MADELEINE: MAssive DEcentralised LEarning In the Network Edge
| Reference number | |
| Coordinator | Linköpings universitet - Linköpings Universitet Inst f systemteknik ISY |
| Funding from Vinnova | SEK 468 003 |
| Project duration | July 2026 - September 2026 |
| Status | Ongoing |
| Venture | 6G - Competence supply |
| Call | Individual mobility within 6G for collaborations with the USA and Singapore |
Purpose and goal
MADELEINE brings together two research strengths: LiU´s mathematical framework for decentralised machine learning, and NTU´s AI-based methods for wireless network optimisation. The goal is to combine them so that the learning algorithm and the wireless resources it runs on can be designed jointly, improving performance, privacy, and energy use in AI-native 6G networks.
Expected effects and result
The visit produces a joint LiU-NTU method connecting decentralised learning with AI-based wireless optimisation, a draft scientific paper, and open-source code. Longer term, it strengthens Swedish competence in AI-native 6G and starts a lasting research partnership with NTU.
Planned approach and implementation
Dr. Rodio spends 90 days at NTU, working directly with Prof. Tan´s group. The work is organised in three monthly tasks: improving learning performance under real wireless constraints, studying privacy from wireless channel noise, and reducing the energy cost of decentralised learning. Results are consolidated into a joint paper and a follow-up proposal.