Main Article Content
Abstract
Large-scale adoption of Agricultural Internet of Things (IoT) technologies is increasing, driven by the growing need for data-driven and automated agricultural systems. This widespread adoption is hindered by several challenges, including high energy consumption of resource-constrained devices, limited device lifetime, communication latency, and the cost of maintaining large and heterogeneous IoT deployments. Energy consumption is particularly critical because it directly affects the operational cost and sustainability of agricultural IoT networks. Several strategies and optimization frameworks have been adopted to mitigate these challenges, including energy-aware routing, resource allocation, clustering, duty-cycling, and Federated Learning (FL)-based approaches. However, existing approaches often address these challenges independently, creating a gap in simultaneously optimizing energy consumption, device lifetime, communication latency, and data throughput in large-scale and heterogeneous agricultural IoT networks. This study introduces a Hybrid Bio inspired federated learning (HBFL) -Based Energy Optimization Framework for large-scale agricultural IoT networks. The framework integrates FL with bio-inspired optimization techniques to enable energy-aware resource allocation while reducing communication overhead. MATLAB-based simulations were conducted using a four-farm agricultural IoT network model and evaluated using four metrics: energy consumption, device lifetime, latency, and data throughput. Comparative evaluation demonstrated a 14% improvement in energy efficiency and a 20.3% improvement in communication performance. The findings demonstrate the potential of HBFL for scalable and efficient agricultural IoT networks.
