Gunshot Localization Based on Acoustics

Prof. Aniruddha Sinha

Acoustic localization of gunshots will have applications in improving camp security and enhancing response to ambush situations faced by our personnel.

The acoustic signature of each gunshot contains information regarding the origin of the gunfire that may be decoded if enough microphones are deployed. With a minimum of 4 microphones simultaneously recording the gunfire sound (see figures below), appropriate real-time algorithms can yield the relative origin of the source to a sufficient degree of certainty for use in the field.

he proposed solution will be implemented on standalone hardware for robust deployment. At present, we are in the initial prototype stage. Also, in its current form, it can provide security to static installations only; subsequently, a mobile version may be developed. We have tested the device with gunfire (and firecracker bursts) that are approximately 20 to 30 meters away from the microphones. In these tests, the bearing angle was estimated accurately up to 2 degrees. However, the range error is much poorer, with maximum inaccuracy up to 40% of the actual range. The location estimate is returned by the device within 0.1 seconds of the arrival of the gunfire event at the microphones. This time interval is also the currently-implemented minimum time-separation between two gunfire events that may be individually localized. The device gives the location of the gunfire origin relative to itself. Knowledge of the devices’ location (which is straightforward as it is fixed) can easily transform this result to a grid reference.

For instance, the figure below shows the recording of a bursting cracker on four channels.