Enhancing Environmental Observatories With Fog Computing

Ammar Kazem ^1* Guillaume Pierre ¹ Laurent Longuevergne ²

• ¹ Univ Rennes, Inria, CNRS, IRISA, Rennes, France
• ² Geosciences Rennes–UMR 6118, Univ. Rennes, CNRS, Rennes, France

The exploitation of natural resources by humans and the generation of waste have transformed the environment, raising concerns about the habitability of our planet for all life forms, as evidenced by the ongoing collapse of biodiversity. In this context, environmental observatories play a crucial role in documenting the state and evolution of socio-ecological systems by capturing inter-linkages between matter, energy and biota at relevant scales, in inter-connected compartments (surface, subsurface). The ultimate goal remains to understand and model the past and future trajectories of our habitats.Classical observation systems rely on a wide range of sensors of heterogeneous nature distributed over a domain, based on manual observations (manual gauges, water sampling) and/or transferred to a cloud. However effective continuous monitoring in any condition without data gap is challenged by the remote location of observatories, including limited access to energy, the large dynamic range in environmental signals, the necessity of maintenance and the need to limit our impact.In this work, we surveyed a set of environmental observatories belonging to three research infrastructures in France and Germany: the French network of critical zone observatories (OZCAR), the Réseau Zone Atelier (RZA) and the TERrestrial ENvironmental Observatories network (TERENO). The site managers and personnel express clearly the need to ensure continuous operations, adapt sampling strategy to effective in-situ events, in a context of decreasing technical staff onsite.The results of our survey highlight the critical need for bringing data processing near the sensors before the data are sent to a cloud platform. Adding in situ local computational power in the observatories themselves may improve reactivity and robustness of observation systems, while taking into account available energy at the same time.Therefore, in this review, we propose to introduce Fog Computing technologies in environmental monitoring systems, highlight its advantages and draft its main characteristics. We explore and review the value of Fog Computing, a technical solution bringing intelligence to operate adaptive heterogeneous sensor networks and comply with the challenge to capture intermittent to long-term temporal variability with an intermittent source of energy.