Singh et al. (2026) Predicting Optimal Irrigation Strategies Using Advanced Neural Networks and IoT-Enabled Data for Precision Agriculture

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Jagendra Singh, Prateek Gupta, Monika Dandotiya, Pongkit Ekvitayavetchanukul, Manoj Rana, Bakshish Singh
• DOI: 10.1007/978-981-95-2875-2_45

Research Groups

• School of Computer Science Engineering and Technology, Bennett University, Greater Noida, India
• Department of Computer Science and Engineering, Poornima University, Jaipur, India
• Board of Khon Kaen University Affairs, Khon Kaen University, Khon Kaen, Thailand
• Computer Science and CDOE, Teerthanker Mahaveer University, Moradabad, India
• Department of Information Technology, Raj Kumar Goel Institute of Technology, Ghaziabad, India

Short Summary

This paper presents an IoT-based precision irrigation system that uses environmental sensor data and a PID controller for real-time water management, while also training neural networks (LSTM, GRU, TFT, MLP) on cloud-stored data to forecast optimal water needs. It concludes that TFT offers the highest accuracy for cloud-based systems, whereas GRU and LSTM provide a balanced performance suitable for real-time applications.

Objective

• To develop and evaluate an IoT-based precision irrigation system that integrates real-time environmental sensor data with advanced neural networks to forecast optimal water requirements, thereby enhancing water use efficiency and promoting sustainable agriculture.

Study Configuration

• Spatial Scale: Agricultural field-level
• Temporal Scale: Real-time monitoring and short-term forecasting of irrigation requirements.

Methodology and Data

• Models used: PID controller, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Temporal Fusion Transformer (TFT), Multi-Layer Perceptron (MLP).
• Data sources: IoT-enabled environmental sensors collecting data on soil moisture, temperature, pH, air humidity, wind speed, and evapotranspiration rates. Sensor readings are saved in a cloud platform.

Main Results

• The Temporal Fusion Transformer (TFT) model achieved the highest accuracy in predicting optimal water needs.
• TFT models require substantial computational power, making them most suitable for cloud-based processing environments.
• Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) models demonstrated a favorable balance between computational efficiency and prediction accuracy.
• GRU and LSTM models are recommended for real-time applications within smart irrigation systems due to their efficient performance.
• The integrated system effectively utilizes deep learning and IoT technologies to improve irrigation strategies, leading to efficient water use and optimal crop growth conditions.

Contributions

• Development of a comprehensive IoT-based precision irrigation system that combines real-time sensor data with advanced deep learning models for predictive irrigation scheduling.
• Comparative analysis of multiple neural network architectures (LSTM, GRU, TFT, MLP) to identify their performance characteristics and suitability for different deployment scenarios (cloud vs. real-time edge) in smart agriculture.
• Demonstration of a practical and effective method for optimizing water resource allocation and minimizing water loss in agriculture through predictive analytics.

Funding

• Not specified in the provided paper text.

Citation

@article{Singh2026Predicting,
  author = {Singh, Jagendra and Gupta, Prateek and Dandotiya, Monika and Ekvitayavetchanukul, Pongkit and Rana, Manoj and Singh, Bakshish},
  title = {Predicting Optimal Irrigation Strategies Using Advanced Neural Networks and IoT-Enabled Data for Precision Agriculture},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-981-95-2875-2_45},
  url = {https://doi.org/10.1007/978-981-95-2875-2_45}
}