Spotting Snipers With Sound

BBN's gunshot-detection system used for trucks in Iraq is being modified for helicopters


In the movies, action heroes can dodge bullets or even catch them with their hands. In the real world, your average soldier is happy just to know where the shooter is.

For three years now, soldiers in Iraq and Afghanistan have had help locating snipers. An acoustics ­system known as Boomerang, developed by BBN Technologies, in Cambridge, Mass., determines the location of a shooter by sensing the blast at the rifle’s ­muzzle and the air pressure of the bullet as it whizzes past. Now BBN is ­taking the technology further, ­building and testing a version for ­helicopters, which it hopes to ship next year.

The U.S. military first began to deploy the system in Iraq in 2005. Today more than 1000 of the units are in Iraq and in Afghanistan. BBN claims that the systems detect 95 percent of all ­bullets that fly anywhere within 30 meters of their sensors, never ­mistaking some other loud noise for a bullet.

BBN is best known for ­building the original Internet network in the 1960s. But it has also done a lot of work in acoustics.

In 1978, the U.S. Congress asked the company to analyze a ­motorcycle ­policeman’s audiotape of the 1963 ­assassination of President John F. Kennedy. ”That was the point at which we started ­thinking about gunshots and gunshot ­detection via acoustic ­signatures,” says Dave Schmitt, BBN ­program manager for Boomerang. In the mid-1990s the Defense Advanced Research Projects Agency (DARPA), ran a small competition to pursue that line of inquiry, which BBN won. The United States wasn’t involved in any hot wars at the time, ”so they put that technology on the shelf,” he says. DARPA’s ­interest revived in the summer of 2003, when postinvasion resistance in Iraq led to ­frequent sniper attacks on U.S. troops. The agency asked BBN to build a ­system that would operate on the move. The company came up with a first-generation design in about two months, and 50 units were quickly assembled and sent over.

Using an array of seven microphones as its sensors, the system first detects the bullet’s supersonic shock wave, from which it determines the angle from the sensor to the shooter. The system then listens for the muzzle blast. The time difference between the shock wave and the arrival of the blast—and some clever algorithms—yields complete information about the direction, elevation, and range, all in less than a second and a half. The shooter’s direction is indicated as a clock position on a small console, the range and elevation are displayed on an LED screen, and all the information is spoken aloud by a recorded voice.

Once the microphone array and the signal processing had been worked out, the principal technical challenge was eliminating false positives—sounds that might be misidentified as gunshots. Early on, slamming a Humvee door would be enough to trigger the system, Schmitt says. ”It would look enough like a [bullet’s] shock wave that we would do the processing and report it falsely.” After some ­tinkering, BBN got Boomerang into shape. The tradeâ''off in reducing false positives was in ­diminishing the system’s ­ability to figure the origin of bullets that do not come within 30 meters of a moving ­vehicle or 50 meters of a stationary one. But a soldier doesn’t care about gunfire that misses by such a wide margin, says Schmitt. And Boomerang hasn’t registered a single false alarm in the past two years in Iraq and Afghanistan, he claims.

At the same time, BBN began ­testing a bullet-detection system that can be mounted on helicopters. Such a ­system is much more complicated, in part because a helicopter can move much faster and in more directions. Bullets are so fast that Boomerang needn’t ­factor in the ­movement of a Humvee, even when ­traveling at 100 ­kilometers per hour. But a helicopter’s greater speed can’t be ignored, so the new ­system uses ­accelerometers to factor it in. A ­single set of sensors arranged on a mast strapped to the top of a vehicle serves all types of cars and trucks, but this won’t work on a helicopter, which can be shot at from below as well as above. The ­microphones have to be carefully ­positioned, and each model will require a custom configuration. Finally, the new system uses only the shock wave—and more sophisticated algorithms—not the muzzle blast, which cannot be reliably measured from a helicopter.

Last June and then again in August, a prototype was tested at Fort Rucker, in Alabama. Previous tests had ­collected data to be processed off-line back at BBN; these were the first to ­process data on board. Both BBN and the Department of Defense declined to ­comment on their outcomes, but Schmitt says that he believes a deployable ­version will be ready sometime in 2009.