Performance Evaluation and Analysis of Federated Learning in Edge Computing Environment
Young-Hwan Choi, Taehong Kim · Journal of Institute of Control Robotics and Systems · 2022
The recent development of the Internet of Things (IoT) in various fields has dramatically increased the size of the IoT market. This development has increased the amount of local data and opened up new opportunities for artificial intelligence technologies. Federated learning has emerged to protect personal information during data processing and mitigate communication costs during data transmission. In this article, we construct a federated learning environment using Django, a python web framework, and analyze the optimal conditions for edge computing through performance evaluations in various scenarios. The performance evaluations show that the accuracy and training time can be improved by increasing the size of training data, reducing the number of clients and by taking higher data ratio of a particular client when the total training dataset is given. These findings can help in significantly improving the performance of federated learning in an edge computing environment.