AI methods for reliable big data transport
A large data flow has to cross a busy network. Available capacity changes, feedback takes time and several destinations may need the same data. AI-RMDT investigates how artificial intelligence can help transport systems adapt to these conditions.
The project builds on RMDT and connects three questions: how to control the sending rate, how to distribute data through application-level trees and how to synchronise files efficiently. Machine learning and multi-objective optimisation are research tools for these tasks; their value has to be assessed in experiments with documented network conditions.
© FILA
The receiver reports on the transfer. The congestion controller inside the sender adjusts packet transmission.
A controller observes the network, chooses a sending rate and receives feedback on the outcome. The challenge is to make these decisions useful beyond the conditions seen during training. A published 2026 study examines the difficulties of moving a simulator-trained controller to a real high-bandwidth path. Read the scaling analysis.
Research topics
- Analysis and further development of RMDT for large-volume data transfer.
- AI-based optimisation of data flow and stability, alongside new congestion control approaches.
- Theoretical and experimental investigation of application-layer multicast and file synchronisation.
- Analysis of secure transmission models and the role of multi-parameter optimisation.
Funding
This project is funded by the European Regional Development Fund (ERDF) and the state of Saxony-Anhalt as part of the Saxony-Anhalt SCIENCE - Research and Innovation (ERDF) 2021-2027 program (funding reference: ZS/2023/12/182323).
Approved project period: 1 March 2024 to 31 December 2027.
© EFRE
