Adaptive Resource Management in Cloud Edge Architectures for IoT-Driven Smart Agriculture Systems

This study proposes an adaptive resource management framework for cloud-edge architectures tailored to IoT-driven smart agriculture. The framework incorporates AI-driven predictive models, federated learning, and dynamic resource allocation to optimize tasks such as crop health monitoring, irrigation scheduling, and pest detection. By utilizing edge computing, the system reduces latency and energy consumption while federated learning enhances privacy and minimizes bandwidth usage. Comprehensive evaluations demonstrate the framework's scalability, achieving high predictive accuracy (91.5% for crop health, 88.7% for irrigation, and 86.9% for pest detection) and reducing bandwidth usage by up to 70%. Further experiments highlight its robustness under varying network conditions and scalability with an increasing number of edge devices. The findings underscore the potential of combining cloud and edge computing with federated learning to advance smart agriculture in resource-constrained environments.