Prediction of shooter localization accuracy in an urban environment

Abstract

The use of acoustic sensor networks for shooter localization can provide a vital contribution to situational awareness in urban environments. During mission planning, the performance prediction for acoustic shooter localization plays a central role as it allows an optimization of the sensor positions in a sensor network. Instead of a sound propagation-based approach, in our work we focus on an information theoretical analysis using the Cramér-Rao bound to predict the achievable shooter localization accuracy. Through this approach, we have shown that accounting for incomplete and heterogeneous acoustic measurement data sets leads to maximization of the fusion gain and consequently to improved achievable localization accuracy. We validated the match between predicted and actual experimental performance in free-field measurements with supersonic gunshots including varying sensor-to-shooter geometries, weapon types, and various measurement types. By measuring signatures of impulsive gas cannon shots in urban terrain, we analyzed the effect of buildings to the sensor performance and achieved a significant improvement in the localization performance prediction by adjusting the sensor model depending on whether line-of-sight or non-line-of-sight conditions to the target exists.