A Novel 5G PMN-Driven Approach for AI-Powered Irrigation and Crop Health Monitoring

Precision agriculture is a crucial field that addresses food security for the growing population. This article presents an innovative approach merging deep learning with a 5G Private Mobile Network (PMN) to construct a smart agriculture system. We simulated a 5G PMN infrastructure using UERANSIM and Free5GC with the Quantum Key Distribution Function (QKDF) integration. The 5G connectivity enhances agricultural practices through real-time crop healthcare monitoring and reliable transmission of irrigation control commands. In addition to using 5G, our proposed system integrates a lightweight model and three YOLOv8 models for AI-powered irrigation and crop health monitoring. While the lightweight model automatically generates irrigation schedules based on current weather conditions and soil moisture levels, optimizing crop water needs, the three YOLOv8 models are trained to predict growth stage, health status, and nutritional status (nitrogen, phosphorus, and potassium). All deep learning models are optimized for lettuce and feature lightweight architecture for easy deployment on low-power devices. Evaluation on test datasets reveals promising accuracy scores: the irrigation schedule generation model achieves 73%, the growth stage model attains 93.8%, the health status model reaches 87%, and the nutritional status model obtains 83%. This research contributes to improving precision agriculture while providing practical solutions to enhance irrigation management and crop health monitoring. We have also developed a QKDF that is integrated with the 5G core network to improve the security of the proposed system by establishing secure session keys with a length of 256 bits.