FLAG: Federated Learning for Sustainable Irrigation in Agriculture 5.0
- Resource Type
- Periodical
- Authors
- Bera, S.; Dey, T.; Mukherjee, A.; De, D.
- Source
- IEEE Transactions on Consumer Electronics IEEE Trans. Consumer Electron. Consumer Electronics, IEEE Transactions on. 70(1):2303-2310 Feb, 2024
- Subject
- Power, Energy and Industry Applications
Components, Circuits, Devices and Systems
Fields, Waves and Electromagnetics
Servers
Cloud computing
Irrigation
Computational modeling
Data models
Sensors
Soil moisture
Dew computing
federated learning
energy-efficient
latency
sustainability
- Language
- ISSN
- 0098-3063
1558-4127
This paper proposes a federated learning-based decision making framework for sustainable irrigation using IoT and dew-edge-cloud paradigm. The federated learning is used to prevent the sharing of user identities and raw data for data privacy protection. Further, gradient encryption is used to prevent the leakage of gradient information. Long short-term memory (LSTM) network and deep neural network (DNN) are used for data analysis in local and global models. Edge computing is used to reduce energy consumption and latency. The cache-based dew computing is used to provide temporary holding of the data when network connectivity is not available. The results present that the proposed framework achieves ~99% prediction accuracy at ~50% lower latency and energy consumption than the conventional edge-cloud framework.