Military vehicle acoustic pattern identification by distributed Ground Sensors

Object identification is an important task of military Intelligence Surveillance and Reconnaissance (ISR) systems. It plays an integral role during threat detection and situation assessment in the monitored area. This paper considers a ground ISR system comprising distributed Unmanned Ground Sensors (UGS), which perform military vehicle identification based on acoustic signal analysis. Each UGS operates autonomously and transmits its identification estimates through a Wireless Sensor Network (WSN) to data aggregation and fusion nodes, where situation assessment is performed. The vehicles are identified by applying a multistage decision-making approach comprising time and frequency domain feature analysis and classification procedures. System operation is validated in a field experiment, during which several moving military vehicles are identified by UGS prototypes implemented on the BeagleBone Black platform. Identification accuracy reaches 89% for the chosen vehicles.