Doubt in concept - Federated Learning

[MalavikaJames]

Hi. I am a post graduate student working on my project on digital twins, and I wish to incorporate the idea of federated learning into it. But I have several doubts regarding the same. Here I am posting one my many doubts. I kindly request for suggestions.

I'm trying to implement a digital twin for an IoT environment. I have some sensors, edge servers, and a central server. The sensor data are stored on the edge servers, and I want to train my deep learning model using these data. I would like to apply federated learning for this training purpose so that I need not send the data to the central server for training the model. Once the training is done I can deploy this model to make predictions on the newly incoming sensor data in real-time. And I can do this on the edge servers itself.

Is my above idea feasible?
Any help is much appreciated. Kindly give me suggestions.
Thank you🙏