Solar-Powered IoT–ML Framework for Energy-Aware Crop Suitability Prediction

Abstract

: The increasing demand for sustainable and energy-aware agricultural practices due to climate change, limited natural resources, and increasing food requirements has accelerated the adoption of Internet of Things (IoT) and machine learning (ML) technologies in smart farming. This study proposes a solar-powered IoT and ML-based framework for energy-aware crop suitability prediction in precision agriculture. The system employs low-power IoT sensors to continuously monitor important environmental and soil parameters, including temperature, humidity, soil moisture, soil temperature, heat index, nitrogen (N), phosphorus (P), and potassium (K) levels. To reduce dependence on conventional energy sources, the framework is powered by solar energy, making it suitable for deployment in remote and resource-constrained agricultural environments. The collected sensor data are transmitted to a cloud-based platform for storage and real-time monitoring. Different ML models, including Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), Naïve Bayes (NB), and Neural Network (NN), are implemented for crop suitability classification while evaluating their predictive performance and computational overhead. A real-world dataset containing 11,644 initial sensor records was collected from agricultural regions of Sindh, Pakistan, covering four major crops: cotton, rice, sugarcane, and wheat. After preprocessing, 11,643 valid records were retained for model training and evaluation. Since the original dataset was imbalanced, particularly for the Sugarcane class, tabular augmentation was applied exclusively to the training set, while the test set remained unchanged to ensure an unbiased evaluation. Experimental results demonstrate that the proposed framework provides reliable crop suitability prediction and supports data-driven agricultural decision-making under resource-constrained field conditions. Among the evaluated models, RF achieved the highest held-out test performance, with an accuracy of 94.3331%, precision of 94.5985%, recall of 94.3331%, and F1-score of 94.3727%. By integrating solar-powered IoT infrastructure with ML-based analytics, class-imbalance handling, and computational-efficiency evaluation, the proposed framework offers a practical and scalable solution for energy-aware suitability assessment in resource-constrained farming environments.