In recent years, both Edge Computing and Federated learning have focused on a great interest in research works. Instead of transmitting all data from the edge devices to a cloud data center to build a machine learning training model, Federated learning has emerged as a promising solution for data privacy by training at the edge, where end devices collaborate to train models without sharing data across distributed clients.
Existing works presented resource management tasks into resource estimation, resource discovery, resource allocation, resource sharing, and resource optimization. However, the edge servers are often resource-constrained, limiting computational efficiency. To tackle this issue, Federated Learning enables computation offloading by deploying multiple Deep Reinforcement Learning agents in distributed devices to indicate their decisions. Federated learning empowers distributed DRL training and proactively handled resources constrained devices, and also it reduces the transmission cost between the IoT devices and edge nodes.