7th KuVS Fachgespräch on Machine Learning in Networking (MaLeNe 2026)

The Seventh KuVS Fachgespräch on Machine Learning in Networking (MaLeNe) was held at the University of Augsburg on March 19–20, 2026, as a successful event spanning two half-days. The workshop was organized by the co-chairs Michael Seufert (University of Augsburg), Andreas Blenk (Siemens AG), and Björn Richerzhagen (Siemens AG). In total, 13 full papers were presented and about 30 participants from industry and academia were brought together to discuss the lates advances in applying machine learning in networking.

Day 1 – From 6G Visions to Intelligent Routing

The workshop opened with a welcome by the organizers, setting the stage for two days of lively exchange. The first highlight was an industry keynote by Dr. Andreas Mäder (Nokia) titled “AI for 6G and AI for Standards – a transformation journey.” In his talk, he highlighted the role of AI as a key enabler for future 6G networks and discussed the convergence of development cycles in mobile communications and machine learning, emphasizing the need to better align innovation with standardization.

The first technical session focused on security and intrusion detection. The presentations showcased how machine learning can strengthen network defenses -from Xenia Wagner’s (ipoque, Germany) SPARTAN dataset for IoT-based APT detection in smart homes, to Marleen Sichermann’s (University of Wuerzburg, Germany) intrusion detection approaches for IPFIX networks. Johannes Schleicher (University of Augsburg, Germany) then addressed the challenge of detecting previously unseen attacks through active learning, before Hannes Schwanzer (RheinMain University of Applied Sciences, Germany) closed the session with his work on Bayesian packet header synthesis and authenticity evaluation. 

In the second session, the focus shifted to learning-based routing and congestion control. Ahmed Al-Dulaimy (University of Koblenz, Germany) presented a machine learning-based approach for predicting optimal parent nodes in software-defined wireless sensor networks to improve routing efficiency. Musab Ahmed (Hamburg University of Technology, Germany) introduced a learning-based method for local routing decisions in sparse aeronautical communication networks to enhance connectivity under challenging conditions. Andreas Boltres (Karlsruhe Institute of Technology, Germany) explored how neural networks can generate latent embeddings for telemetry-aware greedy routing, enabling more informed forwarding decisions. The session concluded with Michael König (Karlsruhe Institute of Technology, Germany), who presented a method for generalizing coordinated congestion control across multiple flows and network topologies.

The discussions continued after a city tour in Augsburg during a dinner event at a traditional Bavarian Wirtshaus.

Day 2 – From Industrial AI to Digital Twins

The second day began with another keynote, this time by Dr. Andreas Blenk (Siemens AG), titled “Industrial AI Today. Autonomous Industry Tomorrow.” His talk highlighted current developments in industrial AI, including automation, digitalization, and future research directions toward autonomous systems.

The third technical session focused on AI for network management. Zineddine Bettouche (Deggendorf Institute of Technology, Germany) introduced a spatial PDE-aware model for mobile traffic forecasting, while Ibrahima Ndiaye (Otto von Guericke University Magdeburg, Germany) presented a federated learning-based approach for proactive load balancing in IIoT gateways. Jonas Wessner (Ulm University, Germany) complemented these contributions by demonstrating how large language models can be used to automate network access control.

In the final session, attention turned to systems, platforms, and digital twins. Johannes Späth (Technical University of Munich, Germany) presented a high-fidelity virtualized digital twin for ISP core networks, enabling realistic experimentation in controlled environments. Sebastian Hauschild (University of Luebeck, Germany) concluded the session with a portable IoT platform designed for experimental cadaverine measurements, showcasing the diversity of application domains covered during the workshop.