MininetFed

An Emulation Environment for Federated Learning algorithms evaluation

MininetFed is a federated learning environment emulation tool based on Mininet and Containernet.

MininetFed Architecture

Its main features include:

• Addition of new Mininet nodes: Server, Client. These containerized nodes allow configuration of connection characteristics, available RAM, CPU, etc.
• Automatic setup of a communication environment using MQTT
• Facilitates the implementation of new aggregation and client selection functions
• Enables the development of new trainers to be executed on the clients (model + dataset + manipulations)

Currently, the following versions of MininetFed are available:

• MininetFed 2.0 - Updated version with improvements – Best Tool at the Demo sessions of SBRC 2026.
• MininetFed Low Energy Extended version considering energy consumption of nodes. Presented at Demo session of SIGCOMM 2025.
• MininetFed-LoWPan Extended version supporting 6LoWPAN and IEEE 802.15.4 emulation.
• MininetFed - An Emulation Environment for Federated Learning algorithms evaluation – Second Best Tool at the Demo sessions of SBRC 2024.