Submarines have used sonar for decades. Bats and dolphins have used it for millions of years. And thanks to a little
math, humans could soon be echolocating with their mobile phones.
At the École Polytechnique Fédérale de Lausanne (EPFL), in Switzerland, experts in signal processing discovered a
mathematical technique that allows ordinary microphones to "see" the shape of a room by picking up ultrasonic
pulses as they bounce off the walls. The work was published in this week's edition of the journal Proceedings of the
National Academy of Sciences (PNAS).
Microphone echolocation is harder than it sounds. Ambient noise in any room interferes with the sounds used to
locate the walls, and the echoes sometimes bounce more than once. There is also the added challenge of figuring
out which echoes are bouncing off which wall. [See also: "How Bats Stay on Target: Bio Sonar "]
Bats have had millions of years to evolve specialized neural circuits to fine-tune their echolocation abilities, said
Ivan Dokmanic, a doctoral researcher and lead author of the PNAS paper. He added that humans can echolocate
too, though not as precisely. (Some blind people have demonstrated this ability.)
One reason echolocation is easier for bats and humans than it is for computers is that bats and humans have
skulls that filter the sound. Tracking where a sound originates is easier for humans because people's two ears hear
slightly different things. This allows humans to pinpoint the origin of a sound.
To enable echolocation in mobile devices, Dokmanic investigated the math behind echolocation. What he found was
that it's possible to treat the echoes of sounds emitted by a speaker as sources, rather than as waves bouncing off
of something.
It's kind of like what happens when you look into mirror: Your eyes see a reflection, but there's the illusion that
there's another person who looks just like you standing at precisely the same distance from the mirror.
That's what Dokmanic did with sound. He assumed that each echo was a source, and created a kind of grid, called
a matrix, of distances. Using some advanced math, he was then able to create an algorithm that could group the
echoes in the correct way to deduce the shape of a room.
First, the team experimented with an ordinary room at the EPFL, using a set of microphones and a laptop computer
to test whether the algorithm worked. It did, and their next step was to test their program in the real world. So they
went to a cathedral and tested it there.
"It was really the opposite environment," Dokmanic said, adding that unlike a controlled lab setting, a cathedral has
a lot of ambient noise and the space isn't perfectly square.
The algorithm worked there too, showing that the echolocation scheme could detect the cathedral's walls.
"The innovation is in the way that they process the signal to calculate the shape of the room," said Tommaso
Melodia, an associate professor of electrical engineering at the University at Buffalo who was not involved in the
study.
Martin Vetterli, professor of communications systems at EPFL and a co-author of the paper, said that mobile
phones could be used to locate people more precisely. One problem with getting anyone's precise location on the
phone is that only certain frequencies penetrate building walls, so GPS signals are sometimes useless.
Moreover, GPS is not always precise — if there's a lot of interference,it's not uncommon for a phone to say it can't
locate you more precisely than within a half mile. Wi-Fi could work, but it depends on the existence of a local
network.
Echolocation partly solves that problem, because it can measure the distance from where a user is standing to the
walls of an individual room, and send that more precise information to tell the network exactly where that person is
located. Instead of knowing where someone is within a city block, you'd be able to see that he or she is inside a
room of a certain size or is surrounded by walls that give an intersection a certain shape.
One other issue is the distance between two microphones on a mobile phone. Many mobile phones have two mics
—the directional mic is used when it's pressed to your head while you're on a phone call, and the other is used for
canceling out the ambient noise.
The two microphones on a phone calculate the distance by triangulating – measuring the small gap between when
an echo reaches each microphone. The distance between the microphones is the base of a triangle, and the time
difference between echoes' time of arrival tells you the length of the other two sides.
But these two microphones usually aren't very far apart on phones, so calculating the distance to a source that's far
away is harder to do.
One solution, Vetterli said, might be to use people's tendency to walk with their phones in order to help echolocate
walls more accurately.
Since you can't make phones much bigger, it is simpler to have the phone take measurements from more than one
spot as the user walks with it, so the base of the triangle is longer, he said.
