Locating Acoustic Events Based on Large-Scale Sensor Networks

Abstract: Research on acoustic source localization is actively being conducted to enhance accuracy and coverage. However, the performance is inherently limited due to the use of expensive sensor nodes and inefficient communication methods. This paper proposes an acoustic source localization algorithm for a large area that uses low-cost sensor nodes. The proposed mechanism efficiently handles multiple acoustic sources by removing false-positive errors that arise from the different propagation ranges of radio and sound. Extensive outdoor experiments with real hardware validated that the proposed mechanism could localize four acoustic sources within a 3 m error in a 60 m by 60 m area, where conventional systems could hardly achieve similar performance.