Towards DAS ship detection using signal-based features

Within the last decade, technological advancement and research development have enabled to exploit existing underwater fiber-optic cables for monitoring their environment as well as their intrinsic properties. Those cables are critical parts of the maritime infrastructure and are thus targets for sabotage through vessels and anchor drops. Vice versa, fiber-optic cables can be used to detect ships for, e.g., monitoring sea traffic in that area or maritime surveillance. Passing vessels perturb the water column and generate waves with distinct signatures at specific frequencies that can be sensed by the cables. In this study, we tested a set of statistical and signal-based features to detect ships using the distributed acoustic sensing (DAS) method. We analyzed the acoustic signature of a ship passing a telecommunication cable off the coast of Oregon. Our results showed that the selected features, energy envelope, spectral entropy, and frequency index, can detect ships, without considering additional environmental data. Thus, this study provides a compliment to existing studies for further automatic ship detection and with that the ability to protect the cable infrastructures.