Federated Learning Architectures: A Performance Evaluation With Crop Yield Prediction Application

ABSTRACT Introduction Federated learning has become an emerging technology in data analysis for IoT applications. Methods This paper implements centralized and decentralized federated learning frameworks for crop yield prediction based on Long Short‐Term Memory Network and Gated Recurrent Unit. For centralized federated learning, multiple clients and one server are considered, where the clients exchange their model updates with the server that works as the aggregator to build the global model. For the decentralized framework, a collaborative network is formed among the devices either using ring topology or using mesh topology. In this network, each device receives model updates from the neighboring devices and performs aggregation to build the upgraded model. Results The performance of the centralized and decentralized federated learning frameworks is evaluated in terms of prediction accuracy, precision, recall, F1‐Score, and training time. The experimental results show that> $$ > $$ 93% prediction accuracy is achieved using the centralized and decentralized federated learning‐based frameworks. The results also show that using centralized federated learning, the response time can be reduced by∼ $$ \sim $$ 75% than the cloud‐only framework. Conclusion Centralized and decentralized federated learning architectures show good performance in terms of prediction accuracy and loss. The training time, including communication for both case studies, is also not very high, as observed from the results. Further, as no raw data is shared, the data privacy is protected. Finally, the future research directions of the use of federated learning in crop yield prediction are proposed.