A Prototype for Lab-Based System Testing of Cyber Physical Systems for Smart Farming

Machine learning models are typically evaluated directly on data, in simulated environments or in real conditions. Because smart farming involves cyber physical systems, location information, environment conditions, hardware configurations and timing issues matter. Therefore, it is desirable to perform system testing in real conditions. However, in the agricultural domain, this is often not feasible due to economic constraints or due to the fact that one would have to wait for the crops to grow before conducting the evaluation. Therefore, we propose an architecture and a prototypical implementation for lab-based system testing of machine learning based cyber physical systems in the agricultural domain.