This WP investigates the methodologies for XInternet including data representation and creation of models for traffic classification and anomaly detection.

Task 2.1: Representation via embedding [UNINA, POLITO]

This task investigates data representation for network traffic using spatio-temporal embeddings. Also, we will map traffic traces in latent spaces where to develop supervised and unsupervised approaches. We will study how to update the embedding with new data, and how to leverage federated and transfer learning approaches to improve the embedding creation and let it evolve over time.

Task 2.2: System knowledge [UNINA, POLITO]
This task investigates the usage of centralized and distributed system knowledge creation where information about known traffic and attacks are maintained to support the network analysts in the identification of traffic generated by services and applications running on the Internet, observe changes in network traffic, identify new services, and unexpected activities. Algorithms for traffic classification and anomaly detection are the focus of this task.

Task 2.3: Model Sharing [UNINA, POLITO]
This task deals with the sharing of models and results from previous tasks. We will make them available on the project website and other repositories using the FAIR principles.

DELIVERABLES:

• D2.1 [M24 - POLITO] - Description of the methodologies for representation and system knowledge definition.
• D2.2 [M24 - UNINA] - Description of the models and labeled dataset made open.