Shooter localization and weapon classification with soldier-wearable networked sensors

TLDR

The paper presents a wireless sensor network–based mobile counter‑sniper system and reports on its design and evaluation. The system employs helmet‑mounted microphone arrays, MICAz motes, and FPGA‑based sensor boards to capture acoustic signals, estimate time‑of‑arrival and angle‑of‑arrival, and fuse these with compass and PDA data to compute trajectory, range, caliber, and weapon type. The evaluation achieved 1‑degree trajectory precision, over 95 % caliber estimation accuracy for all shots, and nearly 100 % weapon estimation accuracy for 4 of 6 tested guns.

Abstract

The paper presents a wireless sensor network-based mobilecountersniper system. A sensor node consists of a helmetmountedmicrophone array, a COTS MICAz mote for internodecommunication and a custom sensorboard that implementsthe acoustic detection and Time of Arrival (ToA) estimationalgorithms on an FPGA. A 3-axis compass providesself orientation and Bluetooth is used for communicationwith the soldier's PDA running the data fusion and the userinterface. The heterogeneous sensor fusion algorithm canwork with data from a single sensor or it can fuse ToA orAngle of Arrival (AoA) observations of muzzle blasts andballistic shockwaves from multiple sensors. The system estimatesthe trajectory, the range, the caliber and the weapontype. The paper presents the system design and the resultsfrom an independent evaluation at the US Army AberdeenTest Center. The system performance is characterized by 1-degree trajectory precision and over 95% caliber estimationaccuracy for all shots, and close to 100% weapon estimationaccuracy for 4 out of 6 guns tested.