Edge-cloud collaborative aiot platform for smart fertigation in greenhouse tomato production

Overview

This solution is an edge-cloud collaborative learning platform designed to advance smart fertigation management in greenhouse tomato cultivation. By integrating sensing (AI), connectivity (IoT), and data analytics, the system enables complex decision-making for optimized crop production. The platform addresses a critical challenge in greenhouse tomato production: inefficient fertigation caused by salt accumulation and high nutrient solution concentrations, which can reduce yield and quality through elevated pH and electrical conductivity (EC).

The technology is relevant to greenhouse operators, agritech companies, and precision agriculture stakeholders seeking to improve nutrient management, reduce input waste, and increase crop performance through data-driven automation.

Technical specifications

System architecture:

• Network layer: LoRa-enabled sensors deployed throughout the greenhouse monitor key parameters including temperature, pH, EC, nitrate, humidity, CO2, and light intensity
• Edge layer: On-site edge devices perform real-time processing and local inference for immediate fertigation control decisions
• Cloud layer: Centralized database and analytics engine aggregate sensor data and host AI models for long-term learning and visualization
• Application layer: Dashboard interface for real-time monitoring of all parameters and crop stress indicators

AI and analytics capabilities:

• AI-enabled crop stress monitoring model for greenhouse tomato with demonstrated 91% accuracy
• Machine learning and deep learning-based data analytics for predicting fertigation requirements
• Federated learning approaches to improve model prediction results while preserving data privacy
• Data fusion combining IoT-sensed environmental data with crop growth information to develop decision support outputs

Key benefits:

• Real-time monitoring and control of nutrient solution parameters
• Reduced risk of salt accumulation and nutrient imbalance
• Potential improvements in tomato yield and quality through optimized fertigation timing and dosing
• Scalable architecture suitable for varying greenhouse sizes and configurations

Technology readiness level

The platform has been validated through initial deployments, including the development of an AI-enabled crop stress monitoring system achieving 91% accuracy on greenhouse tomato data. The system has been successfully deployed on edge devices with data transmitted via the IoT framework to a cloud server, demonstrating functional end-to-end AIoT integration. A real-time monitoring dashboard has also been developed and tested.

Future validation efforts will focus on extending the system into a full decision support tool that predicts fertigation requirements by fusing IoT sensor data with crop growth models. Planned work includes implementing federated learning to enhance prediction accuracy and data privacy, and validating the platform's ability to optimize fertigation decisions in operational greenhouse environments.