MARE

PoC2 focuses on the Network Data Analytics Function (NWDAF), as specified in 3GPP Release 18, adopts a federated learning (FL) approach to enable decentralised data collection and distributed ML model training across multiple NWDAF instances deployed in the network. We will adopt this architecture supporting AI/ML-based attack detection within the service-based architecture (SBA) of the core network, creating datasets to train AI/ML models for attack detection purposes and leverage GANs to augment limited datasets and enhance detection accuracy. Local models will be continuously updated in a privacy-preserving manner through FL, ensuring data remains localised while benefiting from global intelligence.