From Centralized to Distributed and Ubiquitous In-Network Defense for Future 6G Networks

Future telecommunications networks are increasingly vulnerable to external attacks mainly due to the fast development of applications based on the massive use of IoT devices As a consequence, the perimeter of attacks on networks has expanded enormously and traditional security control techniques need to evolve to cope with this issue. A promising approach leverages the potential of Programmable Data Planes, that will characterize future networks, to implement effective in-network Intrusion Detection Systems (IDS). Initial works addressed the in-network IDS development, mainly with a centralized approach. In this paper, instead, we propose a Proof-of-Concept study dealing with a deployment approach that allows to distribute across the network nodes an in-network IDS created via a chain of Weak Learners. This, as demonstrated in the Prof-of-Concept experiments shown, allows to overcome some limitations of in-network centralized approaches, by working faster, and occupying fewer resources of the devices involved, while maintaining the level of precision sought.