A Deep Learning-Driven Smart Agriculture Framework with Privacy-Preserving Federated Learning and Real-Time Monitoring
Procedia Computer Science · 2026 · 283 · 4387-4401
DOI: 10.1016/j.procs.2026.06.497 ↗Kurz gefasst
Fachartikel in der Zeitschrift Procedia Computer Science, 2026. Verfasst von Malathi D und 1 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
The data-driven decision-making can be applied to predict crop growth and manage the resources with the fast development of smart farming technologies. However, existing methods still face noteworthy limitations, including insufficient model generalization on heterogeneous agricultural datasets, inadequate optimization of lightweight deep learning architectures, and limited focus on communication efficiency and privacy risks in decentralized environments. To address these drawbacks, this study introduces Improved SqueezeNet (ISQN) model combined with Adaptive Human Evolutionary Optimization (AHEO) to predict crops efficiently and accurately. ISQN architecture is rich in feature representation and low in computational cost, and AHEO is used to optimize the hyperparameters dynamically, which leads to the convergence and performance stability. The framework is further integrated with Federated Learning (FL) and Digital Twin (DT) technologies, enabling decentralized training, real-time monitoring, and predictive analytics while reducing communication costs and mitigating privacy leakage. Experiments on the Smart Farming 2024 (SF24) dataset show that the proposed model allows reaching 98.85% accuracy, 98.68% F-score, which is better than a number of state-of-the-art methods. These findings affirm that the proposed framework is a lightweight, precise, and privacy-preserving framework that can be applied to smart agriculture applications in the real world.
Zitieren
APA
D, M., & P, S. (2026). A Deep Learning-Driven Smart Agriculture Framework with Privacy-Preserving Federated Learning and Real-Time Monitoring. Procedia Computer Science, 283, 4387-4401. https://doi.org/10.1016/j.procs.2026.06.497
Vancouver
D M, P S. A Deep Learning-Driven Smart Agriculture Framework with Privacy-Preserving Federated Learning and Real-Time Monitoring. Procedia Computer Science. 2026;283:4387-4401. doi:10.1016/j.procs.2026.06.497
BibTeX
@article{D2026,
author = {D, Malathi and P, Selvameena},
title = {A Deep Learning-Driven Smart Agriculture Framework with Privacy-Preserving Federated Learning and Real-Time Monitoring},
journal = {Procedia Computer Science},
year = {2026},
volume = {283},
pages = {4387--4401},
doi = {10.1016/j.procs.2026.06.497}
}
EUVIMED