SAN JOSE, Calif. Intel Research is hosting a workshop on cognitive radio this week, one of many efforts around the globe exploring the road toward more efficient wireless networks. Both Intel and Europe's IMEC research group said they see both progress and challenges ahead in cognitive and software-defined radios.
Software-defined radios use some common hardware elements to handle traffic on more than one wireless network. Cognitive radios go a big step further by actually sensing wireless traffic to make intelligent decisions about using spectrum.
Early implementations of both concepts are available, but researchers agreed cognitive radios still face many fundamental challenges before they can achieve their full promise. The recent decision by U.S. regulators to open up so-called white spaces could pave the way for more work on cognitive radios, researchers said.
Intel hopes to gather at its Hillsboro, Oregon, site this week (Nov. 6-7) as many as 100 researchers from companies including Alcatel-Lucent, Marvell, Motorola, Nokia and Qualcomm for its workshop on cognitive radio.
"Getting the full value chain in line is important," said Liesbet Van der Perre, group science director for wireless research at IMEC who assembles partnerships on similar topics in Europe.
Ultimately handset makers such as Nokia and Samsung need to understand and adopt the new techniques for more effective use of hardware and limited spectrum, she said.
"We could hit a brick wall [with overcrowded spectrum], but sometimes it's only when you hit such a wall that people get interested in a solution," she said. "As long as their pain isn't too bad [systems makers] aren't interested," she added.
Researchers are encountering many walls as they look for ways to implement cognitive radio, such as finding effective ways to sense when spectrum is being used or is free, said Kevin Kahn, director of Intel's communications technology lab who is hosting this week's event.
"It is likely that more cooperative sensing approaches among multiple radios can mitigate this in a practical sense, but it's tricky," said Kahn. "Some proposals rely upon 'information channels' that announce bands in use although this approach also has its issues," he added.
Van der Perre agreed.
"There have been some very disruptive ideas in cooperative social networking so no one terminal has to do all the sensing. That's quite interesting but leaves open questions about how you share the data," she said. If some terminals decide to game the system, taking data on free spectrum without sharing it, "the system breaks down," she said.
Overall, "there has been quite a lot of research on the algorithms, but they have not always been well implemented" in hardware sensing engines, she said. "Algorithms for detection range from simple power detectors to complex feature detectors and a wide range of intermediate solutions," she added
IMEC aims to build some prototype hardware for spectrum sensing soon. "Maybe we can resolve some of the open questions by doing some experiments," she said.
Researchers believe cognitive radios could play significant roles in areas such as next-generation WiMax and cellular networks as well as emerging areas such as devices for so-called white spaces, unused spectrum areas recently opened up by U.S. regulators.
Cognitive radios would "have more direct payoff to the incumbent operators and some form of this probably will begin to appear in next generation broadband standards [such as] WiMax 2 and LTE," said Kahn. In addition, the white spaces "may be a good first entry to allow sensible experimentation where the cost of entry of taking the cognitive approach should be much lower," he added.
Ze'ev Roth, chief technology officer of WiMax system maker Alvarion Ltd. (Tel-Aviv), agreed. "We can expect many commercial products [for the white spaces] to use cognitive radios," he said.
"There is always a shortage of spectrum for radio communications, therefore regulators are trying to make more efficient use of the existing spectrum, and cognitive radio has the potential to enable meeting this goal," Roth said. "To make cognitive radio attractive, relatively large chunks of spectrum should be targeted, and regulator licensing fees to mine these bands should be as low as possible," he added.