Description
Master’s thesis in cooperation with Fraunhofer AISEC
GNN-based Intrusion Detection for Time-critical Data Streams
Real-time distributed systems are currently arising in fields like industrial automation or
autonomous driving. While the implementation for deterministic or soft-real time behavior
is well studied, security-mechanisms in those systems are not well understood or de-
veloped. The tasks of this thesis will be to lay the foundation for intrusion detection for
real-time distributed systems with a controller based networking architecture. Your task
will be to use Graph Neural Networks (GNNs) to do intrusion detection in Time-sensitive
networks utilizing new temporal features emerging from scheduling mechanisms.
Task Description
Your tasks will be manifold, starting from the generation of datasets to actually training a
GNN:
• Literature review on existing approaches of GNN with regards to intrusion detection
• Setting up a simulation environment with OMNET++ allowing configuration of a TSN
network with different real-time scheduling mechanisms (e.g., Time-aware shaping,
Credit-based shaping, asynchronous traffic shaping)
• Generating a dataset with additional temporal features by either simulation of traffic
baselines and attacks (e.g., flooding attacks against time-critical data streams) or
utilizing existing datasets (with possible adaptations for TSN-conformity)
• Utilize GNN to train on the newly generated dataset, evaluating its performance to do
intrusion detection
Requirements
• High motivation and ability to work independently
• Basic understanding of graph theory, networking, and cybersecurity
• Optimally, interest and experience in machine learning
Contact
Please send your application with current CV and transcript of records to:
Lukas Lautenschlager
Fraunhofer Institute for Applied and Integrated Security (AISEC)
Product Protection and Industrial Security
Lichtenbergstr. 11, 85748 Garching near Munich
Mail: lukas.lautenschlager@aisec.fraunhofer.de
Publication Date: 16.03.2026
|
Thesis topic in co-operation with the Fraunhofer Institute for Applied and Integrated Security AISEC, Garching