New ‘Batphone’ App Uses Acoustics to Determine Location

Outside, the global position system allows mobile phone users to pinpoint their location with surprising accuracy.

But indoors, those who are lost are out of luck: GPS satellite signals can’t penetrate roofs.

Researchers at the McCormick School of Engineering and Applied Science have determined one way of figuring out your location inside: by letting your phone listen. Their new mobile phone app, called Batphone, allows users to record ambient noise in a room and tag it with an acoustic fingerprint, which allows future users to use that database of fingerprints to determine their location.

“We have found that the app has been very successful in determining locations,” says app developer Stephen Tarzia, a computer engineering graduate student in the Empathic Systems Project headed by electrical engineering and computer science professors Peter Dinda and Gokhan Memik and adjunct professor Robert Dick.

Tarzia and his collaborators presented their work June 30 at MobiSys, the International Conference on Mobile Systems, Applications, and Services.

Tarzia, who previously created a software program that used sonar to determine whether a user was at a computer (and, if they weren’t, the software would shut off the computer screen to save power), decided to extend that idea to mobile phones. What if the phone could use sonar to determine its location?

That idea turned out to be infeasible: the phone would need to emit a sound and record the sound’s echo, which was too loud and reliant on the phone’s orientation. Tarzia and his collaborators then tried recording acoustic fingerprints based on loudness alone, which didn’t work, and even tried recording just the sound that fluorescent lights emit.

The solution, it turned out, was much simpler.

“The purely passive approach works really well,” he said.

The Batphone app records 10 seconds of noises that humans often ignore: vents, computers, lights, and appliances. The program then looks at how the sound energy is distributed over various frequencies, and after filtering out transient, short-lived sounds (like someone talking), it creates a sound fingerprint for the room.

Tarzia has tagged rooms in his home and around McCormick’s Technological Institute and is currently refining the app so it works in hallways.

“It was an interesting research question because I thought it was going to fail,” he said. “I’m surprised by how accurate it is.”

Right now the app is just a proof-of-concept for the technique; in the future, it could help provide indoor navigation or help determine indoor locations for users of social applications like FourSquare, where users “check-in” at businesses to both let their friends know where they are and to earn points from the business.

An acoustic fingerprint is just one way of determining location indoors; other possibilities include Wi-Fi signals and radio signals from cellular towers.

“Ideally future technology would combine two sources to get better accuracy,” Tarzia said.

In the meantime, Tarzia has also created a website with tags from different rooms around the Technological Institute where users can listen to an acoustic fingerprint and try to determine which room recording matches it.

“It gives people an intuition behind what the app is doing,” Tarzia said.

Take the test to determine your ears’ room identification abilities, or download the app at the iTunes store.