Tutorial 2

Edge computing has become an important paradigm for next-generation distributed systems by enabling computation close to data sources, thereby reducing latency, lowering communication overhead, and improving data privacy.

At the same time, Federated Learning has emerged as a promising approach for training models across decentralized devices without moving raw data to a central server, making it particularly attractive for edge environments.

This tutorial explores the principles, system architectures, and practical realization of federated learning for edge computing. It reviews the basic concepts of FL and the Flower framework, and presents two demonstration settings: a Flower-based federated learning workflow for a Low Earth Orbit edge computing scenario, and a Raspberry Pi-based platform for lightweight federated learning.

Proposed Schedule

• Introduction: 5 minutes
• Applications of Federated Learning in Edge Computing: 10 minutes
• Architectures and Deployment Considerations for FL at the Edge: 10 minutes
• Hands-on Demo 1: Flower-based Federated Learning for a LEO Edge Computing Scenario: 30 minutes
• Hands-on Demo 2: Implementing and Visualization of Federated Learning on Raspberry Pi: 30 minutes
• Final remarks: 5 minutes

Audience and Background

Intended audience: intermediate level.

Assumed knowledge: intermediate Python programming, basic cloud and edge computing, and basic machine learning.

Instructors

• Dr Nan Yang, Western Sydney University, Australia
• A/Prof Rodrigo N. Calheiros, Western Sydney University, Australia
• Prof Massimo Villari, University of Messina, Italy
• Mark Gambito, University of Messina, Italy / Western Sydney University, Australia
• Prof Bahman Javadi, Western Sydney University, Australia

Duration

1.5 hours in one session.