and capabilities, users could work with a single radio that supports multiple parameters.
However, SDR allows only for adaptability. It doesn’t sense what other users are doing in a spectrum range or include the ability to make link-optimization decisions.
According to Mitola, CRs add cognition and the ability to make adjustments to the radio based on algorithms in a cognition engine.
CRs can function without SDRs. However, SDRs give CRs more flexibility in choosing communications options.
Getting to Know Cognitive Radio
CR technology works within a standard radio-based wireless network consisting of transmitters, receivers, and antennas.
CR systems use an RF sensor, which is part of the receiver, for awareness of radio-spectrum usage. These sensors detect where spectrum is being used and find available frequencies suitable for communications
CR works with a cognitive engine to recognize if a wireless channel isn't being used. And they frequently use a policy engine to determine if the radio is allowed to use that channel.
CR augments RF awareness via a database of licensed users of the slices of spectrum being analyzed, as well as their geographic coordinates. This would detect spectrum users that aren't working via RF, such as TV broadcasters.
To enable dynamic spectrum access and create the best link while minimizing interference with other radios, CR uses a set of decision-making algorithms running on a processor within the system to change RF frequencies, protocols, or modulation techniques.
To improve a link, CR systems must understand user needs, such as quality of service requirements for various types of transmissions, whether battery life is more important than high throughput, and what protocol and frequency the intended recipient is using.
The systems also must recognize the RF environment's physical properties such as signal interference and attenuation.
CR uses these capabilities to adjust factors such as signal power, communications protocol, and spectrum.
CR techniques are attracting interest in various areas.
The US military and DARPA are pursuing CR implementation, particularly for interoperability among radios using different technologies and spectrum ranges.
In 1998, the US Department of Defense spelled out the requirements for the SDR-based Joint Tactical Radio System.
As part of JTRS, the DoD and DARPA designed the XG (next generation) radio program to use CR to enable a radio to sense unused RF spectrum and change frequencies in case of interference.
BAE Systems, Harris, and Thales Communications are now developing and beginning to deliver XG-based dynamic spectrum radios for military use.
DARPA has launched the Wireless Network after Next program to build on XG. WNaN would offer better spectrum sensing, decision making, and adaptability, as well as allow greater interoperability between radio systems.
WNaN is also exploring disruption-tolerant networking, which would let radios maintain connections despite unreliable links; and DARPA Interference Multiple Access, which exploits signal interference to enable higher throughput and capacity.
TV white space
In the UK and the US, regulators are considering opening up TV white space — frequencies allocated to a television service but not used — for unlicensed wireless communications applications.
CR technologies are expected to play a significant role in this. For example, the FCC envisions that equipment makers would employ a combination of CR-based spectrum sensing capabilities and a database of licensed broadcasters to ensure that new users don’t interfere with TV signals.
Wireless-spectrum provider Spectrum Bridge has demonstrated this approach in the US (see www.computer.org/portal/web/computingnow/archive/news040).
The United Kingdom's Office of Communications has determined that 90 percent of the UK has available white space that could be used for other communications services.
Femtocells and cellular agility
Cellular operators — including AT&T, China Unicom, Sprint, Verizon, and Vodafone — have deployed or plan to deploy femtocells to homes and businesses to improve cellular and data communications’ speed and coverage.
The proliferation of femtocells and the signals they transmit can create interference and thus spectrum-usage problems.
Service providers and equipment makers are now looking at ways CR technology could improve femtocell operations by providing the intelligence that would let them reduce interference with and from other devices, according to Mitola.
Other CR advances
Vendor Adapt4’s XG2plus is a CR system that uses cognitive frequency management. This capability lets telemetry applications or supervisory control and data acquisition (SCADA) systems — typically used in industrial plants — to automatically change to one of 10 alternate channels upon detecting interference.
However, the product operates only in the 217-to-220-MHz spectrum range and thus doesn’t offer the broad coverage that full CR systems would provide.
Vendor Drakontas and Drexel University have developed the DragonForce tools to leverage CR capabilities.
The tools would let a computer create a visual representation of an emergency situation — including elements such as the nature of the incident and the location of responders and victims — using information transmitted from the equipment of multiple response agencies. This would work even if the equipment uses different technologies.
DragonForce would also enable the equipment to interoperate with one another.
CR and its uses are so potentially complex that the technology faces multiple challenges to reaching its potential.
ISI's Marshall said the biggest barrier CR faces in reaching its potential is changing the way businesses and regulators think about spectrum.
Today, regulators and users leave large amounts of frequency unused to avoid interference. However, this significantly reduces overall available network and spectrum capacity.
Instead, Marshall argued, they should create policies for acceptable interference levels, and vendors should build radios that can adapt to the noise.
To yield maximum benefits, CRs must be used by many devices and need the capability to quickly recognize what is happening in a network environment, said Thales Communications staff engineer Bob Masucci.
A big challenge will be proving the technology's benefits in terms of increased reliability and resource utilization, better response time, and reduced operational expenditures to the managers who would decide whether to buy CR equipment, said Panagiotis Demestichas, associate professor at University of Piraeus and chair of WWRF Working Group 6, which addresses cognitive networks and systems for a wireless future internet.
As Politis noted, "Typically people are uncomfortable with technologies they do not understand and believe will cause catastrophic interference to incumbent systems."
The size, weight, energy consumption, and cost that CRs would add are also big challenges, particularly for mobile devices, added DARPA program manager Bruce Fette.
CR-enabled flexibility will mean wireless devices will have to navigate many different choices of access networks and protocols, based on factors such as performance, link quality, power usage, and cost.
Therefore, said Greg Troxel, directing scientist at Raytheon BBN Technologies, an R&D company, CR systems will need the ability to make these choices optimally before CR can reach its full potential.
CR systems have trouble making intelligent decisions when many are devices interacting across a network.
The field also needs more research on making the systems intelligent enough to improve the aggregate throughput across an entire network's nodes.
Researchers are thus exploring the evolution of CR toward entire cognitive networks, which apply intelligence to improving communications across all devices in a network, not just individual links.
Mitola cautioned that as highly configurable CR capabilities become more widespread, hackers could use them to penetrate wireless systems more easily.
Current CR research addresses many areas, including SDR and the use of AI to quickly find the optimal combination of parameters for increasing a link’s or network’s throughput and reliability.
Marshall said 80 percent of CR research is focusing on maximizing spectrum utilization, which is critical to communications providers worldwide.
Researchers at places such as the Stevens Institute of Technology are exploring low-cost CR devices that will function across very broad ranges of spectrum, noted Marshall.
Research into intelligent-radio technology would evolve CR to enable devices to improve their performance by learning which adaptive strategies work best.
The big CR advances will probably come from the military, which has access to wide swaths of spectrum and the funds to invest in new approaches, DARPA's Fette said.
CR technology is not ready yet for the complete flexibility envisioned by proponents, noted Kingston University's Politis.
Thus, he said, some industry observers expect users to adopt a reduced CR version — which operates in smaller spectrum ranges and has less decision making ability — in noncritical applications such as Internet access.
Cognitive-radio technology still has numerous hurdles to clear before it can become popular. In the longer term, proponents expect CR capabilities will be applied to the evolution of networking and wireless telephony.
"At this point, CR is in the demonstration and refinement phase," said Thales' Masucci. "Once the next generation of wideband radios arrives, CR could help improve reliability since it will be harder to find unused wideband channels and they will be more susceptible to interference."
There is already a lot of momentum for CR, said Politis, "but we need to persuade the regulators that a suitable regulatory framework should be created to support this technology."
Vanu Bose, president of Vanu Inc., noted, "The adoption will be slow and in applications, such as TV white space and femtocells, where it is needed first." Once it gains credibility and people see that it is a viable technology, he said, it will be rolled out to broader networks.
George Lawton is a freelance technology writer based in Monte Rio, California. Contact him at email@example.com.