This project focuses on optimizing network communication for federated learning using Software-Defined Networking (SDN) technologies. By integrating the Flower framework for federated learning with SDN tools like ONOS, Mininet, and PyTorch, the project aims to enhance the efficiency and scalability of distributed machine learning models. The SDN controller (ONOS) manages dynamic routing to minimize communication latency and improve the overall performance of the federated learning process across multiple edge devices. The project is being tested in simulated environments using Mininet and optimized using real-time adjustments based on network conditions, with the goal of accelerating model training while reducing bandwidth consumption.
- Mininet: Used to emulate network topologies and simulate SDN-based communication between federated learning nodes.
- Flower Framework: Provides the federated learning framework for distributed model training across multiple devices.
- ONOS: Acts as the SDN controller to manage dynamic routing and network optimization for federated learning communication.
- PyTorch: Used for implementing and training machine learning models in the federated learning setup.