At a time when news of advanced technologies seems dominated by the likes of robots, nanogadgets and supercomputers on a chip, the subject of radio might seem a tad boring. But software-defined radios and cognitive radios hold promise for making our wireless networks far more powerful and useful.
Andrew Lippman, who leads the Viral Communications Group at the MIT Media Lab, sums it up this way: "The real core of the idea is not to regard radios as individual, discrete units, but as members of a community."
His idea stems from several related concepts. One of them, software- defined radio (SDR), is all about replacing much of the hardware in radio frequency devices -- cell phones, GPS units, wireless laptops -- with software, giving them unheard-of capabilities. For example, an SDR could morph from a cell phone to an FM radio receiver to a GPS receiver as directed by its user or even a remote party.
Cognitive radios go a step further by being aware of their environments. The most important application in the near term is to enable a wireless device to automatically and rapidly hop from frequency to frequency looking for uncongested slices of the radio spectrum. And cognitive radios can learn: Your GPS-equipped cell phone might remember that it has encountered dead spots along a certain street and then seek an alternate communication path the next time you are there.
Lippman says cognitive radios in a local area can work on an ad hoc basis with one another rather than taking direction from a central router, server or cell tower. They can "vote" among themselves -- dynamically, and based on local conditions such as hot spots and dead spots -- to determine the clearest and most power-efficient route for a signal to travel. The radios act like bucket brigades for messages, and the closer they are to one another, the lower their power requirements are and the less interference they generate. And since the radios add spectrum capacity even as they use it, the network "can scale almost without bound," Lippman says.
He calls this kind of arrangement "a virtuous cycle," which explains this way: "An analogy is a wedding -- as more people attend, there are more conversations, but each one is softer, so they don't interfere. Most current systems are a wedding where everyone uses the PA."
Lippman is working on another application, called "fluid voice." "Think about it as a party line where all the phones in an area can hear what anyone else is saying," he says. "You have phones now where you push to talk, but this is push to listen. It allows you to think about communications in a different way." Firefighters responding to an emergency could use fluid voice so if one of them got into trouble, he could "shout" a call to all of his colleagues without having to pause to tell the device who he wanted to contact.
But users of fluid voice would also be able to communicate with just a few other users on the party line. "Think of the user interface as being a screen with a dot for everybody you could be listening to," Lippman says. "The closer to the centre you move the dot, the louder they become. If there is somebody you don't want to hear, you push them to the edge."
Joseph Evans, a computer science professor at the University of Kansas, is involved in research in a similar area. In his CogNet project, he hopes to develop open-source radio protocols that can be used to build cognitive wireless networks that are adaptive and agile. For example, Evans says, such a network might work in broadcast mode to warn residents of a city about an approaching hurricane. After the hurricane strikes, the network might quickly morph into one that provides a dedicated secure channel for emergency workers.
While cognitive radio works at the physical layer -- by varying things like frequency and modulation -- CogNet would span layers of the communication protocol stack (see diagram). Explains Evans, "The application may say to the network layer, 'I want you to be a broadcast channel,' and the radio then tells the physical layer to make my cell phone network behave like it's FM radio. In other words, ring everyone's phone and tell them to get out of the way of the storm."
The Moore's Law-driven ability to put a lot of processing power into small, cheap devices, coupled with the replacement of hardware with software, opens up many promising new paths for radio, Lippman says. "For years, we have tried to make the simplest possible receiver so people could afford them. So we piled all the intelligence into the transmitter," he explains. "But now the whole engineering basis has turned around. It's suddenly feasible to put lots of processing in consumer radio."
Not so boring after all.