gunfire locator

History

Determination of the origin of gunfire by sound was conceived before World War I where it was first used operationally (see: Artillery sound ranging).

In 1990, a unique algorithm was used as a starting point : Metravib defence, working with Délégation Générale pour l’Armement (DGA) – the French defence procurement agency – studied the acoustic signature of submarines. The DGA & Section Technique de l’Armée de Terre (STAT), the French Army’s engineering section

subsequently commissioned Metravib D. to find a solution for shot detection, a way to assist soldiers and peacekeepers who come under fire from snipers without knowing precisely where the shots were coming from.

In the early 1990s, the areas of East Palo Alto and eastern Menlo Park, California, were besieged with crime. During 1992 there were 42 homicides in East Palo Alto, which resulted in East Palo Alto becoming the murder capital of the United States. The Menlo Park police department was often called upon to investigate when residents reported gunshots; however there was no way to determine their source from scattered 911 calls.

In late 1992, John C. Lahr, a PhD seismologist at the nearby United States Geological Survey, approached the Menlo Park police department to ask if they would be interested in applying seismological techniques to locate gunshots. Others had also approached the Menlo Park police department suggesting ways to help the police by means of gunshot location systems. The police chief arranged a meeting with local inventors and entrepreneurs who had expressed an interest in the problem. At that time there were no solutions to tracking gunshots, only a desire to do so. One key attendee was Robert Showen, a Stanford Research Institute employee and expert in acoustics.

Lahr decided to go ahead with his plans to demonstrate the feasibility of locating the gunshots, relying on his background in the earthquake location techniques and monitoring in Alaska. A network consisting of one wired and four radio-telemetered microphones was established, with his home in eastern Menlo Park becoming the command center. Lahr modified the software typically used for locating earthquakes and recorded the data at a higher sample rate than is used for regional seismology. After gunshots were heard, Lahr would determine their location while his wife monitored the police radio for independent confirmation of their source.

Gunfire characteristics

There are three primary attributes that characterize gunfire and hence enable the detection and location of gunfire and similar weapon discharges:

An optical flash that occurs when an explosive charge is ignited to propel a projectile from the chamber of the weapon

A typical muzzle blast generates an impulse sound wave with a sound pressure level (SPL) that ranges from 120 dB to 160 dB

A shock wave that occurs as a projectile moves through the air at supersonic speed. Note, this does not apply to subsonic ammunition, whose bullet projectiles do not exceed 1120 feet per second (i.e. the speed of sound in air).

Optical flashes can be detected using optical and/or infrared sensing techniques; however there must be a line of sight from the sensor to the weapon, otherwise the flash will not be seen. Indirect flashes that bounce off nearby structures such as walls, trees, and rocks assist in exposing concealed or limited line-of-sight detections between the weapon and the sensor. Because only optical flashes are detected, such systems are typically capable of determining only the bearing of a discharge relative to sensor unless multiple systems triangulate the shot range. Multiple gunshots, fired from multiple locations at nearly the same time, are easily discriminated as separate gunshots because the sensors generally utilize a focal plane array consisting of many sensitive pixels. Each pixel in the entire focal plane (e.g. 640×480 pixels) is constantly evaluated.

The projectile generally must travel within 50 to 100 meters of a sensor in order for the sensor to hear the shockwave. The combination of a muzzle blast and a shockwave provides additional information that can be used along with the physics of acoustics and sound propagation to determine the range of a discharge to the sensor, especially if the round or type of projectile is known. Assault rifles are more commonly used in battle scenarios where it is important for potential targets to be immediately alerted to the position of enemy fire. A system that can hear minute differences in the arrival time of the muzzle blast and also hear a projectile's shockwave "snap" can calculate the origin of the discharge. Multiple gunshots, fired from multiple locations at nearly the same time, such as those found in an ambush, can provide ambiguous signals resulting in location ambiguities.

Design

Sensing method

Gunshot location systems generally require one or more sensing modalities to detect either the fact that a weapon has been fired or to detect the projectile fired by the weapon. To date, sound, vibration and visual or infrared light have successfully been used as sensing technologies. Both applications can be implemented to detect gunfire under static and dynamic conditions. Most police-related systems can be permanently mounted, mapped and correlated as the sensors remain in place for long periods. Military and SWAT actions, on the other hand, operate in more dynamic environments requiring a fast setup time or a capability to operate while the sensors are on move.

Acoustic systems "listen" for either the bullet bow shockwave (the sound either of the projectile or bullet as it passes through the air), the sound of the muzzle blast of the weapon when it fires the projectile, or a combination of both.

Due to their ability to sense at great distances, to sense in a non line-of-sight manner, and the relatively low bandwidth required for transmitting sensor telemetry data, systems deployed for law enforcement, public safety and homeland security in the United States have primarily been based on acoustic techniques.

Acoustic-only based systems typically generate their alerts a few seconds slower than optical sensing systems because they rely on the propagation of sound waves. For example, the sound reaching a sensor 1 mile from its origin will take almost 5 seconds. A few seconds to accommodate pickup from distant sensors and to discern the number of rounds fired, often an indicator of incident severity, are both tolerable and a drastic improvement for typical police dispatching scenarios when compared against the several minutes that elapse from when an actual discharge occurs to the cumulative time of several minutes that pass when a person decides to place a 9-1-1 call and that information is captured, processed, and dispatched to patrol officers.

Because such systems have arrays of highly sensitive microphones that are continuously active, there have been concerns over privacy with this broad ability to record conversations without the knowledge of those being recorded (this is "collateral eavesdropping", because capturing conversations is only an inadvertent capability of the system's design, and law enforcement agencies have stated that the recording happens only after shots have been detected.)

Discriminating gunfire

Many techniques can be used to discriminate gunfire (also referred to as "classifying gunfire") from similar noises such as cars backfiring, fireworks, or the sound of a helicopter passing overhead. Analysis of the spectral content of the sound, its envelope, and other heuristics are also commonly used methods to classify whether loud, sudden sounds are gunfire. Identifying the source of the sounds can be subjective, and companies such as ShotSpotter revise their records based upon information they receive from police agencies, so that a sound originally classified by the automated system as the beat of helicopter rotors has been reported first as three, then four, and finally as the sound of five separate gunshots. As a result, this technology has been rejected in court cases as non-scientific for the purpose of legal evidence. It is meant to be an investigative tool rather than a source of primary legal evidence.

Another method of classifying gunfire uses "temporal pattern recognition," as referred by its developer, that employs artificial neural networks that are trained and then listen for a sound signature in acoustic events. Like other acoustic sensing systems, they are fundamentally based on the physics of acoustics, but they analyze the physical acoustic data using a neural network. Information in the network is coded in terms of variation in the sequence of all-or-none (spike) events, or temporal patterns, transmitted between artificial "neurons". Identifying the nonlinear input/output properties of neurons involved in forming memories for new patterns and developing mathematical models of those nonlinear properties enable the identification of specific types of sounds. These neural networks can then be trained as "recognizers" of a target sound, like a gunshot, even in the presence of high noise.

Regardless of the methods used to isolate gunfire from other impulsive sounds or infrared sensing, standard triangulation methods can be used to locate the source of the gunshot once it has been recognized as a gunshot.

Optical discriminating had previously consisted of methods, among them spatial, spectral, and creative temporal filters, to eliminate solar glint as a false alarm. Earlier sensors could not operate at speeds fast enough to allow for the incorporation of matched temporal filters that now eliminate solar glint as a false alarm contributor.

Architectures

Different system architectures have different capabilities and are used for specific applications. In general there are 2 architectures: stand-alone systems with local microphone arrays, and distributed sensor arrays ("wide-area acoustic surveillance"). The former are generally used for immediate detection and alerting to a nearby shooter in the vicinity of the system; such uses are typically used to help protect soldiers, military vehicles and craft, and also to protect small open-space areas (e.g., parking lot, park). The latter are used for protecting large areas such as cities, municipalities, critical infrastructure, transportation hubs, and military operating bases.

Most stand-alone systems have been designed for military use where the goal is immediately alerting human targets so they may take evasive and/or neutralization action. Such systems generally consist of a small array of microphones separated by a precise small distance. Each microphone hears the sounds of gunfire at minute differences in time, allowing the system to calculate the range and bearing of the origin of the gunfire relative to the system. Military systems generally rely on both the muzzle blast and projectile shockwave "snap" sounds to validate their classification of gunfire and to calculate the range to the origin.

Distributed sensor arrays have a distinct advantage over stand-alone systems in that they can successfully classify gunfire with and without hearing a projectile "snap" sound, even amid heavy background noise and echoes. Such systems are the accepted norm for urban public safety as they allow law enforcement agencies to hear gunfire discharges across a broad urban landscape of many square miles. In addition to urban cityscapes, the distributed-array approach is intended for area protection applications, such as critical infrastructure, transportation hubs, and campuses.

Using common data-networking methods, alerts of the discharges can be conveyed to dispatch centers, commanders, and field-based personnel, allowing them to make an immediate assessment of severity and initiate appropriate and decisive force response. Some systems have the capability of capturing and conveying audio clips of the discharges with the alert information that provides additional invaluable information regarding the situation and its severity. Similarly for the protection of critical infrastructure, where the information is clearly and unambiguously conveyed in real-time to regional crisis command and control centers, enabling security personnel to cut through often inaccurate and delayed reports so they may react immediately to thwart attacks and minimize subsequent activity.

Applications

Gunshot location systems are used by public safety agencies as well as military/defense agencies. They have been used primarily in dispatch centers for rapid reaction to gunfire incidents. In military/defense, they are variously known as counter-sniper systems, weapons detection and location systems, or other similar terms. Uses include alerting potential human targets to take evasive action, to direct force response to neutralize threats, including automated weapon cuing.

In addition to using gunshot location systems to convey incident alerts, they also can relay their alert data to video surveillance systems in real-time, enabling them to automatically slew cameras to the scene of an incident. Real-time incident location data makes the video surveillance smart; once cameras have slewed to the scene, the information can be viewed to assess the situation and further plan necessary response; the combined audio and video information can be tagged and stored for subsequent use as forensic evidence.

Infrared-based detection systems can detect not only ordnance blast signatures but also large caliber weapons such as mortars, artillery, Rocket-Propelled munitions, machine guns as well as small arms. These systems can also detect bomb impact explosions, thereby locating the impacts of indirect fire weapons like artillery and mortars. The detector can be used as an automated shot correction sensor for close arms support.

Public safety

In public safety and law enforcement, gunshot location systems are often used in high-crime areas for rapid alerts and awareness into the communications and dispatch center where the alerts are used to direct first responders to the scene of the gunfire, thus increasing arrest rates, improving officer safety, securing witnesses and evidence, and enhancing investigations, as well as in the long run deterring gun crimes, shootings and especially "celebratory gunfire" (the practice of shooting weapons in the air for fun). Gunshot location systems based upon wide-area acoustic surveillance coupled with persistent incident data storage transcends dispatch-only uses because reporting of urban gunfire (via calls to 9-1-1) can be as low as 25%, which means that law enforcement agencies and their crime analysts have incomplete data regarding true activity levels and patterns. With a wide-area acoustic-surveillance-based approach combined with a persistent repository of gunfire activity (i.e., a database), agencies have closer to 100% activity data that can be analyzed for patterns and trends to drive directed patrols and intelligence-led policing. Additional benefits include aiding investigators to find more forensic evidence to solve crimes and provide to prosecutors to strengthen court cases resulting in a higher conviction rate. With the accuracy of a gunshot location system and the ability to geo-reference to a specific street address, versus a dearth of information that typically is the case when citizens report gunfire incidents to 9-1-1, agencies can also infer shooters by comparing with known criminal locations, including those on parole and probation; investigators can also at times infer intended victims and hence predict and prevent reprisals.

Gunshot location systems have been used domestically in urban areas since the mid-1990s by a growing list of cities and municipalities that are embracing gunshot location systems as a mission-essential tool in their arsenal for fighting violent crime. Federal and homeland security agencies too have embraced gunshot location systems and their benefits; notably the FBI successfully used a ShotSpotter gunshot location system during the 2003–2004 Ohio highway sniper attacks, in conjunction with the Franklin County Sheriff.

The technology was tested in Redwood Village, a neighborhood of Redwood City, CA, in April 1996. Through 2007, the manufacturer touted the device as having benefits, but local officials were split as to its effectiveness. It is effective in reducing random gunfire. Surveys conducted for the DOJ showed it was most effective as a "perception" of action.

Military and defense

Determination of the origin of gunfire by sound was conceived before World War I where it was first used operationally. Early sound-based systems were used primarily for large weapons. Weapons detection and location systems and counter-sniper systems have been deployed by the US Department of Defense as well as by the militaries of other countries.

Acoustic threat-detection systems include the Unattended Transient Acoustic MASINT Sensor (UTAMS), Serenity Payload and FireFly, which were developed by the Army Research Laboratory.

Wildlife poaching

In South Africa's Kruger National Park, gunfire locators are being used to prevent rhino poaching.