Page 375 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

• Beyond telecom use cases: The system is extendable to multi-tenant domains like smart
grids, industrial IoT, healthcare, and vehicular networks, with initial trials planned in energy
and healthcare sectors.

3 Use Case Requirements 4.3 - 5G

• REQ-01: It is critical that the anomaly detection system operates across multiple 5G slices
without directly sharing raw data to preserve data isolation.
• REQ-02: It is critical that the federated learning framework effectively generalizes anomaly
patterns and distributes model updates efficiently across participating nodes.
• REQ-03: It is critical that the system ensures data privacy and adheres to applicable
regulatory standards at all times.
• REQ-04: It is critical that the solution supports real-time anomaly detection and mitigation
while maintaining minimal latency to ensure timely response.
• REQ-05: It is critical that the framework is scalable to accommodate dynamic and evolving
5G slice configurations.

4 Sequence Diagram

5 References

[1] Bonawitz, K., Eichner, H., et al. (2019). “Towards Federated Learning at Scale: System
Design.”arXiv preprint arXiv:1902.01046.

339