A Modular Intelligent Resource Architecture: Enhancing Energy Efficiency in Smart Farming with Edge-Cloud Fusion

Abstract

Smart greenhouse systems utilizing IoT technologies for real-time monitoring and autonomous control have emerged as a response to increasing climate variability. However, conventional architectures suffer from scalability limits, high latency, and excessive energy use. This study introduces a novel Edge Fusion 3-Layer architecture that integrates localized anomaly detection via Probabilistic Neural Networks (PNNs) at the edge with cloud-based decision-making through Adaptive Neuro-Fuzzy Inference Systems (ANFIS). The proposed model employs event-driven communication to transmit only critical anomalies, thereby conserving energy and improving responsiveness. Simulation results demonstrate that Edge Fusion reduces total energy consumption by up to 45% (809 J at 1000 nodes) and achieves client- and server-side delays below 30 ms, validated through two-way ANOVA with strong explanatory power (R2 = 0.775 for energy, R2 = 0.748 for server delay). Lifetime analysis further confirms extended sensor operation, particularly at smaller deployment scales. Compared with recent state-of-the-art approaches, including Hybrid PSO + Fuzzy Clustering and ANFIS + DBO optimization, Edge Fusion delivers statistically robust improvements in energy efficiency and responsiveness. These results highlight the framework’s potential as a scalable and sustainable solution for smart agriculture in resource-constrained environments.