June 10, 2005

Roadcasting: A Potential Mesh Network Killer App

Terrestrial radio may soon face yet another challenge with user-created radio broadcasting from any car on the road.

By Eric Hellweg

The Quicktime-formatted promotional video for the roadcasting project begins with a black screen that quickly dissolves into a still black-and-white image of a driver sitting behind the steering wheel, wearing a look of frustration and boredom. The narration begins: "Everyone has experienced the headaches of FM radio. There's the endless commercials, the same old songs over and over again, and the difficulty of finding something that you want to hear. Welcome to the next generation of radio: roadcasting."

The terrestrial radio industry is already fighting a multi-front battle with the ascendant satellite radio business and nascent podcasting community. Now it has another technological innovation to worry about: roadcasting.

The concept was created by a team of five students at Carnegie Mellon University. Their Human-Computer Interaction Institute Masters Program project, which was sponsored by General Motors, according to a company spokeswoman, combines three hot areas: ad hoc (mesh) computer networks, personalized digital music, and open-source software development. While the hardware elements -- the network devices, the touch-screen interface, and the stereo component -- have yet to be created, the working software application is currently being picked over by open-source enthusiasts around the world.

The most straightforward use for the software enables people to create their own personal radio stations -- playlists -- and store them on an in-car stereo hard drive. The real innovation, though, comes from what happens once a playlist is created. While a driver is listening to music from his or her choices, the songs will be broadcast and available for reception by any other car with a roadcast-equipped car stereo. So, if a driver gets bored with a personal playlist, the software's collaborative filtering capabilities will automatically scan the airwaves looking for other roadcast stations that match the driver's stated preferences, and return any matching available stations. Listeners can search by bands, genres, and song titles, and skip through other users' radio stations to find music they want to hear.

Say, for instance, Heather gets into her car in the morning and decides she wants to hear some death metal on a traffic-congested commute. Using a touch screen (the method roadcasting's designers believe will work best for both convenience and safety), Heather can select her favorite songs and bands from her existing song library (stored on a car's hard drive or a digital music player docked into a stereo). Heather then selects an icon for her radio station, names it, and heads off to work. Meanwhile, John, a fellow frazzled commuter and death-rock aficionado, is on the road, and his roadcasting system alerts him that a new station with similar music is in range. John can then select to listen to the station, viewing the name, icon, and songs available. If he would rather hear a song on Heather's list that isn't currently playing, John can fast-forward to that song and play it.

The arrival and success of roadcasting, though, won't happen until further development of mesh networks, which are created through small, inexpensive sensors, typically attached to other devices, with little intelligence other than the ability to connect, reconnect, and conduct an application.