EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Sustainable agriculture through IoT-driven smart irrigation with explainable AI

S. G. Rahul, S. J. Rithikkaa, T. M. Amirthalakshmi, M. Gayathri

Frontiers in Sustainable Food Systems · 30.06.2026 · 10

DOI: 10.3389/fsufs.2026.1847041 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Frontiers in Sustainable Food Systems, online erschienen am 30.06.2026. Verfasst von S. G. Rahul und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Water is becoming scarcer due to inefficient irrigation practices on traditional irrigation schedules; thus, it will become very difficult to keep crops healthy and vibrant if farmers do not strike a balance between this and the responsible use of water. In the face of all these challenges, this paper presents a whole new type of irrigation system with the intelligence of modern internet sensing technology (IoT), machine learning, and artificial intelligence (AI) explained as one. All three have been made into one single working system. The proposed system is validated using real-time data collected from a field-deployed prototype. This system is located in a smallholder paddy field situated in the Ponneri taluk of the Thiruvallur district in the Tamil Nadu State of India. This system monitors soil moisture, air temperature, relative humidity, solar light intensity, and precipitation every 15 min over a paddy and vegetable test plot. This data is then sent to the cloud storage system ThingSpeak via NodeMCU ESP8266. This system was monitored over a period of 90 days (January to March 2024) and managed to collect 25,920 data points over 6 channels. This then gave us a final ML-ready data set of 5,000 data points aggregated every hour. An Isolation Forest anomaly detection module was then implemented to filter out 3.0% of the data points that were identified as anomalies. Five machine learning classifiers were then implemented and benchmarked with stratified 10-fold cross-validation. XGBoost was found to have the highest AUC-ROC value of 0.995 and a cross-validated F1-score value of 0.992 ± 0.003. The XGBoost model was trained in 2.40 s on a standard CPU, with a per-sample inference latency of 0.004 ms, and an end-to-end sensor-to-actuator pipeline latency of approximately 4–6 s negligible relative to the 15-min sampling cycle. This system was found to have reduced water consumption by 35.1% over traditional manual scheduling, thereby contributing to United Nations Sustainable Development Goals including SDG 2 (Zero Hunger), SDG 6 (Clean Water and Sanitation), and SDG 13 (Climate Action).

Zitieren

APA

Rahul, S. G., Rithikkaa, S. J., Amirthalakshmi, T. M., & Gayathri, M. (2026). Sustainable agriculture through IoT-driven smart irrigation with explainable AI. Frontiers in Sustainable Food Systems, 10. https://doi.org/10.3389/fsufs.2026.1847041

Vancouver

Rahul SG, Rithikkaa SJ, Amirthalakshmi TM, Gayathri M. Sustainable agriculture through IoT-driven smart irrigation with explainable AI. Frontiers in Sustainable Food Systems. 2026;10. doi:10.3389/fsufs.2026.1847041

BibTeX
@article{Rahul2026,
  author = {Rahul, S. G. and Rithikkaa, S. J. and Amirthalakshmi, T. M. and Gayathri, M.},
  title = {Sustainable agriculture through IoT-driven smart irrigation with explainable AI},
  journal = {Frontiers in Sustainable Food Systems},
  year = {2026},
  volume = {10},
  doi = {10.3389/fsufs.2026.1847041}
}
Kostenlos starten →