Abstract
<jats:p>This paper presents an IoT-driven smart agriculture framework integrating multi-layer sensing to enable real-time decision-making and long-term optimization of agricultural resources. A heterogeneous sensors monitors soil, environmental, and crop health parameters, while edge-based sensor fusion, anomaly detection, and adaptive communication enhance data reliability and energy efficiency. Real-time control optimizes irrigation, fertilization, and pest management. Field and greenhouse experiments demonstrate significant performance gains: irrigation volume was reduced by 35% in wheat fields and 28% in greenhouse tomatoes, precision NPK dosing improved fertilizer use ratio by 22% and reduced fertilizer waste by 17%, and pesticide usage decreased by 40% in tomato greenhouses. Crop yields increased by 21.4% for wheat and 21.8% for tomatoes. Compared with standalone IoT systems, the proposed framework achieved 12% higher water use efficiency and improved yield predictability. Low-power LoRa nodes operated for 148 days on a single battery, confirming scalability and sustainability.</jats:p>