Multimodal beehive weight forecasting using IoT telemetry and acoustic signal processing

Accurate forecasting of beehive weight is an important challenge in precision apiculture, as it enables continuous assessment of colony development, nectar-flow dynamics, and potential abnormal events. This paper presents a multimodal Internet of Things (IoT) telemetry framework for short-term beehive weight forecasting using data collected by the Intelligent Hives monitoring system. The proposed framework integrates load-cell measurements, environmental sensing, acoustic telemetry, edge data acquisition, and cloud-based analytics. The dataset contains more than 3.4 million telemetry records acquired from over 250 instrumented beehives operating in real-world apiaries between 2022 and 2025. Several forecasting approaches, including multiple linear regression, gradient boosting, and feedforward neural networks, were evaluated using MSE, RMSE, and R² metrics. The experimental results indicate that historical weight measurements constitute the dominant source of predictive information, while environmental and acoustic telemetry provide complementary contextual information for multimodal hive monitoring. Furthermore, lightweight regression-based models were found to offer a favorable trade-off between forecasting accuracy and computational complexity, supporting their deployment on resource-constrained IoT platforms. The presented results demonstrate the practical applicability of multimodal telemetry, signal processing, and machine learning techniques for intelligent monitoring and predictive analytics in precision agriculture.