Master’s thesis “Decentralised federated learning for semantic communications in IIoT Edge-Cloud environments”

Semantic communications shift the goal of data transmission from bit-accurate delivery to task-relevant meaning, reducing bandwidth and latency for industrial applications by transmitting only what matters for a given inference or control task. Existing semantic communication models are, however, fundamentally local optimisations: they treat encoding and decoding as isolated operations at the endpoints and assume that meaning degradation occurs only along a single link and can be fully compensated by better codec design.

In multi-hop IIoT networks, semantic distortion accumulates non-linearly across paths and depends on dynamic interactions between flows and routing decisions that are themselves unaware of semantic content. Extending endpoint-centric semantic models to the network level would require predicting and correcting for congestion dynamics, mobility-induced topology changes, and cross-traffic interference — a fundamental mismatch that current approaches do not address. This gap is formally characterised in Bilen & Akyildiz (arXiv:2603.12695, 2026), which proposes a Knowledge-Defined Networking framework as a first step and demonstrates clear gains in ns-3 simulation. The thesis builds directly on this baseline.

The thesis designs, implements, and evaluates a semantic-aware routing protocol for multi-hop IIoT networks operating over a 6G wireless substrate.

Your tasks:

• Analyse the mechanisms by which semantic distortion accumulates across multi-hop paths, characterising the non-linear interaction between flow dynamics, channel conditions, and routing decisions building on the ns-3 semantic channel module developed as shared infrastructure within SemComIIoT.
• Design a semantic-aware routing protocol that carries semantic distortion budgets as packet-level metadata, selects forwarding paths using a knowledge graph encoding task concepts and contextual relationships, and triggers adaptive re-encoding when accumulated distortion exceeds per-flow thresholds.
• Implement the protocol as a custom Ipv4RoutingProtocol in ns-3, integrating with the shared channel module via ns3-ai for knowledge graph queries and distortion budget management from a Python-side semantic reasoning engine.
• Evaluate the protocol against shortest-path, load-based, and distortion-only routing baselines across representative IIoT topologies — sensor-to-edge inference, edge-to-cloud aggregation, and multi-hop relay scenarios — measuring semantic delivery success rate, accumulated distortion, re-routing frequency, and task accuracy.
• Contribute the routing module as an open tool within SemComIIoT and results to the joint survey and analysis paper. 

__________

Your profile:

• Master student in Computer Science, Electrical Engineering, Communications Engineering, or a related field.
• Background in wireless communications, 5G.
• Practical programming experience in Python, C/C++, or similar.
• Familiarity with network simulation or emulation tools (e.g., ns-3, Mininet, OMNeT++) is desirable.
• Strong analytical skills and interest in real-time systems and network optimization.
• Excellent communication skills in English; German is a plus.

Our offer:

• An exciting and innovative open research environment within a nationally funded R&D project.
• Direct collaboration with industrial partners (Deutsche Telekom, Siemens) and academic research groups.
• Access to a state-of-the-art IIoT Lab with real 5G, TSN, and Wi-Fi 7 hardware infrastructure.
• International, dynamic work environment with highly qualified and motivated colleagues.
• Opportunity to contribute to IEEE/3GPP standardization and publish research results.
• Note: this master thesis position is not remunerated.