| Abstract |
5G networks are becoming popular and important in today's digital era. They offer faster data transfer speeds, lower latency, and greater connectivity density, making them a key driver for many industries. This technology makes it possible to create new possibilities for the Internet of Things (IoT), smart cities, and autonomous driving. However, the large amounts of data from various devices, sensors, or terminals in the 5G network can make it challenging to manage, we have developed a Clustered Federated Learning (CFL) framework. This framework clusters all the clients involved in the FL task based on their data similarity before training a local model. To encourage clients to participate in the FL task, an incentive mechanism based on the Stackelberg game approach is used. This is especially important because clients are usually self-interested and subject to training time and energy consumption. Finally, simulation experiments demonstrate the effectiveness of the proposed incentive method. © 2024 IEEE. |
