Passive acoustic sensing for detection of small vessels

This paper investigates the system requirements and signal processing tools for a semi-autonomous passive acoustic boat detection framework for monitoring marine environments. The objective is to develop a surveillance system which provides wide area coverage, automatic (boat) detection, and ultimately the ability to communicate real-time alerts (i.e., a bell-ringer system). A critical challenge for a successful system is the ability to accurately detect (and possibly classify) targets in the presence of noise. A wavelet detection algorithm has been developed to identify the presence of a boat in acoustic signals overwhelmed by noise from snapping shrimp and humpback whales. A feature extraction tool is introduced to examine the harmonically related structures in shipping noise. These methods are applied on acoustic data collected in a series of field tests in Hawaii, in which small-scale sensor networks were deployed near Marine Protected Areas (MPA) to assess the ability to monitor boat activity. A simulation tool developed based on Bellhop is introduced to study boat signatures under different environmental conditions.