Have you ever considered how many ships are present in our sea? Apparently, underwater acoustics can answer this question. Using detection and classification techniques, a vessel can be identified by its underwater radiated noise (URN). The result can be used as a remote sensing enabling technique for monitoring URN level for noise pollution monitoring; for identifying malfunctions in a vessel for predictive maintenance; or for vessel tracking. But the first step is to collect a large enough dataset that will allow performance evaluation for any detection and classification algorithm.

In this project, we have collected a dataset of acoustic data containing shipping URN and have linked it to our own dataset of AIS. The latter is a device, mandatory for vessels over 12 feet, that continuously reports the ID, speed, and location of the vessel. By connecting acoustic and AIS information, we are able to match between a passing vessel and its URN, thereby providing ground truth for the recorded vessel. 

To obtain the data, we have built a self-made underwater acoustic passive receiver that can record continuously raw acoustic data from several hydrophones (receiving elements) for a few months. The recorder was deployed by our team of scuba divers in the Rosh Carmel area across from Haifa, Israel, at a depth of 30m some 700m from the shipping line exiting the Haifa port. 

The collected dataset contains 5 months of raw acoustic data (flac files of 5min, between 3-5/2022 and 10/2022-1/2023). Analyzing this dataset and linking it to AIS, we have identified the URN of 323 different ships which passed in the range of 200 to 9000 meters from our recorder.

Left panel: AIS dataset collected in the Haifa bay, Upper middle panel: acoustic signature of a merchant vessel collected in the bay, Lower middle panel: picture from the deployment site, right panel: cavitation noise from the thruster of a vessel.

This research is supported by a dataset collection grant from the Data Science Research Center (DSRC), University of Haifa, Israel. For more info on our research please visit our lab website: Underwater Acoustic and Navigation Lab, and AIS dataset page: http://anlais.haifa.ac.il/@signalk/freeboard-sk/