<![CDATA[Conventional hydroponic fodder production relies on manual, non-predictive monitoring, complicating feed planning. This study presents a low-cost edge-computing platform for hydroponic fodder built on a repurposed Linux set-top box (STB), integrated with an environment-driven growth model through a Hardware-in-the-Loop (HIL) scheme. The STB edge node hosts the MQTT broker, time-series database, and monitoring dashboard, while temperature and light are acquired in real time by SHT30 and BH1750 sensors via an ESP32. No physical load cell is used; biomass is therefore simulated by a logistic ordinary differential equation (ODE) with temperature-dependent rate r(T) and light-dependent carrying capacity K(L), driven by real environmental data recorded over a single 10-day cultivation cycle (20-29 June 2026, with interpolated gaps flagged). The model is calibrated to two empirical anchor points (1 kg at day 0, 13 kg reference at day 10); this agreement is reported as a calibration/consistency check rather than out-of-sample predictive validation. Sensitivity analysis shows the endpoint is dominated by carrying capacity (S = 0.96), and a fourth-order Runge-Kutta solver deviates only 0.007 kg from a reference integrator. Crucially, the complete model is executed on the STB itself using only the Python standard library: a full 10-day simulation completes in 46.6 ms (43.7 microseconds per RK4 step) within 15 MB of RAM at a stable 50 C, demonstrating that the edge node alone suffices without cloud computation.]]>
Copyrights © 2026
