MicroPython Testbed for Federated Learning Algorithms

2024-05-15

Telecommunications Forum

AI Summary

Existing Python-based federated learning (FL) testbeds operate exclusively on single machines, failing to support distributed deployment across heterogeneous edge devices—such as PCs and resource-constrained IoT nodes. To address this limitation, we propose MPT-FLA, a lightweight MicroPython-based FL test platform—the first to fully port an FL testing framework onto the MicroPython runtime. MPT-FLA enables cross-platform, decentralized FL experimentation at the edge. It employs a streamlined communication protocol compatible with both Wi-Fi (e.g., PC + Raspberry Pi Pico W) and wired LAN environments, ensuring low-overhead, high-compatibility distributed model collaboration. Experimental evaluation confirms its deployability across heterogeneous hardware and extensibility to diverse FL algorithms. MPT-FLA provides an open-source, lightweight, and reproducible paradigm for developing and empirically evaluating FL systems in edge intelligence scenarios.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Other Foundations of Multi Agent Systems
• Application Domains: Internet of Things, Sensor Networks & Smart Cities

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

The original Python Testbed for Federated Learning Algorithms (PTB-FLA) is a light FL framework, which provides generic centralized and decentralized federated learning algorithms, but has the limitation that all the application instances can run only on a single PC. This paper presents the MicroPyton Testbed for Federated Learning Algorithms (MPT-FLA), the new framework that overcomes this limitation such that individual application instances may run on different network nodes like PCs and IoTs, primarily in edge systems. The new framework was validated and evaluated on a wireless LAN of PCs and RPi Pico W boards and wired LAN of PCs, respectively, by using application examples originally developed for the predecessor framework.

Problem

• Federated Learning
• Multi-device Compatibility
• Edge Systems

Innovation

• MicroPython Federated Learning
• Asynchronous I/O
• Edge Systems Compatibility