In the paper "FedBed: Benchmarking Federated Learning over Virtualized Edge Testbeds", authored by Moysis Symeonides (University of Cyprus), Fotis Nikolaidis (Forth-ICS), Demetris Trihinas (University of Nicosia Cyprus), George Pallis (University of Cyprus), Marios D. Dikaiakos (University of Cyprus), and Angelos Bilas (Forth-ICS), the researchers introduce FedBed, an innovative testing framework that facilitates benchmarking of FL deployments in virtualised environments.
- Why is FedBed important?
Deploying FL across an Edge-Cloud continuum involves overcoming numerous technical and logistical challenges. FedBed addresses these issues by providing a solution that:
- Enables the evaluation of various trade-offs that arise when combining different software and infrastructure configurations in FL.
- Reduces process complexity by automating the setup of physical or emulated testbeds.
- Simplifies experiment configuration and monitoring, accelerating FL research and optimising QoS in Edge-Cloud environments.
Find out more about this award-winning paper here