Energy-efficient OECT Sensor Data Analysis on Constrained Edge Devices

A new frontier in Smart Agriculture is merging nanobiotechnology with edge computing, for on-field raw data collection and processing. Smart plant sensors communicate plant chemical signals to on-field agricultural and phenotyping equipment. Particularly promising are the Organic Electrochemical Transistors (OECTs), i.e., devices that can measure the ionic content of liquid samples and biological systems. In this work, we present and evaluate several algorithms for solving a mathematical model that describes the behavior of OECT devices, in order to translate raw values like electrical currents, to meaningful information about the monitored plant stem, e.g., the concentration of ions and water saturation. Our Rust-based algorithm implementations are energy-efficient and suitable for real-time execution on constrained edge devices, as we demonstrate providing several experimental results that concern the quality of model solution, memory footprint, execution time, and the energy cost. The experiments were carried out using two different Arm Cortex-M processors, an ultra low power one and a high performance one.