Intelligent Underwater Monitoring Systems Combining Distributed Underwater Sensor Networks with Cloud-based Digital Twins
Doctoral Researcher:
Lukas Schattenhofer, GEOMAR and Kiel University, lschattenhofer@geomar.de
Supervisors:
- Dr. Jens Karstens, GEOMAR
- Prof. Olaf Landsiedel, Kiel University
- Prof. Christian Berndt, GEOMAR
Location: Kiel
Disciplines:
Keywords: Underwater Monitoring Systems, Embedded Machine Learning, Smart Underwater Sensor Networks, Digital Twins
Motivation: The transition of the energy sector to renewable sources like offshore wind and the submarine CO2 storage are imperative for meeting greenhouse gas emission targets. However, implementing these technologies at an industrial scale presents challenges like safeguarding marine ecosystems, managing conflicts with established economic sectors, and addressing the increasing vulnerability of coastal communities to climate change impacts and environmental stressors and natural hazards. To effectively address these challenges, underwater observation capabilities need to be made available to a large user base, which requires novel, easy-to-use and cost-effective approaches. We advocate for the application of resource-efficient embedded machine learning in distributed sensor networks for submarine monitoring and data acquisition systems. While remarkable progress has been made in terrestrial Internet of Things (IoT) applications, underwater monitoring is currently mainly performed by passive, unconnected sensor platforms that are incapable of performing autonomous monitoring tasks, required to meet the ever-increasing demand for real-time underwater data. Our approach to bridge the gap and to enable smart, underwater sensor networks is based on devising resourceefficient on-device machine learning algorithms for automated data analysis and event detection. The potential of highly resource-efficient network models deployable on MCUs has already been successfully demonstrated for sensor-level seismological data as part of a previous MarDATA project.
