INTENTIONAL
Responsible:
NRRP
DEIB Role: Coordinator
Start date: 2024-05-15
Length: 14 months
Project abstract
Cybersecurity is of fundamental importance in modern telecommunications networks, given the crucial role these infrastructures play in today's society. The use of Artificial Intelligence and Machine Learning (ML) to enhance protection against cyberattacks within a telecommunication network is widespread. To improve efficiency, it is beneficial to implement ML models directly in the "data plane," i.e., directly within network devices such as routers or switches.
In this context, the INTENTIONAL - In-network attack detection by data plane implementation of machine-learning modelsproject aims to develop ML models suitable for such "in-network" implementation, with the goal of providing the network with adequate protection against cyberattacks. Additionally, since the behavior of malicious traffic is extremely dynamic, it is essential to continuously monitor the evolution of traffic and carefully select useful information in order to update ML models with new data.
The project has three main objectives:
- To develop Continual Learning algorithms compatible with the offloading of ML models into the data plane, in order to enable effective and continuous updates of ML models within the data plane;
- To design mechanisms for the efficient allocation of computational and memory resources to support concurrent ML models;
- To create a realistic dataset to evaluate performance in evolving malicious traffic scenarios.
INTENTIONAL is part of the extended PNRR partnership SERICS - SEcurity and RIghts In the CyberSpace and is integrated with the activities of the project Protect-IT - imPROving The rEsilience to Cyberattacks of distributed ICT InfrastrucTures under Spoke n. 8 - Risk management and governance.
