Internet-of-Things-Based Multiple-Sensor Monitoring System for Soil Information Diagnosis Using a Smartphone

Area of Science:

• Agricultural Technology
• Internet of Things (IoT)
• Data Science

Background:

• The agricultural industry faces increasing demands for food production.
• Smart agriculture, leveraging IoT, offers solutions for increased productivity and reduced waste.
• Real-time soil monitoring is crucial for optimizing crop yields and resource management.

Purpose of the Study:

• To develop and evaluate a novel IoT-based system for real-time agricultural soil measurements.
• To implement an advanced prediction model for soil information using deep reinforcement learning.
• To demonstrate the system's effectiveness in improving soil data acquisition and monitoring.

Main Methods:

• Designed an IoT system with wireless sensors (temperature, moisture), a microprocessor, microcomputer, cloud platform, and mobile app.
• Developed a low-power hardware and software design with a modular power supply and time-saving algorithm.
• Explored a novel soil information prediction strategy using a Deep Q Network (DQN) combined with Bi-LSTM, OS-ELM, and PEML (DQN Bi-OS-P model).

Main Results:

• The system achieved stable, reliable, real-time collection of time-series soil data.
• The DQN Bi-OS-P model demonstrated improved prediction accuracy, reducing RMSE, MAE, and MAPE, and increasing R2 by 0.1% compared to other methods.
• Experimental results showed a time error of no more than 3 seconds between the mobile app display and actual data acquisition.

Conclusions:

• The developed smart soil system effectively enables real-time monitoring of soil quality via mobile applications.
• IoT technology, integrated with advanced prediction models, significantly enhances agricultural data acquisition and management.
• The system offers a reliable and efficient solution for precision agriculture, supporting sustainable food production.