Embedded KETOS Deep Learning Classifier for the AIOS Buoy

The Acoustic Intelligent Ocean Surveillance (AIOS) buoy is an autonomous Passive Acoustic Monitoring (PAM) system that encompasses acoustic data collection, embedded deep learning classifier models, and a buoyancy compensation capability (allowing it to move up and down in the water column under its own control), The current prototype uses as many commercial off the shelf (COTS) components so that each component can be integrated, tested, and independently improved before completing a complete integrated system design. The current AIOS buoy architecture separates functional components into physically distinct modules. This approach allows separately investigating energy requirements for each module. This discussion focuses on the energy required to operate the Ketos deep learning classifier within the current AIOS buoy architecture.