Federated Learning for Resource Management in Edge Computing

Abstract

Federated learning is an innovative machine learning technique that enables the collaborative training of a model on diverse datasets from multiple independent devices. By de-centralizing the training process, federated learning ensures data privacy and security while harnessing the collective intelligence of edge devices. Effective resource management is essential in federated learning frameworks, as it helps in achieving the peak performance while ensuring the maintenance of ideal temperature conditions. Existing research work tends to propose new solutions rather than improving the current hardware capabilities. This research explores the application of Federated Learning for Resource Management in Edge Computing using Flower federated learning simulation framework. A deep learning model is used to predict the device's temperature after a specific time interval. This prediction can be used to adjust the CPU frequency. Various experiments are conducted using different datasets, deep learning models, edge devices, the number of edge devices, and the number of training rounds in the federated learning framework. Results show that our proposed approach with the CNN regression model is very accurate and effective for resource management in federated learning frameworks.